Carbon 42 (2004) 83–94

www.elsevier.com/locate/carbon

Adsorption of organic molecules from aqueous solutionson carbon materials q

Carlos Moreno-Castilla *

Departamento de Qu�ıımica Inorg�aanica, Facultad de Ciencias, Universidad de Granada, Campus Fuentenueva, 18071 Granada, Spain

Received 22 July 2003; accepted 30 September 2003

Abstract

Adsorption of organic molecules from dilute aqueous solutions on carbon materials is a complex interplay between non-elec-

trostatic and electrostatic interactions. Non-electrostatic interactions are essentially due to dispersion and hydrophobic interactions,

whereas the electrostatic or coulombic interactions appear with electrolytes when they are ionized at the experimental conditions

used. Both interactions depend on the characteristics of the adsorbent and the adsorptive and the solution chemistry. Among them

the carbon surface chemistry has a great influence on both electrostatic and non-electrostatic interactions, and can be considered one

of the main factors in the adsorption mechanism from dilute aqueous solutions. In this paper the current knowledge about the

fundamental factors that control the adsorption process from aqueous phase will be presented.

� 2003 Elsevier Ltd. All rights reserved.

Keywords: C. Adsorption; D. Adsorption properties; Surface oxygen complexes

1. Introduction

Adsorption of organic solutes from the aqueous

phase is a very important application of powdered and

granular activated carbons. This covers a wide spectrum

of systems such as drinking water and waste watertreatments, and applications in the food, beverage,

pharmaceutical and chemical industries. Activated car-

bon adsorption has been cited by the US Environmental

Protection Agency as one of the best available envi-

ronmental control technologies [1].

In spite of the large market for activated carbon, the

specific mechanisms by which the adsorption of many

compounds, especially organic compounds, takes placeon this adsorbent are still ambiguous. This is because

liquid-phase adsorption is a more complicated process

than gas or vapour-phase adsorption [2].

Thus, current knowledge about the fundamental

factors that control the adsorption process from the

aqueous phase will be presented, giving examples of the

different situations.

qPlenary lecture delivered at the International Conference on

Carbon, Carbon’03, held in Oviedo (Spain) July 6–10, 2003.* Tel.: +34-958-243-323; fax: +34-958-248-526.

E-mail address: cmoreno@ugr.es (C. Moreno-Castilla).

0008-6223/$ - see front matter � 2003 Elsevier Ltd. All rights reserved.

doi:10.1016/j.carbon.2003.09.022

2. Adsorption from dilute solutions: generalities

When studying adsorption from solutions on solids it

is convenient to differentiate between ‘‘adsorption from

dilute solution’’ and ‘‘adsorption from binary and

multicomponent mixtures covering the entire molefraction scale’’. To judge by the number of papers

published annually on adsorption from dilute solution,

this subject is more important than the adsorption from

binary mixtures. Therefore, reference will be made

hereafter to the adsorption from dilute aqueous solu-

tions.

The main differences between adsorption from the gas

phase and liquid phase are as follows [3]:

(i) A solution is typically a system of more than one

component. In real cases, there are at least two sub-

stances that can adsorb. For a dilute solution, ad-

sorption of one type of molecule (say A) involves

replacement of the other (B). Thus, adsorption

from solution is essentially an exchange process,

and hence molecules adsorb not only because theyare attracted by solids but also because the solution

may reject them. A typical illustration is that the at-

tachment of hydrophobic molecules on hydropho-

bic adsorbents from aqueous solutions is mainly

Effluent volume treated

Efflu

ent c

once

ntra

tion

** 21

Ci

C0

Ci

C0

Breakthrough volume

Fig. 2. Progression of the adsorption front through the adsorber bed.

The mass transfer zone is depicted in grey color.

84 C. Moreno-Castilla / Carbon 42 (2004) 83–94

driven by their dislike of water and not by their at-

traction to the surface.

(ii) Isotherms from solution may exhibit non-ideality,

not only because of lateral interactions between ad-sorbed molecules but also because of non-ideality in

the solution.

(iii) Multilayer adsorption from solution is less com-

mon than from the gas phase, because of the stron-

ger screening interaction forces in condensed fluids.

Adsorption isotherms are normally developed to

evaluate the capacity of activated carbons for the ad-sorption of a particular molecule. They constitute the

first experimental information, which is generally used

as a tool to discriminate among different activated car-

bons and thereby choose the most appropriate one for a

particular application. The shape of the isotherms is the

first experimental tool to diagnose the nature of a spe-

cific adsorption phenomenon, and it is expedient to

classify the most common types phenomenologically [3].There are several types of isotherms but those mainly

found in carbon materials are the five depicted in Fig. 1.

Long-linear isotherms are not common in adsorption

on carbons but are found in the initial part of all iso-

therms on homogeneous surfaces. The Langmuir type

(L) frequently occurs, even when the premises of the

Langmuir theory are not satisfied. Type F, typical for

heterogeneous surfaces, is perhaps the most common.High-affinity isotherms are characterised by a very steep

initial rise, followed by a pseudo-plateau. Sigmoidal

isotherms have been obtained with homogeneous sur-

faces such as the graphitised carbon blacks Graphon

and V3G.

Statistically, adsorption from dilute solutions is sim-

ple because the solvent can be interpreted as primitive,

that is to say as a structureless continuum [3]. Therefore,all the equations derived from monolayer gas adsorption

remain valid after replacing pressure by concentration

and modifying the dimensions of some parameters.

Fig. 1. Most common adsorption isotherms found from dilute aque-

ous solutions on carbon materials.

Some of these equations, such as the Langmuir and

Dubinin–Astakhov, are widely used to determine the

adsorption capacity of activated carbons.

Batch equilibrium tests are often complemented by

dynamic column studies to determine system size re-

quirements, contact time and carbon usage rates. These

parameters can be obtained from the breakthrough

curves. To study these curves it is usual to apply amacro-approach method based on the mass transfer

zone concept (MTZ). The basic principles of this con-

cept are depicted in Fig. 2. The MTZ is the region of the

bed between the already-saturated activated carbon and

the point where the concentration of the adsorptive in

the aqueous phase is at the maximum acceptable limit

in the effluent stream. Breakthrough is reached when the

leading edge of the zone advances to the end of the bed(point 2 in Fig. 2). The height of this zone is a measure

of the adsorption efficiency of the bed. The lower the

height, the more efficient is the carbon bed. Under the

appropriate experimental conditions, this parameter is

independent of the bed depth and generally shows little

variation with in the initial concentration.

Adsorption measurements should preferably be am-

plified by microcalorimetry, such as immersion and flowmicrocalorimetry. These techniques can give additional

information about the surface nature of the adsorbent

and the mode or mechanism of adsorption.

3. Factors that control the adsorption process

Adsorption is a spontaneous process that takes placeif the free energy of adsorption, DGads, is negative. There

is a wide range of energies contributing to the free en-

ergy of adsorption, which can be grouped into non-

electrostatic and electrostatic.

DGads ¼ DGnon-elect: þ DGelect:

C. Moreno-Castilla / Carbon 42 (2004) 83–94 85

Although at atomic level all ionic and molecular in-

teractions can be interpreted as ‘‘electric’’, this term is

restricted to coulombic interactions and all other inter-

actions are termed non-electrostatic, whatever their or-igin.

Electrostatic interactions appear when the adsorptive

is an electrolyte that is dissociated or protonated in

aqueous solution under the experimental conditions

used. These interactions, that can be either attractive or

repulsive, strongly depend on the charge densities for

both the carbon surface and the adsorptive molecule

and on the ionic strength of the solution, as we shall seelater. The non-electrostatic interactions are always at-

tractive, and can include van der Waals forces, hydro-

phobic interactions and hydrogen bonding. Factors that

influence the adsorption process are the characteristics

of the adsorbent and adsorptive, the solution chemistry

and the adsorption temperature. These factors will now

be discussed.

3.1. Characteristics of the adsorbent

The characteristics of carbon that affect the adsorp-

tion process are the pore texture, surface chemistry and

mineral matter content. The adsorption capacity of

carbon materials is not related in a simple form with

their surface area and porosity. The adsorption capacity

will depend on the accessibility of the organic molecules

to the inner surface of the adsorbent, which depends on

their size. Thus, under appropriate experimental con-ditions, small molecules such as phenol can access

micropores, natural organic matter (NOM) can access

mesopores, and bacteria can have access to macropores.

Activated carbon fibres have received increasing at-

tention in recent years as a better adsorbent than

granular activated carbons, because they normally

present much higher adsorption kinetics and adsorption

capacity (Fig. 3). Activated carbon fibres only havemicropores, which are directly accessible from the ex-

ternal surface of the fibre. Thus, adsorptive molecules

reach adsorption sites through micropores without the

0

5

10

20

25

0 2 4 6 8t (days)

Xm (m

g ad

s/g

carb

on) Norit R 0.8

F400

ACF

15

Fig. 3. Adsorption kinetics of aminetriazol on an activated carbon

fibre (ACF) and activated carbons (F400 and Norit R 0.8).

additional diffusion resistance of macropores, which is

usually the rate-controlling step in the case of granular

adsorbents.

The surface chemistry of activated carbons essentiallydepends on their heteroatom content, mainly on their

surface oxygen complex content. They determine the

charge of the surface, its hydrophobicity, and the elec-

tronic density of the graphene layers. Thus, when a solid

such as a carbon material is immersed in an aqueous

solution, it develops a surface charge that comes from

the dissociation of surface groups or the adsorption of

ions from solution [2]. This surface charge (Fig. 4) willdepend on the solution pH and the surface characteris-

tics of the carbon. A negative charge results from the

dissociation of surface oxygen complexes of acid char-

acter such as carboxyl and phenolic groups. Therefore,

these surface acid sites are of Br€oonsted type. The origin

of the positive surface charge is more uncertain because,

in carbons without nitrogen functionalities, it can be due

to surface oxygen complexes of basic character like py-rones or chromenes, or to the existence of electron-rich

regions within the graphene layers acting as Lewis basic

centres, which accept protons from the aqueous solu-

tion.

An indication that surface basicity can be mainly due

to these electron-rich regions is that, the net enthalpy of

neutralization of the total basic sites decreases with the

oxygen content of the surface (Fig. 5), whereas the netenthalpy of neutralization of total acid sites increases [4].

Results obtained with activated carbons of different

origins can be explained by the fact that oxygen com-

plexes reduce the electronic density of the graphene

layers and consequently reduce the basicity of the car-

bon surface.

The surface charge can be determined by electroki-

netics or titration methods. These are complementarymethods, especially in the case of granular porous

Fig. 4. Macroscopic representation of the features of carbon surface

chemistry. From [2].

0

100

200

0 5 10 15

0.4 0.5 0.6 0.7Micropore volume (ml/g)

% Oxygen content

Amou

nt a

dsor

bed

(mg/

g)

Fig. 6. Relationship between the maximum amount of dodecanoic

acid adsorbed on different carbons with their oxygen content and

micropore volume. Adapted from [7].

-50

-40

-30

-20

-10

00 0.4 0.8 1.2

Hydrophilic sites (mmol/g)

Enth

alpy

of d

ispl

acem

ent

(kJ/

mol

)

Fig. 7. Enthalpy of displacement of water by 2-methylisoborneol

(MIB) versus hydrophilic sites on activated carbons. Adapted from [8].

Fig. 5. Variation of the net enthalpy of neutralization DH (NaOH)net,

D; and DH (HCl)net,j, with the total amount of oxygen on the surface.

From [4].

86 C. Moreno-Castilla / Carbon 42 (2004) 83–94

carbons [2]. The first method primarily measures the

surface charge at the more external surface of the par-

ticles whereas the second one provides a measure of the

total surface charge. The pH at which the external sur-

face charge is zero is referred to as the isoelectric point,

pHIEP, while the total surface charge is zero at the point

of zero charge, pHPZC.

In addition, surface oxygen complexes also affect thesurface hydrophobicity, which determines the hydro-

phobic interaction. Hydrophobic interaction or, more

specifically, hydrophobic bonding describes the unusu-

ally strong attraction between hydrophobic molecules

and hydrophobic carbon surfaces. The surface hydro-

phobicity of carbon materials can be determined, for

instance, by measuring their contact angle with test

liquids of known surface tension, by inverse gas chro-matography or by water adsorption.

Hydrophobic bonding occurs exclusively in aqueous

solution: in fact it mainly comes from the strong ten-

dency of water molecules to associate with each other by

hydrogen bonding and from the characteristic structure

of water molecules. Hydrophobic bonding plays an

important role in interface and colloid science [5]. For

instance, the increasing adsorbability of aliphatic acidmolecules with increasing hydrocarbon length, also

known as Traube’s rule, is essentially due to the in-

creasing hydrophobic effect [6].

From this it is derived that it is important to plot the

adsorption isotherms obtained for adsorptives with

different solubilities against the concentration relative to

that in saturated solution. This normalisation eliminates

the differences in hydrophobicity between the adsorptivemolecules, so that the normalised isotherms more truly

reflect the affinity for the surface [5,6].

In general, an increase in the oxygen content of car-

bon brings about a decrease in its hydrophobicity. Thus,

Pendleton et al. [7] showed (Fig. 6) that the adsorption

of dodecanoic acid on different activated carbons line-

arly decreased when the oxygen content of the carbo-

naceous adsorbent increased and that there was no

relation with the micropore volume of the adsorbent.

Water molecules are bound to the surface oxygen

complexes by H-bonds, reducing the accessibility of the

hydrophobic aliphatic chain of dodecanoic acid to the

hydrophobic parts of the carbon surface.

This model is consistent with that found for the de-

crease in methyl isoborneol (MIB) adsorption when theoxygen content of carbons increased [8]. In this case, the

average enthalpy of displacement of water by MIB (Fig.

7) decreased when the concentration of hydrophilic sites

on the surface increased. This finding suggests that it is

more difficult for MIB molecules to displace water

molecules from carbon surfaces when their oxygen

content increases. Results obtained show that the en-

thalpy of displacement reaches a limiting value when thenumber of hydrophilic sites is higher than about 0.7

mmol/g.

Finally, surface oxygen complexes also affect the

electronic density of the graphene layers [9]. This in turn

would affect the dispersion interactions between the

carbon surface and the adsorptive molecules. For in-

stance, carboxyl groups fixed at the edges of the

graphene layers have the ability to withdraw electrons,whereas phenolic groups release electrons.

Table 1

Energies of the highest occupied molecular orbital (HOMO) and

lowest unoccupied molecular orbital (LUMO) for model carbon

clusters. Adapted from [10]

HOMO (eV) LUMO (eV)

AC )6.19 )1.35AC–OH )5.85 )1.09AC–COOH )6.61 )1.92

Table 2

Surface area and surface acidity of activated carbons with different

degrees of oxidation. From [11]

Sample A N2 surface

area (m2/g)

Surface acidity

(meq/g)

As received 975 0.12

As received oxidized 503 5.93

Oxidized and heat treated

at (�C):300 603 4.79

400 606 2.74

500 681 1.60

700 640 0.34

950 675 Nil

C. Moreno-Castilla / Carbon 42 (2004) 83–94 87

Thus, Tamon and Okazaki [10] determined by a semi-

empirical quantum chemical method the HOMO and

LUMO levels of a model aromatic molecule such as a

non-substituted phenanthroperylene cluster (AC), and

the same cluster substituted with two phenolic groups

(AC–OH) and with two carboxyl groups (AC–COOH).

Results are shown in Table 1 and indicate that the

withdrawing groups, such as carboxyl groups, decreaseHOMO and LUMO levels with respect to the original

carbon cluster, whereas the donating groups increase

these levels.

Finally, the other characteristics of the adsorbent that

control the adsorption process is its mineral matter

content. This has in general a deleterious effect on the

adsorption because it can block the porosity of the

carbon matrix and can preferentially adsorb water dueto its hydrophilic character, in this case reducing the

adsorption of the adsorptive.

A-ox

A-ox 300º

A

A-ox 950º

A-ox 700º

A-ox 500º

A-ox 400º

0 100 200 300 400 500

0.5

1.0

1.5

0.0

Equilibrium concentration (ppm)

Phen

ol u

ptak

e ( µµ

mol

/g)

Fig. 8. Adsorption isotherms of phenol on activated carbons with

different degrees of oxidation. From [11].

3.2. Characteristics of the adsorptive

Among the characteristics of the adsorptive that

mainly influence the adsorption process are its molecu-

lar size, solubility, pKa and the nature of the substituent

if they are aromatic. The molecular size controls the

accessibility to the pores of the carbon and the solubility

determines the hydrophobic interactions. The pKa con-

trols the dissociation of the adsorptive if it is an elec-trolyte. This parameter is closely related to the solution

pH and its importance will be discussed later on.

The substituent of the aromatic ring of the adsorptive

molecules, as in the case of the substituent of the

graphene layers, can withdraw or release electrons from

it, which would affect the dispersion interactions be-

tween the aromatic ring of the adsorbate and the

graphene layers of the adsorbent.To date, the adsorption of phenolic compounds on

carbon materials and the effect of surface oxygen com-

plexes on phenol uptake have been the most studied

adsorption processes [2]. Thus, it has long been known

that the increase in surface acidity of activated carbons,

produced after their oxidation, brings about a decrease

in the amount of phenol adsorbed from diluted aqueous

solutions. To show this phenomenon (Table 2) sample Awas oxidized to increase its surface acidity. Heat treat-

ment of the oxidized sample progressively decreased the

surface acidity due to the removal of mainly carboxyl

and phenolic groups [11]. A sample heat treated at 950

�C had no acidity and its surface area was lower than

that of as-received sample. Fig. 8 shows the adsorptionisotherms of phenol on the above carbons. There was a

large decrease in phenol uptake after oxidation. This

phenol uptake progressively increased as the surface

acidity decreased, and the oxidised sample heat treated

at 950 �C had the same adsorption capacity as the as-

received one, despite the lower surface area of the for-

mer sample compared with the latter.

Three mechanisms have been mainly proposed toexplain this behaviour, namely: the p–p dispersion in-

teraction mechanism, the hydrogen bonding formation

mechanism and the electron donor–acceptor complex

mechanism.

The first two mechanisms were proposed by Coughlin

and Ezra [9] in 1968, and the third mechanism was

proposed by Mattson et al. [12] in 1969. At that moment

it was known that phenol was adsorbed in flat positionon the graphene layers, and in this situation the ad-

sorption driving forces would be due to p–p dispersion

interactions between the aromatic ring of phenol and the

aromatic structure of the graphene layers.

Thus, Coughlin and Ezra proposed that acidic sur-

face oxygen groups, which are located at the edges of the

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

0 5 10 15 20 25Equilibrium concentration (ppm)

Phen

ol u

ptak

e (m

mol

/g)

GRAPHITE

GRAPHITE-0.79% B

Fig. 9. Adsorption of phenol on graphite substitutionally doped with

B and undoped. From [11].

1.5

1.0

0.5

0.00.0 100 200

1.5

1.0

0.5

0.0

Equilibrium conc. ( µµmole/l)

Phen

ol u

ptak

e (m

mol

e/g)

Phen

ol u

ptak

e ( µµ

mol

e/m

2 )

Fig. 10. Effect of surface oxygen complexes on adsorption of phenol

from cyclohexanic solution. Oxidized carbon heat treated at 300 �C, �,

and 400 �C, d, SBET with cyclohexane 420 and 500 m2/g, respectively.

Surface acidity: 4.79 and 2.74 meq/g, respectively. From [11].

88 C. Moreno-Castilla / Carbon 42 (2004) 83–94

basal planes, remove electrons from the p-electron sys-

tem, creating positive holes in the conducting p-band of

the graphitic planes. This would lead to weaker inter-

actions between the p-electrons of the phenol aromaticring and the p-electrons of the basal planes, therefore

reducing the phenol uptake.

Coughlin also proposed that the bonding of water

molecules to the oxide functional groups by H-bonding

can play an important role in the uptake of phenolic

compounds. In this case, Coughlin adopted Dubinin’s

proposal that water molecules adsorbed to oxygen

groups become secondary adsorption centres, whichretain other water molecules by means of H-bonds. As a

result, complexes of associated water form within the

pores of a carbon adsorbent. These complexes could

prevent the migration of organic molecules to a large

portion of the active surface area. This mechanism was

ruled out by Coughlin, whose results indicated that

surface oxygen complexes had no influence on phenol

uptake from concentrated solutions (in the second pla-teau of the isotherms).

Mattson and coworkers suggested that aromatic

compounds adsorb on carbons by a donor–acceptor

complex mechanism, with the carbonyl oxygen of the

carbon surface acting as the electron donor and the ar-

omatic ring of the adsorbate acting as the acceptor.

Once the carbonyl groups are exhausted, the aromatic

compounds form donor–acceptor complexes with therings of the basal plane. Thus, Mattson and co-workers

explained the decrease in phenol uptake after carbon

oxidation as due to the oxidation of carbonyl groups to

carboxyl groups. As a result, the electron donor–ac-

ceptor complexes cannot be formed.

At this point it is interesting to indicate that Coughlin

and Ezra pointed out in their key paper [9] that ‘‘evi-

dence for any single explanation (of the decrease ofphenol uptake with increase in surface acidity) is not

overwhelming’’ and added that ‘‘it is hoped that

continuing research will shed more light on this ques-

tion’’. However, this was not, unfortunately, the case in

many instances, as can be seen in the recent review

published by Radovic et al. [2].

Since the initial proposals by Coughlin and Mattson,

many published papers have attempted to elucidate themost appropriate mechanism to explain the adsorption

of phenolic compounds, and in general aromatic com-

pounds on carbon materials. Thus, perhaps one of the

latest papers published on this topic to date was the one

published by Haydar et al. that appeared in 2003 enti-

tled ‘‘Adsorption of p-nitrophenol on an activated car-

bon with different oxidations’’ [13]. This paper gives

some experimental evidence for the p–p dispersion in-teraction mechanism.

Perhaps the first experimental evidence of this

mechanism was given by Mahajan et al. [11] in their

study of phenol adsorption on graphite and boron-

doped graphite samples (Fig. 9). Results show that thepresence of substitutional boron in the lattice of poly-

crystalline graphite, which removes p-electrons from the

solid, results in a lowering of the phenol uptake from

water.

These authors also indicated that both phenol and

water can compete to form H-bonds with surface oxy-

gen groups, such as carboxyl groups. In this competition

water molecules are preferentially bound by H-bonds[11].

This idea of competition between phenol and water

molecules for H-bonding to surface oxygen groups

suggested that phenol uptake from its cyclohexane so-

lution would be favoured by the surface oxygen com-

plexes. This is shown in Fig. 10. When adsorption of

phenol per unit cyclohexane area is plotted for two heat-

treated carbons, the carbon with higher surface acidityexhibited a higher phenol uptake.

More recently, Pinto and co-workers [14] revisited

this issue by studying the adsorption of phenol aniline,

nitrobenzene and benzoic acid from both aqueous and

cyclohexanic solutions. Although the interpretation of

40

45

50

55

60

1.0 2.0 3.0 4.0 5.0Enthalpy of immersion (-kJ/mol)

Enth

alpy

of e

xcha

nge

(-kJ/

mol

)

Fig. 12. The enthalpy of exchange for phenol–water versus the molar

enthalpy of immersion in water for a series of BPL carbons. Adapted

from [16].

C. Moreno-Castilla / Carbon 42 (2004) 83–94 89

the authors of the p–p dispersion interactions is con-

fused, their experimental results seem to indicate that

the adsorption mechanism is both by water H-bonding

with carboxyl groups and p–p dispersion interactionsbetween the aromatic ring of the adsorptive and the

graphene layers. These authors also concluded that

Mattson’s mechanism is not the driving force for the

adsorption of aromatics on activated carbons.

One of the weak points of the Mattson mechanism is

that there is much experimental evidence that although

oxidation of carbons increases their concentration of

CO2-evolving groups, the CO-evolving groups also in-crease or remain essentially unchanged.

Additional experimental evidence of the influence of

water adsorption comes from the enthalpy of immersion

into water of activated carbons, which reflects the spe-

cific and non-specific interactions between the liquid and

the solid [15] as shown in Fig. 11. The study of a large

number of carbons oxidised to different degrees reveals a

simple correlation between the enthalpy of immersion,the oxygen content of the surface, the basic groups ti-

trated by HCl, the micropore filling and the wetting of

the external surface. Thus, the specific interaction be-

tween the surface oxygen and water is 12.1 kJ/mol ox-

ygen, involving an average of two water molecules per

oxygen.

More recently, MacDonald and Evans [16] measured

the enthalpy of exchange of phenol–water from dilutedaqueous solutions on BPL carbons with different oxygen

contents. They found (Fig. 12) that phenol exchange

enthalpy is inversely proportional to the net molar en-

thalpy of immersion into water. This indicates that the

strong preference of water to H-bond to surface oxygen

complexes lowers the phenol exchange enthalpy from

aqueous solution.

The two mechanisms proposed by Coughlin canbetter explain many of the experimental results obtained

to date. However, an electron donor–acceptor mecha-

12.1[O] + 10.3[meq.HCl] ++ 0.8(Nam-2.0[O] - 1.0[meq.HCl]) + Se×0.031

10

30

50

70

90

110

10 30 50 70 90 110

_ ∆iH

(H2O

) / J

g-1

Fig. 11. Correlation between the enthalpy of immersion into water,

the total surface oxygen, the basic groups titrated with HCl, the mi-

cropore filling and the wetting of the non-porous surface area. From

[15].

nism cannot be ruled out. Thus, it is well known that

adsorption of phenolic compounds is partly physical

and partly chemical. The physisorbed part can be de-

sorbed by treating with different solvents or by heat

treatments in inert atmosphere. However, the chemi-

sorbed part cannot be desorbed, even at high temper-

atures, and it is converted to light gases and heavyproducts that evolve from the surface of the carbon, and

also to a polymeric carbon residue that remains on the

surface. This residue reduces the adsorption capacity of

thermally regenerated activated carbons, as depicted in

Fig. 13, which shows that the adsorption capacity for

different phenolic compounds decreased when the re-

generation cycle increased [17].

Better regeneration results could be achieved bytreating the exhausted carbons with liquid water at 150

atm and 350 �C in the absence of oxygen. Some of the

results obtained in a recent study [18] of the regeneration

of exhausted carbons with ortho-chlorophenol (OCP)

are shown in Table 3. In no case was there any loss of

carbon due to the treatment. The efficacy of the treat-

ment was slightly above 100% for the first regeneration

cycles. However, for successive regeneration cycles there

0

70

140

210

0 1 2 3 4 5 6 7Number of cycles

Ads

optio

n ca

paci

ty (m

g/g)

Fig. 13. Variation of the adsorption capacity of the activated carbon

AP-10 as a function of the adsorption–regeneration cycle. Phenol, D;m-aminophenol, �; p-cresol, �; p-nitrophenol, }. From [19].

65

90

115

140

165

2 7 12pH

mg/

g

Fig. 14. Adsorption capacity of activated carbon CP-10 as a function

of solution pH. �, Phenol (pKa ¼ 9.96); �, m-chlorophenol

(pKa ¼ 8.80); D, p-nitrophenol (pKa ¼ 7.13). From [19].

Table 3

Amount of ortho-chlorophenol adsorbed and regeneration yield in

successive adsorption–regeneration cycles. Regeneration conditions:

liquid water at 350 �C and 150 atm. Adapted from [18]

Activated

carbon

Run number Amount adsorbed

(mmol/g)

Regeneration

yield (%)

H-30 0 1.63

1 1.69 104

2 1.61 99

3 1.58 97

H-13 0 0.92

1 0.95 104

2 0.96 105

3 0.87 95

RO-0.8 0 1.05

1 1.07 103

2 1.09 104

3 1.04 99

90 C. Moreno-Castilla / Carbon 42 (2004) 83–94

was a slight decrease in the OCP adsorbed, because,

according to the authors, part of it was strongly ad-

sorbed and was not completely removed from the car-

bon surface.

All these data indicate that chemisorbed phenolic

compounds are strongly bound by other forces than

those of dispersion. These would likely involve charge-

transfer complexes in which the direction of the electrontransfer could be either similar to that proposed by

Mattson or in a reverse direction.

Table 4

Characteristics of activated carbon beds when removing ortho-chlo-

rophenol at pHs 2 and 10. Particle size between 0.15 and 0.25 mm.

From [20]

Sample pH¼ 2 pH¼ 10

VB (cm3) HMTZ (cm) VB (cm3) HMTZ (cm)

C-2 4450 1.02 550 2.02

C-13 6500 0.95 1820 1.54

C-24 10,650 0.48 2800 1.32

3.3. Solution chemistry and adsorption temperature

Factors from solution chemistry that influence the

adsorption process are the solution pH and ionic

strength. Solution pH is one of the key factors that

control the adsorption process of organic weak elec-

trolytes and polyelectrolytes on carbon materials be-

cause it controls the electrostatic interactions between

the adsorbent and the adsorbate. Thus, solution pHdetermines the carbon surface charge and the dissocia-

tion or protonation of the electrolyte. At a solution pH

lower than the pHPZC or the pHIEP, the total or external

surface charge, respectively, will be on average positive,

whereas at a higher solution pH they will be negative.

In addition, the solution pH also controls the disso-

ciation or ionization of the electrolyte through its pKa.

Thus, for instance, acidic electrolytes will be dissociatedat pH>pKa. Therefore, the solution pH controls the

adsorptive–adsorbent and adsorptive–adsorptive elec-

trostatic interactions, which can have a profound effect

on the adsorption process.

Thus, the adsorption of substituted phenols on acti-

vated carbon depends on the solution pH (Fig. 14). At

acidic pH, the uptake was maximal because phenols

were undissociated and the dispersion interactions pre-dominated. At basic pH, however, the uptake was lower

because of electrostatic repulsions between the negative

surface charge and the phenolate anions and between

phenolate–phenolate anions in solution. The pH at

which the uptake decreased was found to be dependenton the adsorptive pKa and the difference between the

pHPZC and the pHIEP [19].

Solution pH also affects the characteristics of carbon

beds, as shown in Table 4. Thus, adsorption of OCP is

favoured at acidic pH, as illustrated by the higher values

of the breakthrough volumes and the lower values of the

height of the mass transfer zone found at this pH. The

lower adsorption at basic pH is due, as explained before,to the repulsion between the net negatively charged

carbon surface and the ortho-chlorophenolate anions

[20].

The effect of pH and the nature of functional groups

of both the aromatic adsorptive and the adsorbent on

the adsorption process have recently been investigated

by Radovic et al. [21]. For this purpose, they used an as-

received activated carbon oxidized with nitric acid andnitrided with ammonia to study the adsorption of ani-

line and nitrobenzene, which are, respectively, electron-

donating and -withdrawing groups. Some of the results

obtained from the adsorption of aniline at different pHs

are shown in Table 5. At pH 2, anilinium cations were

predominant in solution and the surface charge was

positive. Carbon oxidation enhanced adsorption with

respect to the as-received carbon, whereas nitriding de-

Table 5

Aniline uptake (mmol/g) at a relative concentration of 10�3 on Norit

GCW carbon at different solution pHs

Carbon pH¼ 2 pH¼ 11 pH¼ pHPZC

As received 8.0a 0.48 1.10 1.10

Oxidized 2.6 1.14 0.86 1.67

Nitrided 8.9 0.33 0.86 0.80

Aniline pKa ¼ 4.6. Adapted from [21].a pHPZC values.

C. Moreno-Castilla / Carbon 42 (2004) 83–94 91

creased it. The authors indicated that in this case the

electrostatic adsorbent/adsorbate interactions were

more important than the dispersion interactions. Thus,

oxidation shifted the pHPZC to a value close to the so-

lution pH, whereas nitriding increased the pHPZC, en-

hancing the adsorbate–adsorbent interactions.

At pH 11, the surface charge was predominantly

negative and aniline was not protonated in solution. Inthis case, oxidation and nitriding were detrimental.

These results can be explained because dispersion in-

teractions predominate and are not favoured, because

the presence of some oxygen and nitrogen-containing

functional groups withdraws electrons from the graph-

ene layers. However, in this case the preferential H-

bonding of water to some oxygen and nitrogen surface

functional groups can also explain these results.At a solution pH equal to the pHPZC, aniline uptake

was relatively high on all adsorbents. According to the

authors this is consistent with the minimized electro-

static repulsions.

Some of the results obtained with nitrobenzene are

shown in Table 6. In this case solution pH has little

effect on equilibrium uptakes because nitrobenzene

molecules are the dominant species across the entirerange of solution pH. Therefore the effect of surface

chemistry is much more important. Across the pH range

the as-received carbon exhibits the highest uptake. The

authors explained these results as due to the disper-

sion interactions, which are dominant in this case.

Thus, surface oxygen and nitrogen complexes with-

draw electrons from the graphene layers decreasing the

p–p interactions. However, as commented above, theH-bonding of water to some surface functionalities can

also play an important role in this case. Maximal uptake

Table 6

Nitrobenzene uptake (mmol/g) at a relative concentration of 2.10�2 on

Norit GCW carbon at different solution pHs

Carbon pH¼ 2 pH¼ 11 pH¼ pHPZC

As received 8.0a 2.43 2.61 3.04

Oxidized 2.6 0.78 0.70 0.95

Nitrided 8.9 1.30 1.65 1.91

Adapted from [21].a pHPZC values.

is obtained at pH¼ pHPZC because dispersion interac-

tions are maximized.

These results, together with others found in the litera-

ture, indicate that functionalization of either the adsorp-tive or the adsorbent, which increases the p-electrondensity, leads to enhanced or stronger adsorption when

the adsorption process is governed by dispersion forces.

The converse is also true.

Ionic strength is the other key factor that controls the

electrostatic interactions. Thus, these interactions, either

attractive or repulsive, can be reduced by increasing the

ionic strength of the solution. This is due to a screeningeffect of the surface charge produced by the added salt.

Therefore, when the electrostatic interaction between

the surface and the adsorptive is repulsive, or the surface

concentration is sufficiently high, an increase in ionic

strength will increase the adsorption. Conversely, when

the electrostatic interactions are attractive, or the sur-

face concentration is sufficiently low, an increase of the

ionic strength will diminish the adsorption. These effectshave been shown to take place with many weak organic

electrolytes and polyelectrolytes, for instance in the case

of bisphenol A and NOM [22,23].

In the case of bisphenol A, Fig. 15 shows that its

uptake increased with the ionic strength [22]. At solution

pH 7, the surface charge would be on average positively

charged and screened by the increase in the electrolyte

concentration, resulting in an enhancement of the bis-phenol uptake.

NOM is found in varying concentrations in all nat-

ural water sources. It is a complex mixture of com-

pounds varying from small hydrophilic acids, proteins

and amino acids to larger humic and fulvic acids.

Newcombe and Drikas [23] have studied the adsorption

of NOM (with nominal molecular weight between 500

and 3000 Da) on different activated carbons. Fig. 16shows the adsorption of NOM at pH 7 on carbon W. At

this pH, both the carbon surface and NOM are nega-

tively charged. At very low surface coverage direct

surface–adsorbate interactions should predominate.

An increase in salt concentration effectively screens re-

pulsive interactions, resulting in increased adsorption.

In the case of carbon C, Fig. 17, the adsorption of

NOM was carried out at pH 4. In these conditions thecarbon surface will be positively charged whereas NOM

will be negatively charged. The adsorption isotherms

show an intersection point indicating a transition from a

screening-reduced to a screening-enhanced regime. At

low surface concentrations below the intersection point

of the isotherms, attractive interactions between the

adsorbent and the adsorbate are screened by an increase

in salt concentration, decreasing the adsorption. Abovethe intersection point, where the surface concentration is

higher, salt screens the repulsion between charged seg-

ments of the polyelectrolyte and between adsorbed

NOM and NOM in solution, increasing the adsorption.

0

100

200

300

80400Ce (mg/l)

X (m

g/g)

Fig. 15. Adsorption isotherms of bisphenol A on activated carbon A

(pHPZC ¼ 10) in the absence of NaCl, d, and in the presence of NaCl:

0.01 M, N; 0.1 M, j. Solution pH¼ 7.

0

20

40

60

80

0 4 8 12 16Solution Concentration (mg/l)

Am

ount

ads

orbe

d (m

g/g)

0.30 M NaCl

0.01 M NaCl

Fig. 16. Adsorption of NOM on carbon W (pHPZC ¼ 4.5) at two ionic

strengths and at solution pH¼ 7. Adapted from [23].

0

20

40

60

80

0 4 8 12 16

Am

ount

ads

orbe

d (m

g/g) 0.30 M NaCl

0.01 M NaCl

Solution Concentration (mg/l)

Fig. 17. Adsorption of NOM on carbon C (pHPZC ¼ 7.5) at two ionic

strengths and at solution pH¼ 4. Adapted from [23].

Table 7

Surface area (m2/g) of different carbons from adsorption isotherms

applying DRK equation and immersion calorimetry

Carbon STotala Sphenol Sm-chlorophenol

PLW 1097 1016 (1037)b 1050 (1047)

AP-5 404 453 453

CP-5 477 596 522

From [24].a From CO2 at 273 K.bFrom immersion calorimetry.

0

200

400

0 0.2 0.4 0.6 0.8Pore volume (cm3/g)between 0.8 - 50 nm

Am

ount

ads

orbe

d (m

g/g)

Fig. 18. Relationship between NOM adsorption capacity and avail-

able adsorption volume for different activated carbons at solution

pH¼ 3. Adapted from [23].

92 C. Moreno-Castilla / Carbon 42 (2004) 83–94

Therefore, both solution pH and ionic strength con-

trol the electrostatic interactions between the adsorbent

and the adsorptive.

When the adsorption of organic molecules is gov-

erned by non-electrostatic interactions, such as p–pdispersion or hydrophobic interactions, the area of the

adsorbent occupied by the adsorbate depends on the

porosity of the former and the molecular size of the

latter. Thus, it has been shown [24] in adsorption from

diluted aqueous solution and immersion calorimetry

measurements (Table 7) that phenol and metachloro-

phenol are adsorbed as monolayers by both porous and

non-porous carbons with basic surface properties,

provided that the adsorptive is undissociated at the so-

lution pH. This did not apply where molecular sieveeffects reduced the accessibility of the micropore system.

In the case of the adsorption of NOM (Fig. 18) in the

molecular weight range between 500 and 3000 on dif-

ferent activated carbons at solution pH 3, where elec-

trostatic interactions are minimal and non-electrostatic

interactions predominate, the amount adsorbed depends

on the volume of pores between 0.8 and 50 nm [23].

Another example is the adsorption of trichloroethene(Fig. 19) from highly diluted aqueous solutions on ac-

tivated carbon fibres and granular activated carbons

with similar surface chemistry [25]. In this case, the tri-

chloroethene adsorption on the more hydrophobic car-

bons was primarily controlled by pore volume in the

0.7–1 nm width range. According to the authors, this

micropore range is between 1.3 to 1.8 times the tri-

chloroethene kinetic diameter (0.56 nm).With regard to the effect of temperature on the ad-

sorption, an increasing uptake of organic molecules is

expected when the adsorption temperature decreases

because adsorption is a spontaneous process. However,

10

20

30

40

0.1 0.15 0.2 0.25Pore volume in the 0.7 - 1nm

range (cm3/g)

Am

ount

ads

orbe

d (m

g/g)

AAWAWHAWG219F600

width

Fig. 19. Effect of pore volume on trichloroethene adsorption. Adapted

from [25].

C. Moreno-Castilla / Carbon 42 (2004) 83–94 93

some examples have been reported where the amount

adsorbed increased with temperature. Thus, for in-

stance, it has recently been reported [26] that the ad-sorption of paracetamol from diluted aqueous solution

increased with adsorption temperature. This effect was

independent of the type of carbon, its surface chemistry

or solution pH. The authors explained this behaviour as

due to phase changes in the crystal form of the ad-

sorptive.

4. Conclusions

Adsorption of organic molecules from dilute aqueous

solutions on carbon materials is a complex interplaybetween electrostatic and non-electrostatic interactions.

Electrostatic interactions appear with electrolytes, es-

sentially when they are ionized at the experimental

conditions used. Both interactions depend on the char-

acteristics of the adsorbent and the adsorptive, and the

solution chemistry. Non-electrostatic interactions are

essentially due to dispersion and hydrophobic interac-

tions. The surface chemistry of the carbon has a greatinfluence on both electrostatic and non-electrostatic in-

teractions, and can be considered the main factor in the

adsorption mechanism from dilute aqueous solutions.

Aromatic compounds are physisorbed on carbon

materials essentially by dispersion interactions between

the p-electrons of the aromatic ring and those of the

graphene layers. Functionalization of either the adsor-

bent or the adsorptive profoundly affects these disper-sion interactions. In addition, the presence on carbon

surfaces of chemical functionalities that can give rise to

H-bonds with water can effectively reduce the adsorp-

tion of the adsorptive molecules.

Strongly adsorbed aromatic compounds on the car-

bon surface, which cannot be desorbed even at high

temperatures, would probably be adsorbed on the sur-

face by stronger interactions than the dispersion ones.

This likely involves an electron donor–acceptor or

charge-transfer mechanism. Thus, further research

would be needed in this area.On the other hand, it has been shown that the surface

of the carbon material is more effectively used if non-

electrostatic interactions are the driving force for the

adsorption. This condition can be reached by control-

ling the surface chemistry of carbon, and the pH and

ionic strength of the solution.

Finally, some research trends on this topic are the

study of competitive adsorption, for instance betweenNOM and micropollutants such as pesticides; adsorp-

tion of bacteria which in turn can catalyse the decom-

position of the adsorptive; adsorption–desorption of

drugs for medical applications and modifications of the

adsorptive depending on the solution and carbon sur-

face chemistry.

Acknowledgements

The author wants to acknowledge to Drs. Jos�ee Ri-

vera-Utrilla and Victoria L�oopez-Ram�oon for fruitfull

discussions about the manuscript. Financial support

from the Ministerio de Ciencia y Tecnolog�ııa and Feder

is also acknowledged.

References

[1] Derbyshire F, Jagtoyen M, Andrews R, Rao A, Mart�ıın-Gull�oon I,

Grulke E. Carbon materials in environmental applications. In:

Radovic LR, editor. Chemistry and physics of carbon, Vol. 27.

Marcel Dekker: New York; 2001. p. 1–66.

[2] Radovic LR, Moreno-Castilla C, Rivera-Utrilla J. Carbon mate-

rials as adsorbents in aqueous solutions. In: Radovic LR, editor.

Chemistry and physics of carbon, Vol. 27. Marcel Dekker: New

York; 2000. p. 227–405.

[3] Lyklema J. Fundamentals of interface and colloid science, Vol. II.

solid–liquid interfaces. New York: Academic Press; 1995.

[4] L�oopez-Ram�oon MV, Stoeckli F, Moreno-Castilla C, Carrasco-

Mar�ıın F. On the characterization of acidic and basic surface sites

on carbon by various techniques. Carbon 1999;37:1215–21.

[5] Lyklema J. Fundamentals of interface and colloid science, Vol. I.

fundamentals. New York: Academic Press; 1993.

[6] Mattson JS, Mark Jr HB. Activated carbon: surface chemistry

and adsorption from solution. New York: Marcel Dekker; 1971.

[7] Pendleton P, Wu SH, Badalyan A. Activated carbon oxygen

content influence on water and surfactant adsorption. J Colloid

Interface Sci 2002;246:235–40.

[8] Pendleton P, Wong SH, Schumann R, Levay G, Denoyel R,

Rouquerol J. Properties of activated carbon controlling 2-meth-

ylisoborneol adsorption. Carbon 1997;35:1141–9.

[9] Coughlin R, Ezra FS. Role of surface acidity in the adsorption of

organic pollutants on the surface of carbon. Environ Sci Technol

1968;2:291–7.

[10] Tamon H, Okazaki M. Desorption characteristics of aromatic

compounds in aqueous solution on solid adsorbents. J Colloid

Interface Sci 1996;179:181–7.

94 C. Moreno-Castilla / Carbon 42 (2004) 83–94

[11] Mahajan OP, Moreno-Castilla C, Walker Jr PL. Surface-treated

activated carbon for removal of phenol from water. Sep Sci

Technol 1980;15:1733–52.

[12] Mattson JS, Mark Jr HB, Malbin MD, Weber Jr WJ, Crittenden

JC. Surface chemistry of active carbon: specific adsorption of

phenols. J Colloid Interface Sci 1969;31:116–30.

[13] Haydar S, Ferro-Garc�ııa MA, Rivera-Utrilla J, Joly JP. Adsorp-

tion of p-nitrophenol on an activated carbon with different

oxidations. Carbon 2003;41:387–95.

[14] Franz M, Arafat HA, Pinto NG. Effect of chemical surface

heterogeneity on the adsorption mechanism of dissolved aromat-

ics on activated carbon. Carbon 2000;38:1807–19.

[15] L�oopez-Ram�oon MV, Stoeckli F, Moreno-Castilla C, Carrasco-

Mar�ıın F. Specific and non-specific interactions of water moleculeswith carbon surfaces from immersion calorimetry. Carbon

2000;38:825–9.

[16] MacDonald JAF, Evans MJB. Adsorption and enthalpy of

phenol on BPL carbon. Carbon 2002;40:703–7.

[17] Moreno-Castilla C, Rivera-Utrilla J, Joly JP, L�oopez-Ram�oon MV,

Ferro-Garc�ııa MA, Carrasco-Mar�ıın F. Thermal regeneration of an

activated carbon exhausted with different substituted phenols.

Carbon 1995;33:1417–23.

[18] Rivera-Utrilla J, Ferro-Garc�ııa MA, Bautista-Toledo I, S�aanchez-

Jim�eenez C, Salvador F, Merch�aan MD. Regeneration of ortho-

chlorophenol-exhausted activated carbons with liquid water at

high pressure and temperature. Water Res 2003;37:1905–11.

[19] Moreno-Castilla C, Rivera-Utrilla J, L�oopez-Ram�oon MV, Carr-

asco-Mar�ıın F. Adsorption of some substituted phenols on

activated carbons from a bituminous coal. Carbon 1995;33:845–51.

[20] Rivera-Utrilla J, Utrera-Hidalgo E, Ferro-Garc�ııa MA, Moreno-

Castilla C. Comparison of activated carbons prepared from

agricultural raw materials and Spanish lignites when removing

chlorophenols from aqueous solutions. Carbon 1991;29:613–9.

[21] Radovic LR, Silva IF, Ume JI, Men�eendez JA, Le�oon y Le�oon C,

Scaroni AW. An experimental and theoretical study of the

adsorption of aromatics possessing electron-withdrawing and

electron-donating functional groups by chemically modified

activated carbons. Carbon 1997;35:1339–48.

[22] Rivera-Utrilla J, Bautista-Toledo I, Ferro-Garc�ııa MA, Moreno-

Castilla C. Bisphenol A removal from water by activated carbon.

Effects of the carbon characteristics and solution chemistry.

Extended abstracts (in CD format) Carbon’03 (publishers Lin-

ares-Solano A, Cazorla-Amor�oos D). Oviedo, Spain. July 2003.

[23] Newcombe G, Drikas M. Adsorption of NOM onto activated

carbon: electrostatic and non-electrostatic effects. Carbon

1997;35:1239–50.

[24] Stoeckli F, L�oopez-Ram�oon MV, Moreno-Castilla C. The adsorp-

tion of phenolic compounds from aqueous solutions by activated

carbons described by the Dubinin–Astakhov equation. Langmuir

2001;17:3301–6.

[25] Li L, Quinlivan PA, Knappe DRU. Effects of activated carbon

surface chemistry and pore structure on the adsorption of organic

contaminants from aqueous solution. Carbon 2002;40:2085–100.

[26] Terzyk AP, Rychlicki G, Biniak S, Lukaszewicz JP. New

correlations between the composition of the surface layer of

carbon and its physicochemical properties exposed while paracet-

amol is adsorbed at different temperatures and pH. J Colloid

Interface Sci 2003;257:13–30.