Water Research 85 (2015) 521e531

Contents lists available at ScienceDirect

Water Research
journal homepage: www.elsevier.com/locate/watres

Predicting the adsorption of organic pollutants from water onto

activated carbons based on the pore size distribution and molecular
connectivity index
Warisa Bunmahotama a, Wei-Nung Hung b, c, Tsair-Fuh Lin a, b, *
a
b
c

Department of Environmental Engineering, National Cheng Kung University, Tainan 70101, Taiwan
Global Water Quality Research Center, National Cheng Kung University, Tainan 70955, Taiwan
Green Energy and Environment Research Laboratories, Industrial Technology Research Institute, Hsinchu 30011, Taiwan

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 10 April 2015
Received in revised form
3 August 2015
Accepted 3 August 2015
Available online 29 August 2015

A new model approach is developed to predict the adsorption isotherms of low-molecular-weight

nonpolar organic compounds (LMWNPOCs) onto activated carbons (ACs). The model is based on the
Polanyi-Dubinin (PD) equation, with the limiting pore volume of adsorbent estimated from the pore size
distribution (PSD) data, and the adsorption afnity of adsorbate described by the molecular connectivity
index (MCI). To obtain the MCI parameters, the model was rst tested for the adsorption of 34
LMWNPOCs primarily on F400 AC from 3 reports. The models t the experimental data well, with only
39.2% of errors. The approach was further employed to predict the adsorption capacity of 40 LMWNPOCs
on F400 AC, 12 LMWNPOCs onto 9 other ACs, and 8 LMWNPOCs onto 5 ACs with unknown PSD, with the
errors of 41.9%, showing the model being reasonable. The model approach may provide a simple means
for predicting adsorption capacities of LMWNPOCs onto different ACs.
2015 Elsevier Ltd. All rights reserved.

Keywords:
Adsorption isotherm
Polanyi-Dubinin model
MCIs
Micropore volume
Organic compounds

1. Introduction
Adsorption with activated carbon (AC) is a common technology
to remove organic contaminants in water treatment. To design the
adsorption process, the equilibrium contaminant adsorption
isotherm has to be estimated in advance. The US Environmental
Protection Administration (USEPA) estimates that there are 15,000
compounds manufactured in the U.S. in quantities greater than
10,000 pounds (USEPA, 2002). Many of these compounds may be
present in different water-treatment systems. It is infeasible to
directly measure the adsorption capacities (or isotherms) for all
compounds of concern with many ACs of different properties,
because such an experimental work would be too costly and timeconsuming (Lee et al., 1981; Corapcioglu and Huang, 1987).
Therefore, a model capable of predicting adsorption isotherms with
reasonable accuracy based on specic properties of organic contaminants and ACs would relieve the experimental work to yield
needed data for process design.

* Corresponding author. Department of Environmental Engineering, National

Cheng Kung University, Tainan 70101, Taiwan.
E-mail address: tin@mail.ncku.edu.tw (T.-F. Lin).
http://dx.doi.org/10.1016/j.watres.2015.08.008
0043-1354/ 2015 Elsevier Ltd. All rights reserved.

Many studies have been directed to construct models for estimating the adsorption capacities of organic contaminants onto ACs.
However, no universal model has been successfully developed for
all chemicals. The current developed models usually estimate the
adsorption of compounds from a specic class based on the functional group (Crittenden et al., 1999), the hydrogen bonding capability (de Ridder et al., 2010), and/or the molecular size (Mezzari,
2006). In light that small-sized hydrocarbons and halogenated
hydrocarbons are of a major concern in water treatment (USEPA,
2014) and in drinking water contamination (USEPA, 2015), a predictive model aimed for estimating the adsorption of these chemicals by ACs would be especially valued.
Prediction of adsorption capacities in pure water system involves both the properties of chemical and ACs. PolanyieDubinin
(PD) equation is one of the models commonly employed to predict
the adsorption isotherms of organic compounds from both vapor
phase (Dubinin and Astakhov, 1971; Kapoor et al., 1989) and
aqueous phase (Manes and Hofer, 1969; Manes, 1998;
Rozwadowski et al., 1989; Stoeckli et al., 2001). The equation is a
combination of adsorption potential theory developed by Polanyi
(1920) and pore lling theory proposed by Dubinin and coworkers (Bering et al., 1966; Dubinin, 1985). The PD equation may
be expressed as

522

W. Bunmahotama et al. / Water Research 85 (2015) 521e531

W Wo exp


bE0

n 
(1)

where
W volume of solute adsorbed (mL/g),
Wo the limiting volume of the adsorption space, i.e., the
micropore volume (mL/g),
E0 the characteristic adsorption energy of the reference
adsorbate (cal/mol),
b afnity coefcient of the characteristic curve (),
n the exponential constant (),
RT ln (Cs/C) in aqueous system (cal/mol),
C the aqueous phase concentration (mg/L),
Cs the aqueous solubility (mg/L),
R the gas constant (1.987 cal/mol/K), and
T the absolute temperature (K).
The PD equation is widely used for adsorption on porous
activated-carbon adsorbents, including the adsorption for gases
with relative pressure >0.0015 (Ozawa et al., 1976; Hung and Lin,
2007), organic vapors with concentrations of 100e10,000 parts
per million (in volume) (Noll et al., 1989), and organic contaminants
in water with concentrations at the mg/L-mg/L level (Crittenden
et al., 1999). The PD equation requires only 3 parameters to determine the adsorption capacity. In the equation, b  Eo may be
considered as one parameter, as it is only dependent on the properties of adsorbates. The other two parameters, Wo and n, are only
relevant to the nature of adsorbents used.
Attempt has been made by many researchers on the estimation
of Wo (Dubinin, 1989; Aukett et al., 1992). Based on the micropore
volume-lling theory, Urano et al. (1982b) found that Wo for vapor
phase adsorption was approximately equal to 0.055 mL plus the
micropore volume for pores with diameter less than 3.2 nm of
different granular activated carbons (GACs). Hung and Lin (2007)
also employed the same approach to obtain the Wo for two ACs,
which allowed the adsorption capacity to be reasonably predicted.
Urano et al. (1982a) employed a similar approach for estimating Wo
of 16 organic compounds onto 5 GACs in aqueous solution. In their
study, a modied Freundlich equation was developed to describe
the adsorption isotherms, and both Wo and the exponential
parameter of the modied Freundlich equation were pore size
dependent. Mezzari (2006) proposed that Wo in aqueous phase,
called the accessible pore volume in their study, may be estimated
by the adsorption curves of reference compounds, such as nitrogen
and carbon dioxide in gaseous phase, and benzene in aqueous
phase. Hsieh and Teng (2000) measured the pore size distribution
(PSD) of 4 GACs with different activation levels and linked the PSD
information with Wo of phenol in water. The above-mentioned
results showed that the PSD data has the potential to be used to
predict the adsorption capacity of organic compounds onto ACs.
The exponent, n, has also been studied extensively in both
gaseous and aqueous phases. Dubinin and Astakhov (1971) showed
that the exponent n varies between 1 and 3, depending on the nature
of adsorbent used. The value of n approaches 3 for adsorbents with
homogeneous micropores, whereas n approaches 1 for adsorbents
with highly heterogeneous micropores. Crittenden et al. (1999)
showed that range of n values for adsorption of organic chemicals
from water on the activated carbons is from 1.05 to 1.61. Condon
(2000a, b) suggested that n values in the PD equation depended on
the methods used to treat the adsorption mechanisms. In case of
heterogeneous adsorbent, the Dubinin and Astakhov (DA) model
may be simplied to Dubinin-Radushkevich (DR) equation (n 2)
when the energy distribution of the surface follows lognormal

energy distribution with a standard deviation (s) 0.25, or to the

Freundlich model (n 1) if s is equal to 0.5 (Condon, 2000a, b).
b is a measure of the similarities of the characteristic of
adsorption between studied and reference adsorbates, and may be
dened as E/E0 (characteristic adsorption energy of the studied
compound to the reference compound) (Dubinin, 1975). A few
approaches have been proposed to link b with the properties of
adsorbates, including parachor (Dubinin, 1960; Vaskovsky, 1950),
polarizability (Dubinin and Sawerina, 1936) and molar volume
(Vaskovsky, 1950; Dubinin and Tomofeyev, 1946).
In aqueous phase adsorption, the net attractive forces involve
the solute, solvent, and the adsorbent. Crittenden et al. (1999)
replaced b  Eo with a new term, called the normalizing factor
(N), to account for the total interaction forces in aqueous system. In
the study, N for different chemicals is described with linear solvation energy relationships (LSERs). LSERs (Kamlet et al., 1977) treat
the solvation effect in solvent-solute systems, and usually four
parameters, including intrinsic molar volume, the polarity/polarizability parameter, the hydrogen-bonding acceptor parameter, and
the hydrogen-bonding donor parameter, were used (Crittenden
et al., 1999; Luehrs et al., 1995). Luehrs et al. (1995) applied the
LSERs to the prediction of partition constants for 353 organic
compounds onto activated carbon from the aqueous phase. However, the errors between predictions and experimental data were
relatively high, usually >100%. Crittenden et al. (1999) applied
LSERs with an improved prediction of the adsorption capacity of 89
organic compounds onto activated carbon by grouping the chemicals into three categories: halogenated aliphatic compounds, aromatics/halogenated
aromatics,
and
poly-functional-group
hydrocarbons. Mezzari (2006) employed quantitative structureeproperty relationships (QSPR) for the description of the afnity
coefcient, b. Among the 14 models they developed, QSPR_4, which
accounted for surface volume scaled, dielectric energy, highest
occupied molecular orbital energy, and electrostatic hydrogen bond
basicity, had the best-simulated results with the data. However, for
the 62 compounds they simulated, four molecular size dependent
sub-models are needed.
PD equation has been used since 1960 (Dubinin, 1960) for the
adsorption in gaseous (Ozawa et al., 1976; Hung and Lin, 2007) and
aqueous systems (Crittenden et al., 1999). Most studies link the
adsorption data with chemical properties, such as molar volume
(Manes and Hofer, 1969; Arbuckle, 1981; Manes and Wohleber,
1971; Wohleber and Manes, 1971; Kuennen et al., 1989), polarizability (Dubinin and Sawerina, 1936), parachor (Dubinin, 1960
Vaskovsky, 1950), and more recently LSERs (Crittenden et al.,
1999). Only few studies investigated the inuence of AC properties on the adsorption capacity (Urano et al., 1982b; Hsieh and Teng,
2000). Although attempts have been made to predict the adsorption capacity of organic compounds using purely the properties of
ACs and chemicals (Li et al., 2005; Zhang et al., 2013), only very
limited numbers of chemicals and ACs were examined and veried.
In this study, a new approach was proposed to estimate the
adsorption capacity of a compound onto AC using simply the
properties of chemicals and the AC pore size distribution data. All
predicted parameters were either obtained from literature or from
chemical database and no advanced experiments are needed. The
approach was validated with extensive adsorption data of 42
organic chemicals onto 15 different activated carbons from 12
published reports.
2. Methods
2.1. Modeling approach
The PD equation (Dubinin, 1960) modied by Crittenden et al.

W. Bunmahotama et al. / Water Research 85 (2015) 521e531

523

(1999) was adopted in this study, as shown below,

h  n i
W Wo exp 
100N

(2)

where N is a normalizing factor and 100 is a scaling factor.

To be able to predict the adsorption capacity, W, three parameters need to be determined in advance, i.e., Wo, n, and N. Generally,
Wo for a target AC was obtained from the regression of isotherm
data, which was also employed in Crittenden et al. (1999). Although
this approach may provide a precise estimation of Wo for a specic
GAC/chemical system, experiments are needed to obtain the
isotherm data. In this study, the approach suggested by Urano et al.
(1982b) was applied. Urano et al. (1982b) and Hung and Lin (2007)
suggested that Wo can be estimated from PSD of GACs, as described
in equation (3)

Wo 0:55 mL=g V3:2

(3)

where V3.2 is the micropore volume for pores with diameter less
than 3.2 nm for GAC. Although Urano et al. (1982a) developed
another correlation for estimating Wo of GACs in aqueous phase,
both Wo and adsorption afnity depended on PSD, making the
correlation not applicable in this study.
For the evaluation of n, the isotherms of 31 non-polar halogenated aliphatic compounds and aromatics/halogenated aromatics,
according to Crittenden et al. (1999), are rst selected. The data of
these 31 compounds were all from Speth and Miltner (1990) for a
GAC from Calgon (F400). In Crittenden et al. (1999), n 1.05e1.2
was obtained for three studied chemical groups. Therefore, we
estimate Wo in the current study by setting n 1, 1.1, and 1.2 and
using the same N as used in Crittenden et al. (1999). The best-tted
Wo of the 31 chemicals are 0.422 mL/g (R2 0.866), 0.466 mL/g
(R2 0.901) and 0.500 mL/g (R2 0.909), for the case of n 1.0, 1.1,
and 1.2, respectively. As shown in Fig. 1a and the correlation coefcients (R2), the isotherm equations tted the data better for the
cases of n 1.1 and 1.2. The obtained Wo for n 1.1 (0.466 mL/g) is
very similar to that obtained experimentally for benzene onto F400
by Mezzari (2006) (0.457 mL/g) and that estimated from equation
(3) (0.465 mL/g), suggesting that the present approach may provide
a simple means to calculate Wo for GAC.
To further justify the best n for the model, the adsorbed volumes
per unit mass of adsorbent (W, volume per gram) for the 31
chemicals were further analyzed. The experimental W values for
the 31 compounds were calculated from the experimental isotherms of Speth and Miltner (1990). Note that the concentration of
each compound was set at 100 mg/L except that p-xylene was at
10 mg/L, as the experimental concentration range is more limited for
p-xylene. In predicting the W, the PSD approach (equation (3)) with
Wo 0.465 mL/g was used. In addition, the N as in Crittenden et al.
(1999) was also adopted. Fig. 1b shows a comparison of experimental and predicted W for the three cases of n. To compare the
difference between model predictions and experimental data, the
sum of errors (SE) as derived from equation (4),

SE

Wp  We
W2e

2 !#1 2
=

"

(4)

is also obtained, where Wp and We are predicted and experimental

adsorbed volume (W), respectively.
Fig. 1b illustrates the correlations between predicted and
observed adsorption capacities for the 31 compounds with
different n values. The calculated SE between the models and data
were 4.322, 2.511 and 3.565, for the cases of n 1.0, 1.1, and 1.2,
respectively. The gure and SE both suggest that n 1.1 gives the

Fig. 1. Correlation of (a) Wo and (b) n for 31 compounds onto F400 GAC. [bromodichloromethane, bromoform, carbon tetrachloride, chloroform, dibromochloromethane,
dibromomethane,
1,1-dichloroethane,
1,2-dichloroethane,
1,2dichloropropane, 1,3-dichloropropane, methylene chloride, 1,1,1,2-tetrachloroethane,
1,1,1-trichloroethane, 1,1,2-trichloroethane, 1,2,3-trichloropropane, 1,2-dibromoethane,
1,1-dichloroethene, cis-1,2-dichloroethene, tetrachloroethene, trichloroethene, odichlorobenzene, p-dichlorobenzene, ethylbenzene, styrene, toluene, p-xylene, benzene, bromobenzene, chlorobenzene, hexachlorocyclopentadiene and lindane].

best prediction. In addition, at n 1.1, the extracted Wo (0.466 mL/

g) is almost identical to that obtained based on the PSD calculation
(0.465 mL/g), showing that PSD may be used to estimate Wo. Unlike
the approach proposed by Crittenden et al. (1999), in which n may
be inuenced by Wo and N, our results showed that n may be
considered as a constant for the tested conditions.
For the third parameter in the PD-DA model, the normalizing
factor, N, is dependent on the adsorbate properties (Polanyi, 1932),
and is estimated based on molecular connectivity indices (MCIs),
developed by Kier and Hall (1976). MCIs explain molecular structures based on the bonding and branching patterns of molecules. To
obtain MCIs for a specic compound, its three dimensional molecular structure needs to be known in advance. In this study, the
software PaDELeDescriptor developed by National Singapore University (Yap, 2011) was employed to obtain the molecular descriptors of the studied chemicals. Once the descriptors were
obtained, they were plugged into the MCI analysis. During the
analysis, both simple indices, which encode sigma-bonded electrons, and valence indices, which include sigma, pi and lone electrons are considered in the analysis. The MCI values used in this
study are listed in Supplementary Tables S1eS4. The N's used in the

524

W. Bunmahotama et al. / Water Research 85 (2015) 521e531

regression were obtained from tted isotherms for the 31 compounds, using the PD-DA equation (equation (2)), Wo from the PSD
approach (equation (3)), and the n value of 1.1.
Fig. 2 shows the regression results, expressed as R2 and the sum
of errors (SEs) between the experimental and tted N's for different
pairs of MCIs. For all the four cases of n's, it is clearly shown in the
gure that increasing MCI pairs results in better tting of the N's, as
more details of the compounds are provided. Therefore, the N's
were best described with 15 MCIs pairs (till Cluster 6), and the
extracted regression constants are listed in Supplementary
Table S5. As the SEs and R2 for n 1.1 (0.080 and 0.997) and
n 1.2 (0.073 and 0.997) are very similar for the data described
with 15 MCIs pairs (cluster 6), n 1.1 was selected due to the
agreement of Wo with the experimental data observed in the previous section.
As suggested in Crittenden et al. (1999), the percent sample
deviation (SDEV), based on the relative error between experimental
data and the correlation, was used to describe the tting between
the model and experimental data, which is expressed as

v
u
2 3

u2P W
correlation  Wdata
u
u6
Wdata
7
% sample deviation SDEV u
5  100
t4
Ndata  1
(5)
where Wcorrelation and Wdata are the volume adsorbed determined
from the correlation and the experimental data for various
aqueous-phase concentrations, respectively, and Ndata is the number of data.
2.2. Isotherm data and physicalechemical properties of adsorbates
To evaluate the model predictions, the adsorption isotherm data
in deionized water from 12 reports, including Mezzari (2006),
Speth and Miltner (1990, 1998), Urano et al. (1991), El-Dib and
Badawy (1979), Shih and Gschwend (2009), Erto et al. (2011),
Hindarso et al. (2001), Huang et al. (2011), Parker (1995), Alben
et al. (1988) and Hand et al. (1985), were used. The information of
these adsorption isotherms data is listed in Table 1. The molecular
weight (MW), water solubility, chemical density and 3D molecular
structure for the chemicals examined were obtained forChemSpider, 2013 (ChemSpider), IPCS INCHEM, 2013 (IPCS
INCHEM) and ChemicalBook, 2013 (ChemicalBook) online databases. An open source software, PaDELeDescriptor software
version 2.9 (Yap, 2011), combined with the 3D structure is used to
obtain the MCI parameters of tested organic compounds.
For some of the ACs studied, the PSD information is not available
in the literature. Therefore, their PSD's were analyzed using low
temperature nitrogen adsorption/desorption data at 77.4 K determined with a Micromeritics ASAP2010. The PSD data calculated
from the nitrogen adsorption isotherm using the BJH method
(Barrett et al., 1951) are listed in Supplementary Table S6. All the
chemicals examined in this study, including 40 compounds onto
F400 GAC, 12 onto 9 other GACs and 8 onto 5 additional GACs with
unknown pore size distribution, are listed in Supplementary
Table S7.
2.3. Model calibration and modication
As indicated in Supplementary Tables S8 and S9, the 42 compounds tested cover the chemicals with carbon number (C) 1 to 7,
with one ring and 1e3 chlorine and/or bromine. However, the 31
compounds used in the original model development (Version 1) for

correlating N did not cover 6 or 7 carbon hexane and heptane, or

halogenated hydrocarbons with a combination of chlorine and
bromine for C 3. Therefore, dibromochloropropane taken from
Speth and Miltner (1990), and hexane and heptane obtained from
Shih and Gschwend (2009) for Darco, were also included in the
regression analysis of N. The comparison between the two versions
of models, also shown in Supplementary Table S8, indicated that
Version 2 model is able to better describe the N's for the 34 training
compounds. Fig. 3a also shows that Version 2 model ts the data for
the 34 training compounds reasonably well, with 39.1% SDE.
Therefore, Version 2 model was used for further predictions.
3. Results and discussion
3.1. Predictions of the adsorption capacity on F400 activated carbon
Fig. 3a shows the model prediction of the adsorption capacities
of 40 small, non-polar organic compounds onto F400 GAC. It is
noted that the experimental data were extracted from 8 reports
(Speth and Miltner, 1990, 1998; Urano et al., 1991; El-Dib and
Badawy, 1979; Huang et al., 2011; Parker, 1995; Alben et al., 1988;
Hand et al., 1985). In the model, the PD-DA equation (equation (2))
was used, with Wo of 0.465 mL/g obtained from the PSD approach
(equation (3)) for F400, and n value of 1.1 as discussed in the Model
Approach Section. The tted equations for N with MCI are listed in
Table 2 and Supplementary Table S5.
Fig. 3a and b demonstrates that the models account for the
experimental data reasonably well, with only 43.1% SDEV.
Crittenden et al. (1999) and de Ridder et al. (2010) developed
models to simulate the adsorption data collected by Speth and
Miltner (1990, 1998), with the observed difference between
model and data being 36.1% and 54.3% SDEV, respectively. Our
model shows similar degree of predicted results, 39.8% SDEV, when
the same set of experimental data was used. Even if 6 more reports
were included in our predictions, the models still give good predictions of the experimental data as shown in Fig. 3b.
Crittenden et al. (1999) and de Ridder et al. (2010) developed
models to simulate the adsorption capacities of various types of
organic compounds onto F400 GAC. The former study separated 40
compounds studied by Speth and Miltner (1990) into 3 chemical
groups, including aliphatic/halogenated (20 compounds), aromatic/
halogenated (10), and polyfunctional groups (10). The PD-LSER
model they developed was able to describe the adsorption capacity of the chemicals. However, three sub-models with distinct Wo, n,
and N were needed for the compounds. When extrapolating the
models to other compounds listed in Speth and Miltner (1990) but
not included in their correlation, including o-chlorotoluene, pchlorotoluene, dibromochloropropane, trans-1,2-dichloroethene,
1,1-dichloropropene and 1,3,5-trichlorobenzene, the models also
show good predictions, with SDEV of between 0 and 40%. However,
in each of the sub-model, Wo was obtained from the correlation of
experimental data, which complicates the adsorption prediction for
other GACs because the experimental data may not be available.
de Ridder et al. (2010) developed models to describe the
adsorption capacity of F400 GAC for 71 out of 72 compounds obtained from Speth and Miltner (1990, 1998), in which the three
benzene ring compound, diquat, was not considered. Based on
regression analysis, three parameters were considered in the
models, equilibrium aqueous concentration, solute hydrophobicity
and polarizability. In attempt to improve the predictive ability, the
correlations were separated into four groups, i.e. aromatic with no
hydrogen bond donor and acceptor (d/a), aromatic with hydrogen
bond d/a, aliphatic with hydrogen bond d/a, and aliphatic with
hydrogen bond d/a. Although only simple parameters are needed
for the models, the models were based purely on the regression of

1.4
1.2
1.0
0.8
0.6
0.4
0.2

2
R = 0.363
2
R = 0.767
2
R = 0.797
2
R = 0.880
2
R = 0.953
2
R = 0.957
2
R = 0.980
2
R = 0.982
2
R = 0.982
2
R = 0.986
2
R = 0.992
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997

W. Bunmahotama et al. / Water Research 85 (2015) 521e531

(a) n = 1

1.2
1.0
0.8
0.6

0.2
0.0

1.0
0.8
0.6
0.4
0.2

2
R = 0.362

1.2

(c) n = 1.2

2
R = 0.764
2
R = 0.794
2
R = 0.878
2
R = 0.955
2
R = 0.960
2
R = 0.980
2
R = 0.982
2
R = 0.982
2
R = 0.986
2
R = 0.992
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997

1.4

1.0
0.8
0.6
0.4
0.2

2
R = 0.905
2
R = 0.936
2
R = 0.939
2
R = 0.941
2
R = 0.968
2
R = 0.972
2
R = 0.975
2
R = 0.977
2
R = 0.986
2
R = 0.986
2
R = 0.990
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997

1.2

(d) n = 1.3

2
R = 0.799
2
R = 0.800

1.4

2
R = 0.372

0.0

adsorption data for a specic GAC, F400. No information about the

properties of GAC was included in the analysis and therefore, in the
models, the parameters are only related to chemical properties of
adsorbates. This prohibits the extrapolation of the models to other
ACs, unless a large set of experimental adsorption data are available
for a new model regression.
Table 3 listed all the N values obtained from tting of the
experimental isotherms and from calculation based on the MCI
model shown in Table 2. It is noted that the discrepancy between
the tted N and calculated N is very small for the studied compounds, all <5%, suggesting the models conform to the data very
well. The N values of the studied compounds, in range of
12.88e25.95, increase with the number of carbon, chlorine, and
bromine atoms, and with the number of double bonds and aromatic
rings.
Attempts were made to simplify the MCI model by reducing the
number of possible parameters that would affect adsorption afnity. A regression analysis was conducted by correlating the tted
Ns with the properties of the studied compounds, including
numbers of C, Cl, and Br atoms, and numbers of double bonds and
benzene rings. In addition, a few parameters commonly used to link
with adsorption afnity, including molar volume (Vaskovsky, 1950;
Dubinin and Tomofeyev, 1946), refractive index (Manes and Hofer,
1969) and Vi/100, p*, b, and a (Crittenden et al., 1999) were also
included in the analysis. The input parameters and correlation results are shown in Supplementary Tables S9 and S10, respectively.
For all the compounds tested, the number of double bonds, benzene
rings, C atoms and refractive index are positively correlated with N
values, with R2 0.685, 0.576, 0.438 and 0.258, respectively. As the
correlations are not very high, it is difcult to simplify the MCI
model with one of the studied parameters.
Another analysis of N was further conducted using multiple
parameter correlation by combining all the studied parameters
listed in Supplementary Table S10. A combination of the ve molecular parameters, including number of carbon, chlorine, and
bromine atoms, and number of double bonds and benzene rings
give the high correlation, with R2 0.751 (Supplementary
Table S11). Compared with the high correlation between the
tted N and calculated N obtained from the MCI model (R2 0.992),
the simplied correlation is still not enough to provide good
description of adsorption afnity. Therefore, the MCI approach,
although with more parameters, may still be a better model to
describe and predict the adsorption of organic compounds onto
ACs.
3.2. Applications of the models to other ACs
The PD-MCI model was further applied to predict the adsorption
capacity of 12 small, non-polar organic compounds onto 9 ACs,
including Turumi HC-30, Hokuetu Y-20, Mitubisi 005-S, Mitubisi
007-S, and Fujisawa ACW (Urano et al., 1991), Darco (Shih and
Gschwend, 2009), and F600 and G219 (Mezzari, 2006). Note that
in predicting the adsorption capacity, Wo's were estimated from
PSD, as shown in Supplementary Table S6, for 8ACs, in which the
PSD for 3ACs (F400, F300, and Darco) were measured in this current
study, and those for 5 other ACs, including Turumi HC-30, Hokuetu

0.0
pa
th
pa 0
th
pa 1
th
pa 2
th
pa 3
th
pa 4
th
pa 5
th
pa
th p 6
pa clu ath
th s 7
p a clu ter
th st 4
clu er
s 5
clu ter
s 6
clu ter
ste 3
cl r
us 4
clu ter
st 5
ch er 6
a
ch in 3
a
ch in 4
a
ch in 5
a
ch in 6
ain
7

Sum of errors

0.4

(b) n = 1.1

2
R = 0.766
2
R = 0.795
2
R = 0.879
2
R = 0.954
2
R = 0.958
2
R = 0.980
2
R = 0.982
2
R = 0.982
2
R = 0.986
2
R = 0.992
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997
2
R = 0.997

1.4

2
R = 0.362

0.0

525

MCI order

Fig. 2. Correlation of MCI order for 31 compounds onto F400 GAC under different n
values. [bromodichloromethane, bromoform, carbon tetrachloride, chloroform, dibromochloromethane, dibromomethane, 1,1-dichloroethane, 1,2-dichloroethane, 1,2dichloropropane, 1,3-dichloropropane, methylene chloride, 1,1,1,2-tetrachloroethane,
1,1,1-trichloroethane, 1,1,2-trichloroethane, 1,2,3-trichloropropane, 1,2-dibromoethane,
1,1-dichloroethene, cis-1,2-dichloroethene, tetrachloroethene, trichloroethene, odichlorobenzene, p-dichlorobenzene, ethylbenzene, styrene, toluene, p-xylene, benzene, bromobenzene, chlorobenzene, hexachlorocyclopentadiene and lindane].

526

Table 1
Summary of literature and adsorption isotherms used in this study.
References
Hindarso et al.
(2001)

Huang et al.
(2011)

Parker
(1995)

Alben et al.
(1988)

Hand et al.
(1985)

37

21
-b
25

-a
-b
24

-a
6.3 and 7.9
24

148
e
25

28
e
25 2

78
e
24 1

36
7
20

108
e
30, 40, 50

10
7.3
25

Carbon type

Filtrasorb 600
G 219

Filtrasorb 400

Filtrasorb 400

Turumi HC e 30
Hokuetu Y e 20
Mitubisi 005 e S
Mitubisi 007 e S
Filtrasorb 400
Fujisawa ACW

Filtrasorb 400

Darco

Aquacarb
207 EA

TOG

Mesh size
Classes of
chemicals

e
Halogenated
aliphatic

100e200
Wide range
including
environmental
pollutants

100e200
Single benzene
ring

e
Halogenated
aliphatic

e
20e40
Single benzene Wide range
ring
including
environmental
pollutants

Number of
studied
compounds
Data points
pH
Temp ( C)

a
b

Isotherms were given in the reports.

not specied in the reports.

4
e
Room
temperature
Filtrasorb 400 Filtrasorb
Filtrasorb 300 400

e
20 e 50
e
Halogenated Single benzene Halogenated
aliphatic
ring
aliphatic

178
e
e
6
4, 15, 30, 45 10e22
Filtrasorb
400

Filtrasorb 400
Westvaco
(WV eG)
Westvaco
(WVeW)
Darco
(HD3000)
e
12e40
e
Halogenated Halogenated Halogenated
aliphatic
aliphatic
aliphatic

W. Bunmahotama et al. / Water Research 85 (2015) 521e531

Mezzari (2006) Speth and Miltner Speth and Miltner Urano et al. (1991) El-Dib and
Shih and
Erto et al.
(1990)
(1998)
Badawy (1979) Gschwend (2009) (2011)

W. Bunmahotama et al. / Water Research 85 (2015) 521e531

527

et al. (2001) studied the adsorption of benzene and toluene onto

a GAC, called TOG, at three difference temperatures (30, 40, and
50  C) and the PSD was not reported. In determining Wo for TOG,
the isotherm of toluene at 40  C was tted with the D-R equation.
Good tting between the equation and data was found, with
extracted Wo being 0.495 mL/g and R2 of 0.742. The extracted Wo
was then combined with the MCI model to predict all other
isotherm data for TOG.
Two more cases of adsorption data with unknown PSD of GACs
are also tested for this approach, including Erto et al. (2011) and
Hand et al. (1985). The former reported the adsorption of 2 chloroethenes onto a GAC, Aquacarb 207EA, and the latter included 6
halogenated hydrocarbons onto 3 GACs, WV-G and WV-W of
Westvaco and HD3000 of Darco. In determining Wo of the GACs, the
isotherm of trichloroethene (TCE) on each GAC was rst tted to
extract the corresponding Wo, and combined with the MCI
approach to predict adsorption capacities for other compounds.
Fittings of the TCE isotherms to the D-R equations give good correlations for all the 4 GACs, with Wo (R2) being 0.618 (0.916), 0.534
(0.990), 0.411 (0.996) and 0.232 (0.999) mL/g for Aquacarb 207EA,
WV-G, WV-W and HD3000, respectively.
Fig. 5 shows that the models well predict all the experimental
data collected from the three studies (Erto et al., 2011; Hindarso
et al., 2001; Hand et al., 1985), with only 31.4% SDEV. The good
tting between model predictions and experimental data further
demonstrates the applicability of the model. If pore size distribution is not known, for PD-MCI model predictions, only one set of
experiment is needed to obtain Wo. The extracted Wo can then be
incorporated with MCI to predict the adsorption capacity of other
organic compounds.
3.4. Applications and limitations of the model

Fig. 3. Comparison of adsorption capacity between models and measured data for
F400 GAC, where (a) is model ts for 34 training compounds, and (b) is model predictions for 40 organic compounds. Inner gure in (b) is the best predicted (chloroform
from Alben et al. (1988) and trichloroethene from Hand et al. (1985) and worst predicted cases (p- chlorotoluene from Speth and Miltner (1990) and chloroform from
Urano et al. (1991)).

Y-20, Mitubisi 005-S, Mitubisi 007-S, and Fujisawa ACW, were obtained from Urano et al. (1991). For 2 other ACs, F600 and G219, the
Wo's were assumed to be the same as total accessible pore volume
(Vadsorbed, max) reported by Mezzari (2006), because their Vadsorbed,
max of F400 (0.471 and 0.457 ml/g) is similar to our Wo of F400
(0.465 ml/g). Table 2 summarizes the sources of experimental data
and the parameters used (N and Wo) in the model predictions.
As shown in Fig. 4, the models follow the experimental data
excellently, with only 37.3% SDEV on the average. The good ts
between model predictions and experimental data for different
adsorbate/AC combinations from different reports prove that the
PSD-MCI model is able to predict adsorption capacities of the
studied small, non-polar organic compounds onto other ACs.

3.3. Applications to other ACs without pore size distribution

information
On many occasions, the properties of ACs may not be available,
making the determination of Wo from pore size distribution not
possible. Under this condition, Wo may be estimated from a reference compound by tting its isotherm to the D-R equation, as
suggested by Mezzari (2006) and Hung and Lin (2007). Hindarso

The PD-MCI model developed in this study has been successfully

tested in simulation and prediction of the adsorption capacity for
40 low molecular non-polar organic chemicals onto F400 GAC. The
model was also extrapolated to the prediction of adsorption of 12
chemicals onto 9 other GACs, with good agreement between model
estimates and experimental data. In summary, the model was
tested by the adsorption data collected from 12 reports for 42
organic chemicals onto 15 different activated carbons, proving that
the model is applicable to different chemical/GAC combinations.
In developing the model, three key parameters were determined, Wo, n, and N. The best n was found to be to 1.1 for the
chemical/GAC combinations in this study. The limiting adsorption
volume Wo was proved to be linked to pore size distribution of AC
(equation (3)), while N was estimated from MCI parameters listed
in Table 2. With all these three parameters, the adsorption capacity
of a given low-molecular non-polar compound onto a specic AC
may be predicted purely based on the chemical properties (MCIs) of
the adsorbate and the PSD of the AC. Although previous studies,
such as Crittenden et al. (1999), de Ridder et al. (2010) and Shih and
Gschwend (2009), have developed models to improve the prediction of the organic compound adsorption on GACs, their models
have to be divided to several sub-models for different groups of
organic compounds. In addition, their models were limited to only
one AC. Extrapolation of the models developed by Crittenden et al.
(1999), de Ridder et al. (2010) and Shih and Gschwend (2009), to
other GACs are not possible, as no carbon properties are considered
in the models. The PSD-MCI model proposed in this study incorporated the AC pore size information for estimation of Wo for
different ACs, provided the PSD information is available, for
extending the model performance.
Although the present PD-MCI model has been found successful
for treating many combinations of organic chemicals and ACs, the

528

Wo

Mezzari (2006)

Speth and
Miltner
(1990, 1998)

Urano et al.
(1991)

El-Dib and
Badawy
(1979)

Shih and
Gschwend
(2009)

Erto et al.
(2011)

Hindarso et al.
(2001)

Huang et al.
(2011)

Parker (1995)

Alben et al.
(1988)

Hand et al.
(1985)

0.449 (Filtrasorb 600)

0.682 (G 219)

0.465

0.505a (Turumi HC e 30)

0.415a (Hokuetu Y e 20)
0.525a (Mitubisi 005 e S)
0.475a (Mitubisi 007 e S)
0.465 (Filtrasorb 400)
0.415a (Fujisawa ACW)

0.465

0.270

0.618b

0.495b

0.465 (Filtrasorb 400)

0.422 (Filtrasorb 300)

0.465

0.465

0.465 (Filtrasorb 400)

0.534b (Westvaco: WV eG)
0.411b (Westvaco: WVeW)
0.232b (Darco: HD3000)

n
N

1.1
28.432 e 7.861*0cp 10.685*3cp e 0.466*4cp e 1.26*5cp 8.647*6cp 31.662*7cp 1.147*0cpv 2.387*1cpv  2.053*2cpv 0.443*3cpv 8.767*4cpv 0.352*5cpv e 7.383*6cpv e 75.587*7cpv e 2.725*4cpc e
0.517*6cpc 2.79*4cvpc e 4.282*5cvpc 12.415*3cc e 41.776*4cc 13.796*5cc 1.697*3cvc 10.895*4cvc e 17.199*5cvc  44.694*6cvc
35.867
57.808
16.926
41.934 (Filtrasorb 400) 48.273
48.038
34.934 (Filtrasorb 400)
%SDEV 63.048 (Filtrasorb 600) 39.808
45.898 (Turumi HC e 30) 31.820
37.535 (Filtrasorb 300)
31.887 (Westvaco: WV eG)
21.942 (G 219)
33.990 (Hokuetu Y e 20)
22.705 (Westvaco: WVeW)
28.443 (Mitubisi 005 e S)
35.106 (Darco: HD3000)
22.318 (Mitubisi 007 e S)
65.320 (Filtrasorb 400)
34.396 (Fujisawa ACW)
a
b

Obtained from pore size distribution data using Equation (3).

Fitted with one isotherm.

W. Bunmahotama et al. / Water Research 85 (2015) 521e531

Table 2
Summary of the parameters used in the PD-MCI model.

W. Bunmahotama et al. / Water Research 85 (2015) 521e531

Table 3
The tted and calculated N's for all the 42 compounds studied.
Compounds

Chemical class

Fitted N

Calculated N

%Error

Methylene chloride
Chloroform
Bromodichlomethane
Dibromochloromethane
Dibromomethane
Bromoform
Carbon tetrachloride
1,1-dichloroethane
1,2-dichloroethane
1,2-dibromoethane
1,1,1-trichloroethane
1,1,2-trichloroethane
1,1,1,2-tetrachloroethane
1,1-dichloroethene
cis-1,2-dichloroethene
Trichloroethene
Tetrachloroethene
1,2-dichloropropane
1,3-dichloropropane
1,2,3-trichloropropane
Dibromochloropropane
Hexane
Heptane
Hexachlorocyclopentadiene
Lindane
Benzene
Chlorobenzene
Bromobenzene
o-dichlorobenzene
p-dichlorobenzene
Toluene
p-xylene
Ethylbenzene
Styrene
prediction
tran-1,2-dichloroethene
1,1-dichloropropene
m-dichlorobenzene
o-xylene
m-xylene
o-chlorotoluene
p-chlorotoluene
1,3,5-trichlorobenzene

C1

11.389
14.143
13.888
15.021
13.805
17.010
14.225
13.668
13.618
16.125
12.500
16.726
17.611
13.425
13.860
14.825
16.310
15.738
15.811
18.336
21.870
13.000
13.021
17.212
13.450
19.246
22.636
24.440
25.950
25.230
21.792
23.762
24.361
23.898

12.525
13.587
14.497
15.356
13.498
16.725
14.223
13.440
13.567
16.000
12.875
16.532
17.075
12.820
13.040
15.667
16.272
16.212
15.250
18.417
22.051
13.192
13.069
17.204
13.457
19.203
22.470
24.434
25.951
24.722
22.153
24.100
24.146
24.081

0.82
0.17
0.18
0.05
0.05
0.03
0
0.03
0
0.01
0.09
0.01
0.10
0.22
0.40
0.29
0
0.09
0.14
0
0.01
0.02
0
0
0
0
0.01
0
0
0.04
0.03
0.02
0.01
0.01

e
e
e
e
e
e
e
e

13.040
15.309
25.517
26.327
24.832
26.171
24.411
28.117

e
e
e
e
e
e
e
e

C3

C6
C7
Ring

C2
C3
Ring

PD - MCI model, % SDEV = 31.3

Fitted ( ) and predicted ( ) Erto et al. (2011) - Aquacarb 207EA
Fitted ( ) and predicted ( ) Hindarso et al. (2001) - TOG
Fitted ( ) and predicted ( ) Hand et al. (1985) - Westvaco (WV -G),
Westvaco (WV -W) and Darco (HD3000)

log qe measured (mol/g)

0
0

log qe predicted (mol/g)

Fig. 5. Model ts and predictions of adsorption capacity for 8 organic compounds onto
other 5 GACs with unknown pore size distribution.

model still has limitations. Similar to the correlation models

developed by Crittenden et al. (1999), de Ridder et al. (2010), and
Shih and Gschwend (2009), the effectiveness of the present model
is limited mainly for small-sized nonpolar compounds. It is noted
that a large discrepancy occurs between the PD-MCI model predicted and experimental adsorption capacities for large-sized
compounds, such as aromatics with multiple benzene rings and
hydrocarbons with more than 8 carbons in a chain, and those polar
compounds with oxygen and nitrogen. Therefore, more model inputs are needed to extend the adsorption prediction for these
chemicals.
Another limitation of the model is that it is only successfully
tested in thermo-activated ACs. Attempts have been made to predict chemically activated GAC. However, large discrepancy between
model predictions and experimental observations was found, as
indicated in Supplementary Fig. S1 for WVB, an acid activated wood
based AC from Westvaco (Kilduff et al. (2002); Rahman (2014)). The
oxidized surface of AC may have effect on the adsorption capacity,
even when the PSDs are the same for different ACs (Dastgheib and
Karanl, 2005). In addition, potential size exclusion effects of prohibiting large-molecule adsorbates to enter microporous pores of
GAC, as described in Hung and Lin (2007), was not considered in
this study. Application of the developed model to chemically activated ACs and AC/adsorbate combinations with potential size
exclusion effects should be further tested. Finally, to be able to
apply the model in water treatment processes, effect of natural
organic matters (NOMs) on the adsorption of the targeted chemicals should also be considered.

PD - MCI model, % SDEV = 37.3

Mezzari (2006) - Filtrasorb 600 and G 219
Urano et al. (1991) - Turumi HC - 30, Hokuetu Y - 20, Mitubisi 005 - S,
Mitubisi 007 - S and Fujisawa ACW
Shih and Gschwend (2009) - Darco
Huang et al. (2011) - Filtrasorb 300

4
log qe measured (mol/g)

C2

529

4. Conclusions
2

0
0

log qe predicted (mol/g)

Fig. 4. Model predictions of adsorption capacity for 12 organic compounds onto other
9 GACs.

In this study, a model based on the PD equation, the AC pore size

distribution, and the MCI of targeted chemicals, has been developed for estimating the AC adsorption capacities of small nonpolar
organic compounds. Two key parameters are needed in the model
predictions, Wo to be determined by the Urano et al. (1982b)
approach along with the pore size distribution and N to be determined by the MCI approach correlated from adsorption data
available for F400 GAC. The model parameters were evaluated
based on 34 low-molecular non-polar organic compounds primarily on F400 GAC from 3 reports. The extracted parameters were
further employed to predict the adsorption capacity obtained from

530

W. Bunmahotama et al. / Water Research 85 (2015) 521e531

8 reports for 40 compounds on F400 GAC. As the properties of GACs

and adsorbates are considered in the model, the model was successfully extended to predict the adsorption of 12 compounds onto
9 other GACs. The model may provide a simple approach for the
prediction of adsorption capacity for low-molecular non-polar
compounds onto different ACs, provided that the PSD is available.
Further studies are suggested in order to test and extend the model
to the adsorption of other chemical groups, such as polar organic
and multiple benzene ring compounds, and under natural and
engineered water systems.
Acknowledgments
This work was supported in part by the Headquarters of University Advancement at the National Cheng Kung University
sponsored by the Ministry of Education, in part by Industrial
Technology Research Institute Project (DF55RN8000), and in part
by Ministry of Science and Technology (NSC 101-2221-E-006-152MY3), Taiwan, ROC.
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.watres.2015.08.008.
References
Alben, K.T., Shpirt, E., Kaczmarczyk, J.H., 1988. Temperature dependence of
trihalomethane adsorption on activated carbon: implications for systems with
seasonal variations in temperature and concentration. Environ. Sci. Technol. 22
(4), 406e412.
Arbuckle, W.B., 1981. Estimating equilibrium adsorption of organic compounds on
activated carbon from aqueous solution. Environ. Sci. Technol. 15 (7), 812e819.
Aukett, P.N., Quirke, N., Riddiford, S., Tennison, S.R., 1992. Methane adsorption on
microporous carbons e a comparison of experiment, theory, and simulation.
Carbon 30 (6), 913e924.
Barrett, E.P., Joyner, L.G., Halenda, P.P., 1951. The determination of pore volume and
area distributions in porous substances. I. computations from nitrogen isotherms. J. Am. Chem. Soc. 73 (1), 373e380.
Bering, B.P., Dubinin, M.M., Serpinsky, V.V., 1966. Theory of volume lling for vapor
adsorption. J. Colloid Interface Sci. 21 (4), 378e393.
ChemicalBook, ChemicalBook Inc. http://www.chemicalbook.com. (downloaded on
22, August, 2013, 2013).
ChemSpider, 2013. Royal Society of Chemistry, Thomas Graham House. Science
Park, Milton road, Cambridge downloaded on 22, August, 2013. http://www.
chemspider.com/.
Condon, J.B., 2000a. Equivalency of the Dubinin e Polanyi equations and the QM
based sorption isotherm equation. A. mathematical derivation. Microporous
Mesoporous Mater. 38 (2e3), 359e376.
Condon, J.B., 2000b. Equivalency of the Dubinin e Polanyi equations and the QM
based sorption isotherm equation. B. Simulations of heterogeneous surfaces.
Microporous Mesoporous Mater. 38 (2e3), 377e383.
Corapcioglu, M.O., Huang, C.P., 1987. The adsorption of heavy metals onto hydrous
activated carbon. Water Res. 21 (9), 1031e1044.
Crittenden, J.C., Sanongraj, S., Bulloch, J.L., Hand, D.W., Rogers, T.N., Speth, T.F.,
Ulmer, M., 1999. Correlation of aqueous-phase adsorption isotherms. Environ.
Sci. Technol. 33 (17), 2926e2933.
Dastgheib, S.A., Karanl, T., 2005. The effect of the physical and chemical characteristics of activated carbons on the adsorption energy and afnity coefcient of
Dubinin equation. J. Colloid Interface Sci. 292 (2), 312e321.
de Ridder, D.J., Villacorte, L., Verliefde, A.R.D., Verberk, J.Q.J.C., Heijman, S.G.J.,
Amy, G.L., van Dijk, J.C., 2010. Modeling equilibrium adsorption of organic
micropollutants onto activated carbon. Water Res. 44 (10), 3077e3086.
Dubinin, M.M., 1960. The potential theory of adsorption of gases and vapors for
adsorbents with energetically nonuniform surfaces. Chem. Rev. 60 (2),
235e241.
Dubinin, M.M., 1975. Physical adsorption of gases and vapors in micropores. Prog.
Surf. Membr. Sci. 9, 1e70.
Dubinin, M.M., 1985. Generalization of the theory of volume lling of micropores to
nonhomogeneous microporous structures. Carbon 23 (4), 373e380.
Dubinin, M.M., 1989. Fundamentals of the theory of adsorption in micropores of
carbon adsorbents: characteristics of their adsorption properties and microporous structures. Carbon 27 (3), 457e467.
Dubinin, M.M., Astakhov, V.A., 1971. Development of the concepts of volume lling
of micropores in the adsorption of gases and vapors by microporous adsorbents. Bull. Acad. Sci. USSR Div. Chem. Sci. 20 (1), 3e7.

Dubinin, M.M., Sawerina, E., 1936. Charakter der porositats und sorptionseigenschaften aktiver kohle. Acta Physicochem. URSS 4, 647e674.
Dubinin, M.M., Tomofeyev, P., 1946. Adsorption of vapors on active carbons in
relation to the properties of the adsorbate. Dokl. Akad. Nauk. SSSR 54, 701e704.
El-Dib, M.A., Badawy, M.I., 1979. Adsorption of soluble aromatic hydrocarbons on
granular activated carbon. Water Res. 13 (3), 255e258.
Erto, A., Lancia, A., Musmarra, D., 2011. A modelling analysis of PCE/TCE mixture
adsorption based on ideal adsorbed solution theory. Sep. Purif. Technol. 80 (1),
140e147.
Hand, D.W., Loper, S., Ari, M., Crittenden, J.C., 1985. Prediction of multicomponent
adsorption equilibria using ideal adsorbed solution theory. Environ. Sci. Technol. 19 (11), 1037e1043.
Hindarso, H., Ismadji, S., Wicaksana, F., Mudjijati, Indraswati, N., 2001. Adsorption of
benzene and toluene from aqueous solution onto granular activated carbon.
J. Chem. Eng. Data 46 (4), 788e791.
Hsieh, C.T., Teng, H., 2000. Liquid-phase adsorption of phenol onto activated carbons prepared with different activation levels. J. Colloid Interface Sci. 230 (1),
171e175.
Huang, L., Yang, Z., Li, B., Hu, J., Zhang, W., Ying, W.C., 2011. Granular activated
carbon adsorption process for removing trichloroethylene from groundwater.
AIChE J. 57 (2), 542e550.
Hung, H.W., Lin, T.F., 2007. Prediction of the adsorption capacity for volatile organic
compounds onto activated carbons by the Dubinin - Radushkevich - Langmuir
Model. J. Air Waste Manag. Assoc. 57 (4), 497e506.
IPCS INCHEM, 2013. International Programme on Chemical Safety (IPCS) downloaded on 22 August, 2013. http://www.inchem.org.
Kamlet, M.J., Taft, R.W., Abboud, J.L., 1977. Regarding a generalized scale of solvent
polarities. J. Am. Chem. Soc. 99 (25), 8325e8327.
Kapoor, A., Ritter, J.A., Yang, R.T., 1989. On the Dubinin-Radushkevich equation for
adsorption in microporous solids in the Henry's law region. Langmuir 5 (4),
1118e1121.
Kier, L.B., Hall, L.H., 1976. Molecular Connectivity in Chemistry and Drug Research.
Academic, New York.
Kilduff, J.E., Srivastava, R., Karanl, T., 2002. Preloading of GAC by natural organic
matter: effect of surface chemistry on TCE uptake. Stud. Surf. Sci. Catal. 144,
553e560.
Kuennen, R.W., Dyke, K.V., Crittenden, J.C., Hand, D.W., 1989. Predicting the
multicomponent removal of surrogate compounds by a xed-bed adsorber.
J. AWWA 81 (12), 46e58.
Lee, M.C., Vernon, L.S., Crittenden, J.C., 1981. Activated carbon adsorption of humic
substances. J. AWWA 73 (8), 440e446.
Li, L., Quinlivan, P.A., Knappe, D.R.U., 2005. Predicting adsorption isotherms for
aqueous organic micropollutants from activated carbon and pollutant properties. Environ. Sci. Technol. 39 (9), 3393e3400.
Luehrs, D.C., Hickey, J.P., Nilsen, P.E., Godbole, K.A., Rogers, T.N., 1995. Linear solvation energy relationship of the limiting partition coefcient of organic solutes
between water and activated carbon. Environ. Sci. Technol. 30 (1), 143e152.
Manes, M., 1998. Activated carbon adsorption fundamentals. In: Meyers, R.A. (Ed.),
Encyclopedia of Environmental Analysis and Remediation. Wiley, New York,
pp. 26e68.
Manes, M., Hofer, L.J.E., 1969. Application of the polanyi adsorption potential theory
to adsorption from solution on activated carbon. J. Phys. Chem. 73 (3), 584e590.
Manes, M., Wohleber, D.A., 1971. Application of the polanyi adsorption potential
theory to adsorption from solution on activated carbon. II. Adsorption of
partially miscible organic liquids from water solution. J. Phys. Chem. 75 (1),
61e64.
Mezzari, I.A., 2006. Predicting the Adsorption Capacity of Activated Carbon for
Organic Contaminants from Fundamental Adsorbent and Adsorbate Properties
(Master's thesis). North Carolina State University, Raleigh, NC, USA.
Noll, K.E., Wang, D., Shen, T., 1989. Comparison of three methods to predict
adsorption isotherms for organic vapors from similar polarity and nonsimilar
polarity reference vapors. Carbon 27 (2), 239e245.
Ozawa, S., Kusumi, S., Ogino, Y., 1976. Physical adsorption of gases at high pressure.
IV. An improvement of the Dubinin-Astakhov adsorption equation. J. Colloid
Interface Sci. 56 (1), 83e91.
Parker, G.J., 1995. Optimum isotherm equation and thermodynamic interpretation
for aqueous 1,1,2-trichloroethene adsorption isotherms on three adsorbents.
Adsorption 1 (2), 113e132.
Polanyi, M., 1932. Section III. Theories of the adsorption of gases. A general survey
and some additional remarks. Introductory paper to section III. Trans. Faraday
Soc. 28 (0), 316e333.
Polanyi, M.Z., 1920. Adsorption from solutions of substances of limited solubility.
Physik 2, 111e116.
Rahman, M.F., 2014. Removal of Peruorinated Compounds from Ultrapure and
Surface Waters by Adsorption and Ion Exchange (PhD's thesis). University of
Waterloo, Ontario, Canada.
Rozwadowski, M., Wojsz, R., Wisniewski, K.E., Kornatowski, J., 1989. Description of
adsorption equilibrium on type A zeolites with use of the Polanyi-Dubinin
potential theory. Zeolites 9 (6), 503e508.
Shih, Y.H., Gschwend, P.M., 2009. Evaluating activated carbon-water sorption coefcients of organic compounds using a linear solvation energy relationship
approach and sorbate chemical activities. Environ. Sci. Technol. 43 (3), 851e857.
Speth, T.F., Miltner, R.J., 1990. Technical note: adsorption capacity of GAC for synthetic organics. J. AWWA 82 (2), 72e75.
Speth, T.F., Miltner, R.J., 1998. Adsorption capacity of GAC for synthetic organics.

W. Bunmahotama et al. / Water Research 85 (2015) 521e531

J. AWWA 90 (4), 171e174.
pez-Ramo
 n, M., Moreno-Castilla, C., 2001. Adsorption of
Stoeckli, F., Victoria Lo
phenolic compounds from aqueous solutions, by activated carbons, described
by the Dubinin-Astakhov equation. Langmuir 17 (11), 3301e3306.
Urano, K., Koichi, Y., Yamamoto, E., 1982a. Equilibria for adsorption of organic
compounds on activated carbons in aqueous solutions: II. generalization and a
prediction method of adsorption isotherms. J. Colloid Interface Sci. 86 (1),
43e50.
Urano, K., Omori, S., Yamamoto, E., 1982b. Prediction method for adsorption capacities of commercial activated carbons in removal of organic vapors. Environ.
Sci. Technol. 16 (1), 10e14.
Urano, K., Yamamoto, E., Tonegawa, M., Fujie, K., 1991. Adsorption of chlorinated
organic compounds on activated carbon from water. Water Res. 25 (12),
1459e1464.
USEPA, 2002. Chemical Testing and Information, 2002.
USEPA, November 12, 2014. QAPP for Assessment of the Fate of Contaminants in

531

Hydraulic Fracturing Wastewater Treatment Process and Characterization of

Wastewater Residuals.
USEPA, March 4, 2015. Plant Assisted Remediation of Soil and Groundwater
Contaminated by Hazardous Organic Substances: Experimental and Modeling
Studies.
Vaskovsky, B.A., Cited in Dubinin, M.M., Zaverina, E.D., 1950. Adsorption of gases by
activated carbons. Dokl. Akad. Nauk. SSSR 72, 319.
Wohleber, D.A., Manes, M., 1971. Application of the polanyi adsorption potential
theory to adsorption from solution on activated carbon. III. Adsorption of
miscible organic liquids from water solution. J. Phys. Chem. 75 (24), 3720e3723.
Yap, C.W., 2011. PaDEL-descriptor: an open source software to calculate molecular
descriptors and ngerprints. J. Comput. Chem. 32 (7), 1466e1474.
Zhang, S., Liu, X., Karanl, T., 2013. Applicability of the linear solvation energy
relationships in the prediction for adsorption of aromatic compounds on
activated carbons from aqueous solutions. Sep. Purif. Technol. 117 (0),
111e117.