Download as pdf or txt
Download as pdf or txt
You are on page 1of 29

Water Research 120 (2017) 88e116

Contents lists available at ScienceDirect

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

Review

Mistakes and inconsistencies regarding adsorption of contaminants


from aqueous solutions: A critical review
Hai Nguyen Tran a, b, *, Sheng-Jie You b, Ahmad Hosseini-Bandegharaei c, d,
Huan-Ping Chao b, *
a
Department of Civil Engineering, Chung Yuan Christian University, Chungli 320, Taiwan
b
Department of Environmental Engineering, Chung Yuan Christian University, Chungli 320, Taiwan
c
Wastewater Division, Faculty of Health, Sabzevar University of Medical Sciences, PO Box 319, Sabzevar, Iran
d
Department of Engineering, Kashmar Branch, Islamic Azad University, PO Box 161, Kashmar, Iran

a r t i c l e i n f o a b s t r a c t

Article history: In recent years, adsorption science and technology for water and wastewater treatment has attracted
Received 29 October 2016 substantial attention from the scientific community. However, the number of publications containing
Received in revised form inconsistent concepts is increasing. Many publications either reiterate previously discussed mistakes or
29 March 2017
create new mistakes. The inconsistencies are reflected by the increasing publication of certain types of
Accepted 6 April 2017
Available online 18 April 2017
article in this field, including “short communications”, “discussions”, “critical reviews”, “comments”, “letters
to the editor”, and “correspondence (comment/rebuttal)”. This article aims to discuss (1) the inaccurate use
of technical terms, (2) the problem associated with quantities for measuring adsorption performance, (3)
Keywords:
Adsorption
the important roles of the adsorbate and adsorbent pKa, (4) mistakes related to the study of adsorption
Mistake kinetics, isotherms, and thermodynamics, (5) several problems related to adsorption mechanisms, (6)
Comment inconsistent data points in experimental data and model fitting, (7) mistakes in measuring the specific
Inconsistency surface area of an adsorbent, and (8) other mistakes found in the literature. Furthermore, correct ex-
Critical review pressions and original citations of the relevant models (i.e., adsorption kinetics and isotherms) are
Water treatment provided. The authors hope that this work will be helpful for readers, researchers, reviewers, and editors
who are interested in the field of adsorption studies.
© 2017 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
2. Technical terms used in the study of adsorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
3. Quantities for measuring adsorption performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
4. Incorrect assumptions regarding pKa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
5. Adsorption kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
5.1. The important role of initial contact time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
5.2. Adsorption equilibrium time for porous materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.3. Pseudo-first-order (PFO) equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.3.1. Derivation of the PFO equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.3.2. Problems in the application of the PFO equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.4. Pseudo-second-order (PSO) equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.4.1. Derivation of the PSO equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.4.2. Problems in the application of the PSO equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
5.5. Elovich equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96

* Corresponding authors. Department of Environmental Engineering, Chung


Yuan Christian University, Chungli 320, Taiwan.
E-mail addresses: trannguyenhai2512@gmail.com (H.N. Tran), hpchao@cycu.
edu.tw (H.-P. Chao).

http://dx.doi.org/10.1016/j.watres.2017.04.014
0043-1354/© 2017 Elsevier Ltd. All rights reserved.
H.N. Tran et al. / Water Research 120 (2017) 88e116 89

5.5.1. Derivation of the Elovich equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96


5.5.2. Incorrect expression of the Elovich equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
5.6. Intra-particle diffusion equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
6. Adsorption isotherms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
6.1. Adsorption equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
6.2. Langmuir equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
6.2.1. Derivation of the Langmuir equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
6.2.2. Derivation of the separation factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
6.2.3. Mistakes concerning the Langmuir model and the separation factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
6.3. Freundlich equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
6.3.1. Derivation of the Freundlich equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
6.3.2. Mistakes concerning the Freundlich equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
6.4. RedlichePeterson equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
6.4.1. Derivation of the RedlichePeterson equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
6.4.2. Mistakes concerning the RedlichePeterson equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
6.5. DubinineRadushkevich equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
6.5.1. Derivation of the DubinineRadushkevich equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
6.5.2. Mistakes concerning the DubinineRadushkevich equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
7. Adsorption thermodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
7.1. Principles of adsorption thermodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
7.2. Equilibrium constant derived from the Langmuir constant (KL) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
7.3. Equilibrium constant derived from the Freundlich constant (KF) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
7.4. Equilibrium constant derived from the partition coefficient (Kp) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
7.5. Equilibrium constant derived from the distribution coefficient (Kd) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
8. Others mistakes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
8.1. Inconsistent data points in experimental data and model fitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
8.2. Oxidation state of chromium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
8.3. Incorrect labels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
8.4. BET specific surface area of an adsorbent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
8.5. Maximum absorption wavelength in dye adsorption studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
8.6. Cp-cation and p-p interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
8.6.1. Cp-cation interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
8.6.2. p-p interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
8.7. Other miscellaneous errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
9. Nonlinear-optimization technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
10. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

1. Introduction related to the adsorption of contaminants in aqueous solutions


reveals that some less obvious inconsistencies and mistakes have
Nowadays, the number of publications in international scientific unavoidably slipped the attention of authors and were repeated in
journals has increasingly become a standard for the assessment of subsequent publications.
scientists (Geckeis and Rabung, 2004). As the number of scientific The observation of these inconsistencies and mistakes, regard-
publications increase, mistakes and misconceptions enter scientific less of their origin, along with comments and open discussions can
literature and some can also be repeated in subsequent publica- prevent their propagation in scientific knowledge transfer. There-
tions. Once errors and mistakes enter the literature, it is difficult to fore, the presentation of a comprehensive review on common
eradicate them. The entrance of mistakes in scientific publications, mistakes in adsorption studies, besides introducing the most cor-
as well as their propagation, can result from the subjectivity of the rect approaches to use and giving a source of up-to-date literature
authors and/or objective reasons. on this issue, seems to be valuable to both readers and researchers
A typical study on the adsorption of organic and inorganic in this field.
contaminants comprises several sections, such as the effects of The main purpose of this review is to identify mistakes in past
solution pH and ionic strength, studies of adsorption kinetics, iso- publications and prevent the propagation of such mistakes in
therms, thermodynamics, desorption, and regeneration. Adsorp- future scientific literature. Herein, we present some common
tion processes can be conducted using either batch or column mistakes in various aspects of adsorption studies and analyze the
techniques. It has been found that many scientific articles in the reasons for their entrance into the literature. This was accom-
field of adsorption science and technology contain several mistakes plished by discussing the published papers in the field of adsorp-
and misconceptions including the use of technical terms, applica- tion of contaminants from aqueous solutions, commonly including
tion of quantities, determining the roles of some constants like pKa, “comments on”, “reply for comments”, “response to”, “authors'
performing kinetic studies and modeling kinetic adsorption data, response to comments on”, “comments on the authors” response to
finding a suitable isotherm model to fit equilibrium data, calcu- comments on”, “comments on comment on”, “in reaction to”, “critical
lating thermodynamic parameters, correct interpretation of review”, and “a note on the comments” published works. In addition
adsorption mechanisms, and measuring the specific surface area of to our comprehensive review on the mistakes existing in publica-
an adsorbent. Moreover, a survey of the scientific publications tions, some correct approaches are suggested where necessary to
90 H.N. Tran et al. / Water Research 120 (2017) 88e116

avoid the propagation of such mistakes in the future scientific chelation), physisorption, or microprecipitation” (Robalds et al.,
literature. This paper primarily focuses on three basic constituents 2016). In addition, the use of the term “electrostatic adsorption”
of adsorption theory: adsorption equilibria, kinetics, and may cause misunderstanding and confusion; therefore, depending
thermodynamics. on the intended meaning, the appropriate terms (i.e., “electrostatic
attraction”, “electrostatic force”, or “electrostatic interaction”)
are recommended. Notably, the authors also suggested a new
2. Technical terms used in the study of adsorption classification system for the (bio)sorption mechanisms of heavy
metals (Fig. 3).
Adsorption processes have their own vernacular. Therefore, The electrical state of an adsorbent's surface in solution is usu-
correctly understandings the technical terms used in adsorption ally characterized by either the point of zero charge (PZC) or the
technology can prevent the introduction of several unexpected isoelectric point (IEP). Many authors have complained about the
ambiguities and discrepancies. Some basic adsorption terms are misuse of the terms IEP and PZC outside their normal meanings
summarized in Fig. 1. (Somasundaran, 1968; Kosmulski, 2009). The PZC is defined as the
To some extent, a thorough understanding of the properties of solution conditions under which the surface charge density equals
adsorbents might prevent common mistakes made in interpreting zero. Potentiometric titration, or related methods, is used to
adsorption processes and adsorption mechanisms. Numerous determine the PZC by finding the point at which the apparent
techniques can be used to characterize an adsorbent (Fig. 2). More surface charge density in the presence of an inert electrolyte is not
information on the basic properties of adsorbents (i.e., biosorbent, dependent on ionic strength (Fig. 4a). The IEP occurs when the
biochar, hydrochar, and activated carbons) and the relations be- electrokinetic (z) potential at the shear plane equals zero; z is
tween these properties has been published in our recent work determined from measuring electrokinetic (electrophoresis and
(Tran et al., 2017c). streaming potential) and electroacoustic phenomena (Fig. 4b)
The appropriate use of technical words regarding the bio- (Kosmulski, 2009). IEP values clearly represent only the external
sorption mechanism of heavy metals has been highlighted by surface charge of particles in solution, whereas the PZC varies in
Robalds et al. (2016). To avoid reader confusion, they recommended response to the net total (external and internal) surface charge of
that “surface precipitation” be used instead of “microprecipitation” the adsorbent (as the characteristic of amphoteric surfaces).
(also written as “micro-precipitation”). Similarly, it is necessary to Therefore, the difference between the PZC and IEP values for a
distinguish between “chelation”, “complexation”, and “coordina- porous carbon system can be interpreted as a measure of the sur-
tion”. The word “complexation” includes two meanings: “chelation” face charge distribution. A difference (PZC  IEP) of greater than
and/or “coordination”. For example, the incorrect sentence “in zero indicates that the external particle surface is more negatively
another sense, it can also be defined as a collective term for a number charged than the internal particle surface, and a difference of close
of passive accumulation processes which in any particular case may to zero corresponds to a more homogeneous distribution of surface
include ion exchange, coordination, complexation, chelation, adsorp- charge. The PZC is only equal to the IEP in the absence of specific or
tion and microprecipitation” should be revised as “biosorption may counter ions. This means that equality between the PZC and IEP
include ion exchange, complexation (including coordination and/or

Fig. 1. Some basic terms used in adsorption science and technology.


H.N. Tran et al. / Water Research 120 (2017) 88e116 91

Fig. 2. Basic properties of an adsorbent determined by various common techniques.

the percentage of removed adsorbate. The amount of adsorbate


adsorbed at equilibrium (qe; mg/g) is often calculated using the
material balance of an adsorption system; the adsorbate, which has
disappeared from the solution, must be in the adsorbent (Volesky,
2007).

ðCo  Ce Þ
qe ¼ V (1)
m

where C0 (mg/L) and Ce (mg/L) are the initial and equilibrium


adsorbate concentration in solution, respectively; m (g) is the dried
mass of used adsorbent; V (L) is the volume of the adsorbate so-
lution; and m/V (g/L) is defined as a solid/liquid ratio.
Depending on the purpose of the study, the parameter qe may be
expressed in different units as follows: (1) mg/g for evaluating
practical and engineering processes, (2) mmol/g or meq/g for
examining the stoichiometry of a process and for studying func-
tional groups and metal-binding mechanisms, and (3) mmol/g for
comparing the selective adsorption performance of various adsor-
Fig. 3. Schematic illustration of a new classification system for metal (bio)sorption
bates (Volesky, 2007). Of these units, mg/g (milligrams of adsorbate
mechanisms (Adapted from Robalds et al., 2016. Copyright (2016), with permission
from Elsevier). adsorbed per gram of dried adsorbent) is used most commonly for
qe in adsorption studies.
In addition to qe, the adsorption performance can also be
exists only if no specific adsorption of counter ions from the solu- expressed as the percentage of removed adsorbate (%removal) from
tion occurs (Radovic, 1999; Onjia and Milonjic, 2002). solution. However, the unit of %removal needs to be used cautiously,
as it is very approximate and can cause misleading conclusions
3. Quantities for measuring adsorption performance about relative adsorption performance. This unit is only appro-
priate for the purpose of crude measurements and perhaps for
In the field of adsorption, adsorption performance can be quick and very approximate screening of adsorbent materials
expressed as the amount of adsorbate adsorbed at equilibrium or (Kratochvil and Volesky, 1998; Hai, 2017).
92 H.N. Tran et al. / Water Research 120 (2017) 88e116

ReOHþ
2 4 ReOH þ H
þ
(3)

ReOH 4 ReO þ Hþ (4)

ReO þ Cu2þ 4 ReOCuþ (5)

ReO þ Cu(OH)þ 4 ReOCu(OH) (6)

where eR represents the surface adsorbent; ReOHþ 2 , ReOH, and


ReO represent protonated, neutral, and ionized surface hydroxyl
functional groups, respectively; and ReOCuþ and ReOCu(OH) are
the formed bonding complexes.
The authors (Ben Ali et al., 2016) explained that, “the effect of pH
could be also explained by considering the point of zero charge of the
adsorbent (pHpzc). The adsorbent surface is positively charged for
pH < 4.7 and it becomes negatively charged at pH value above 4.7.
Therefore, for pH values < 4.7, the adsorption is unfavorable because of
repulsive electrostatic interactions between metal ions and positively
charged functional groups. The maximum adsorption of Cu(II) occurs
at pH above pHpzc value when the adsorbent surface is negatively
highly charged.” However, the important role of the pKa values of
carboxylic acid groups (eCOOH; pKa 1.7e4.7) and hydroxyl groups
(eOH; pKa 9.5e13) was ignored (Volesky, 2007). Correspondingly,
Eqs. (3)e(6) indicate a misunderstanding of the surface chemistry
of the adsorbent. The authors investigated the effect of solution pH
on the adsorption process of Cu(II) in solutions with pH values from
2.0 to 6.0. Therefore, dissociation of the eOH groups to eO did not
occur at these solutions pH values. Instead, the eCOOH groups
dissociated, forming negatively charged carboxylate groups
(eCOO‒) in the aforementioned pH range, and were the dominant
groups that interacted with copper ions (Tran, 2017).
Moreover, Bui and Choi (2010) made a critical comment that
Oleszczuk et al. (2010) mistyped the properties of two pharma-
ceuticals: oxytetracycline (logKow ¼ 2.45; pKa ¼ 7) and carbamaz-
epine (logKow ¼ 1.22; pKa ¼ 3.27, 7.32, and 9.11). The correct
expression should be oxytetracycline (Sw ¼ 121 mg/L ;
Fig. 4. A typical example of difference between (a) point of zero charge (PZC) and (b) logKow ¼ 1.22; pKa ¼ 3.27, 7.32, and 9.11; density ¼ 1.63 g/cm3)
isoelectric point (IEP) glucose-derived spherical biochar prepared at 800  C (Data and carbamazepine (Sw ¼ 112 mg/L; logKow ¼ 2.45; pKa ¼ 7;
unpublished).
density ¼ 1.15 g/cm3). Bui and Choi (2010) also emphasized that the
pKa value of carbamazepine reported by Oleszczuk et al. (2010) is
still debated. Carbamazepine (CBZ) is commonly characterized with
ðCo  Ce Þ two pKa values: pKa1 (¼ 1 or 2.3) for the equilibrium RCONHþ 3 #
%removal ¼  100 (2)
Co RCONH2 þ Hþ and pKa2 (¼ 13.9) for the equilibrium RCONH2 #
RCONH þ Hþ. The misreporting of the CBZ pKa can cause inac-
To some extent, the use of %removal in the study of adsorption
curate conclusions. For example, Oleszczuk et al. (2010) suggested
equilibria can cause incorrect observations and, as a result, inac-
that an electrostatic interaction occurred between CBZ and multi-
curate conclusions (a detailed discussion is provided in Section 6.1).
walled carbon nanotubes (MWCNT) at pH < 7.0 because they
opined that CBZ existed in its cationic form at pH < 7.0 (pKa ¼ 7.0).
4. Incorrect assumptions regarding pKa However, CBZ with pKa1 < 2.3 and pKa2  13.9 actually exists
entirely as a neutral species in the pH range of 1.9e11.9, as calcu-
A mistake regarding the fundamentals of chemistry was com- lated using ACDLab. As a result, the electrostatic interaction was
mented on by Rayne (2013). The commented paper reported the ruled out as a contributing interaction in the CBZ-adsorption
adsorption of n-perfluorooctanesulfonamide (PFOSA) onto multi- mechanisms (Bui and Choi, 2010).
walled carbon nanotubes with differing oxygen content. However,
PFOSA was mistakenly assumed to be an “organic base”, resulting in 5. Adsorption kinetics
the following invalid interpretation: “for PFOSA (pKa ¼ 6.52), when
pH < pKa, protonation occurs on the amino group, and the decreased 5.1. The important role of initial contact time
protonation leads to the increased adsorption, but when pH > pKa,
PFOSA exists as neutral molecule in water.” In fact, PFOSA is an acid In the study of adsorption kinetics, Azizian proposed that by the
and, thus, is a neutral compound below its pKa value (estimated at observation of breaking point or deviating from linearity at initial
between 6.0 and 7.0) and a dissociated anion at pH > pKa (Rayne, regions of t/qt versus t plots, means presence of at least two
2013). different sites for adsorption Azizian (2006) or contribution of
Recently, the effect of pH on the interaction of copper with the diffusion on the rate of adsorption (Haerifar and Azizian, 2013) and
surface functional groups of raw pomegranate-peel biosorbent was therefore fitting with the PSO model should be avoided. Adsorption
reported as follows: kinetics data in the initial contact time periods play an important
H.N. Tran et al. / Water Research 120 (2017) 88e116 93

role in performing accurate modeling and drawing valid involving desorption from the adsorbent surface after reaching
conclusions. adsorption equilibrium (Fig. 5d) has been reported in the literature
Depending on the affinity between the adsorbent and adsor- (Sharma et al., 2013; Halder et al., 2016). Therefore, studies of the
bate, the initial adsorption rate may be very fast. For the adsorption adsorption kinetics play an important role in identifying the
of heavy metals, Tran et al. (2015) investigated the adsorption of required equilibration time and the optimal contact time and for an
Cd2þ onto orange-peel-generated biochars, and found that equi- adsorption process.
librium can be reached rapidly in kinetic experiments with a In aqueous-phase adsorption, two kinetic reaction models (i.e.,
removal rate of 80.6e96.9% (within 1 min). An analogous perfor- pseudo-first-order and pseudo-second-order equations), the Elo-
mance was reported by Guo et al. (2015), with approximately vich or RoginskyeZeldovich model, and the intraparticle diffusion
70.0e96.6% of the total Cd2þ in solution removed within the first model have been widely applied to mathematically describe the
minute of contact. Additionally, Müller et al. (2012) noted that intrinsic kinetic adsorption constant. Furthermore, most mistakes
approximately 90% of Zn(II) adsorption onto Lewatit S1468 occurs found and discussed in the literature correspond to the use of ki-
within 1 min, and this adsorption process reached equilibrium in netic models. Therefore, the following Sections (5.3e5.6) include a
approximately 5 min. thorough discussion of these models.
For the adsorption of organic contaminants, kinetic studies
indicated that the adsorption equilibrium of cationic methylene 5.3. Pseudo-first-order (PFO) equation
green 5 was quickly established, with a low activation energy
required for adsorption (Ea; 3.30e27.8 kJ/mol), and activated car- 5.3.1. Derivation of the PFO equation
bons removed 50e73% of the dye from solution within 1 min (Tran Recently, Ho published a series of critical papers (Ho and McKay,
et al., 2017d, 2017b). Similarly, Canzano et al. (2012) noted that the 1998b; Ho, 2005, 2014a, 2015, 2016e) discussing two main points.
adsorption process of anionic Congo-red dye onto raw and acid- First, Ho (Ho and McKay, 1998b) claimed to offer the correct
treated pine cone powder reached equilibrium in the first few reference style for citing Lagergren's paper, with the original
minutes (<5 min), during which no measurements were carried translated from “Lagergren, S. (1898), Zur theorie der sogenannten
out. adsorption gelo€ster stoffe. Kungliga Svenska Vetenskapsakademiens
As time goes on, the amount of adsorbate uptake per mass of Handlingar, Band 24, No. 4, 1e39” into English as “Lagergren, S.,
adsorbent at any time t (qt) approaches that of at equilibrium (qe) About the theory of so-called adsorption of soluble substances, Kun-
and, therefore, plotting the kinetic data is bound to produce a gliga Svenska Vetenskapsakademiens. Handlingar, Band 24, No. 4,
straight line, independent of the real kinetic order followed by the 1898, pp. 1e39.” Second, Ho stated that Lagergren's first-order rate
system. In other words, equilibrium data does not describe the way equation has been called pseudo-first-order since 1998 (Ho, 2005)
by which the system reaches equilibrium. As a result, the use of in order to distinguish kinetic equations based on the adsorption
data recorded over a significant time interval after the attainment capacity of solids from those based on the concentration of a so-
of equilibrium, or very close to equilibrium, is likely to lead to lution. However, the responses of many authors (Ozacar, € 2005;
erroneous conclusions regarding adsorption kinetics. Kumar, 2006d; Utomo et al., 2010) state that Lagergren's first-
order rate expression had been referred to as a pseudo-first-order
5.2. Adsorption equilibrium time for porous materials equation before Ho's publication in 1998. For example, in 1990,
Sharma et al. (1990) studied the removal of Ni(II) from an aquatic
The adsorption processes for inorganic and organic contami- environment by wollastonite. In the section on adsorption kinetics,
nants onto porous adsorbents (i.e., activated carbon, biochar, acti- they stated: “the adsorption of Ni (II) on wollastonite follows the first
vated carbon spheres, zeolite, and macroreticular resin) can take order adsorption rate expression of Lagergren (1898).”
longer (i.e., several days or weeks) to approach true equilibrium The great contributions of Ho toward correcting the citation
than nonporous adsorbents (i.e., biosorbent and hydrochar). This style format for the PFO equation cannot be denied. However, the
difference is attributed to the different adsorption mechanisms of PFO equation was not originally proposed or initially expressed by
these adsorbents. Pore filling is the most common adsorption Ho. In the scientific community, the authors who first propose a
mechanism for porous adsorbents, along with several other in- theoretical model should always be given credit for their contri-
teractions (i.e., electrostatic attraction, hydrogen bonding, surface butions via correct citation (Vasanth Kumar and Rattanaphani,
precipitation, p-p interaction, n-p interaction, and cation ex- 2007; Wu, 2009; Utomo et al., 2010). Therefore, the original work
change) that depend on the characteristics of the porous adsor- by Lagergren (1898), who first presented the first-order-rate
bents and adsorbates. Many studies have opted for a 24 h equation for the adsorption of oxalic acid and malonic acid onto
equilibrating time for isotherm adsorption experiments, without charcoal, should be cited. The PFO equation can be expressed
providing evidence that adsorption in their system attained true correctly in nonlinear (Eq. (7)) and linear forms (Eq. (8) or Eq. (9)).
equilibrium within the appropriate timeframe.  
Therefore, it is necessary to provide the specified timeframe in qt ¼ qe 1  ek1 t (7)
which the adsorption process approaches a true equilibrium.
Fig. 5aec represent the influence of contact time on the adsorption
processes of various contaminants onto porous adsorbents (Hung k
logðqe  qt Þ ¼  1 t þ logðqe Þ (8)
and Lin, 2006; Masson et al., 2016; Kim et al., 2016). Clearly, the 2:303
amount of contaminants adsorbed onto the porous adsorbents
continued to increase after 24 h. For example, Kim et al. (2016) lnðqe  qt Þ ¼ k1 t þ lnðqe Þ (9)
speculated on the process of methyl violet (MV) adsorption onto
synthesized granular mesoporous carbon (GMC) and commercial where qe and qt are the amounts of adsorbate uptake per mass of
granular activated carbon (GAC). The results of their kinetic study adsorbent at equilibrium and at any time t (min), respectively; and
demonstrated that MV adsorption reached an equilibrium state k1 (1/min) is the rate constant of the PFO equation.
within 20 d for GMC-polyvinyl alcohol and 26 d for GMC-
carboxymethyl cellulose, while adsorption onto GAC had not 5.3.2. Problems in the application of the PFO equation
reached equilibrium even after 75 d. Furthermore, a phenomenon It cannot be denied that the PFO equation is widely applied in
94 H.N. Tran et al. / Water Research 120 (2017) 88e116

Fig. 5. Effect of contact time on the adsorption process of (a) methyl violet onto granulated mesoporous carbon (GMC) and granular activated carbon (GAC) (Reprinted from Kim
et al. 2016. Copyright (2016), with permission from Elsevier); (b) methyl tert-butyl ether onto carbonaceous resin (Ambersorb 563) (Reprinted from Hung and Lin 2006. Copyright
(2006), with permission from Elsevier); (c) ibuprofen (IBP), carbamazepine (CBZ), ofloxacin (OFX), bisphenol-A (BPA), diclofenac (DFN), mecoprop (MCP), pentachlorophenol (PCP),
benzotriazol (BZT), and caffeine (CAF) onto activated carbon cloths (resin phenolic precursor) (Reprinted from Masson et al. 2016. Copyright (2016) with permission from Elsevier);
and (d) methyl blue onto graphene oxide (Reprinted with permission from Sharma et al. 2013. Copyright (2013) American Chemical Society).

adsorption kinetics. However, several problems related to the


application of the PFO equation have been discussed in the litera- k1 t
logðqe  qt Þ ¼ logðqe Þ (13)
ture. First, many incorrect expressions for its linear forms have been 2:303
published. Various incorrect forms are shown in Eq. (10)
(Alagumuthu et al., 2010), Eq. (11) (Fu et al., 2007; Jing et al., Ci k t
log ¼ 1 (14)
2014; Nekouei and Nekouei, 2017), Eq. (12) (Ho 2014a, 2014e; Ct 2:303
Jing et al., 2014; Ho 2016a, 2016b, 2016c, 2016d, 2016e, 2017), Eq.
 
(13) (Ho, 2015), Eqs. (14) and (15) (Jing et al., 2014), Eq. (16) (Kumar, C  Ct k t
2006a), and Eq. (17) (Kumar and Porkodi, 2007b), which should be log 1  i ¼ 1 (15)
Ci  Ce 2:303
avoided in future studies. As a result of these incorrect expressions,
recalculation of the PFO equation parameters was necessary to
qe k1
obtain the correct parameters (Wu et al., 2017). log ¼ (16)
qe  qt 2:303t

logðqe Þ  k1 k1
logðqe  qt Þ ¼ t (10) logðqe  qt Þ ¼ logðqe Þ  (17)
2:303 2:303

where Ci (mg/L), Ct (mg/L), and Ce (mg/L) are the concentrations of


logðqe Þ  k1 adsorbate at the initial time (t ¼ 0), any time t, and equilibrium,
logðqe  qt Þ ¼ (11) respectively.
2:303t
The second problem is the two unknown parameters (qe and k1).
Furthermore, in most cases, the PFO equation is only appropriate
1 k 1 for the initial 20e30 min of contact time, not for the whole range
¼ 1 þ (12) (Ho and McKay, 1998a). Thus, plots of Eq. (8) or Eq. (9) are only
qt qe t qe
linear over approximately the first 30 min; beyond this initial
period, the experimental and theoretical data will not fit
H.N. Tran et al. / Water Research 120 (2017) 88e116 95

adequately (McKay et al., 1999). Another important problem is the Favere, 2006; Kumar and Guha, 2006; Vasanth Kumar and
selection of an appropriate qe value. Notably, the adjusted qe value Rattanaphani, 2007) to clarify the correct citation of the PSO
cannot be lower than the maximum measured value of qt. This is equation. Thus, the nonlinear form of the PSO equation for solid-
because mathematical errors associated with taking the logarithm liquid adsorption systems was not originally reported by Ho
of a negative number will occur when using Eq. (8) or Eq. (9) (1995) or first applied by Ho et al. (1996), but initially proposed
(Plazinski et al., 2009). Finally, the qe value calculated using the by Blanchard et al. (1984).
PFO equation is not equal to the qe value obtained from experi- In the literature, the PSO equation can be correctly expressed in
ments, which further indicates the inability of the PFO equation to four linear forms (Eqs. (23)e(26)), and the initial adsorption rate (h)
fit kinetic adsorption data (Febrianto et al., 2009). The presence of a can be described by Eq. (29).
boundary layer or external resistance controlling the beginning of Expression for the linear form (Type 1) with a plot of t/qt versus
the sorption process was argued to be responsible for this t:
discrepancy (McKay et al., 1999).  
Therefore, two methods have been recommended to obtain an t 1 1
¼ tþ (23)
accurate estimation of the kinetic parameters in the PFO equation: qt qe k2 q2e
(1) a trial and error process to obtain the optimal qe value (Ho and
McKay, 1998a), or (2) application of a nonlinear optimization Expression of the linear form (Type 2) with a plot of 1/qt versus
1/t:
technique. A simple guide to using the nonlinear method is intro-
duced in Section 9.  
1 1 1 1
¼ þ (24)
qt k2 q2e t qe
5.4. Pseudo-second-order (PSO) equation
Expression of the linear form (Type 3) with a plot of qt versus qt/
5.4.1. Derivation of the PSO equation t:
In 1984, Blanchard et al. (1984) initially proposed a second-  
order rate equation for the removal of heavy metals from water 1 qt
qt ¼  þ qe (25)
using natural zeolites, as follows: k2 qe t
Expression of the linear form (Type 4) with a plot of qt/t versus
dn
¼ Kðno  nÞ2 (18) qt:
dt
Integration of Eq. (18) gives: qt
¼ ðk2 qe Þqt þ k2 q2e (26)
t
1
 a ¼ Kt (19) The initial adsorption rate (h) was proposed by Ho et al. (1996).
no  n First, Eq. (22) can be rearranged to:
After rearranging, Eq. (19) becomes:
1
qt ¼ (27)
Ktno þ ano  1 1 þ qte
n¼ (20) k2 q2e
Kt þ a
Subsequently, when t approaches 0, Eqs. (23) and (27) can be
where n is the amount of M2þ fixed or the amount of NHþ 4 released expressed as Eq. (28). Setting h ¼ qt/t, the initial adsorption rate (h;
at each instant; no is an exchange capacity; and K is a rate constant. mg/g  min) can then be determined by Eq. (29).
Plotting 1/(n0  n) as a function of time (t) must produce a
straight line, the slope of which gives the rate constant K and the t 1 qt
intercept leads to the exchange capacity. Considering the boundary ¼ 4 ¼ k2 q2e (28)
qt k2 q2e t
condition of n ¼ 0 for t ¼ 0, it follows that a ¼ 1/n, and Eq. (20) can
be rearranged to:
h ¼ k2 q2e (29)
n2
þ Kt
no ¼ (21) Clearly, only the linear form (Type 1, Eq. (23)) might have been
1 þ Knt reported by Ho et al. (1996) for the adsorption of dyes from waste
If qt ¼ no, qe ¼ n, and k2 ¼ K, Eq. (21) becomes: €
streams by peat (Ozacar, 2005; Kumar, 2006d; Kumar and F avere,
2006; Wu, 2009). Kumar and Fa vere (2006) hypothesized that, in
q2e k2 t adsorption research, the Langmuir equation (Langmuir, 1918) has
qt ¼ (22)
1 þ k2 qe t been the most widely used isotherm to describe adsorption pro-
cesses under equilibrium conditions. In the literature, four linear
where qe (mg/g) and qt (mg/g) are the amount of adsorbate types of the Langmuir isotherm have been reported. Irrespective of
adsorbed at equilibrium and at any t (min), respectively; and k2 (g/ the linear expression reported, these equations have been widely
mg  min) is the rate constant of the PSO equation. called the Langmuir isotherm. It is not correct to claim that the
The nonlinear form of the PSO equation (Eq. (22)) has been Langmuir model was derived or proposed by other authors who
applied elsewhere (Ho, 1995; Ho et al., 1996). Recently, a series of transformed the original Langmuir expression into a new linear
comments made by Ho implied that the derivation and first expression (Kumar and Guha, 2006). In scientific literature, the
application of the PSO equation belongs to his work. Any applica- names of scientists (i.e., Langmuir or Freundlich) have always been
tion of the PSO equation in the field of adsorption studies without presented, even after modification of their equations (Arica, 2003).
citation of the original work or with a misquotation (i.e., the cita- Similarly, Utomo et al. (2010) suggested that it might not be
tion of a secondary reference) might suffer from his comments. necessary to cite papers that deal with mathematical modification
However, in reaction to Ho's comments, a series of notes were and different mathematical expressions of the models (i.e., PFO and
published by other scholars (Kumar, 2006c, 2006d; Kumar and PSO, and Langmuir). However, from the scientific point of view, the
96 H.N. Tran et al. / Water Research 120 (2017) 88e116

authors who first proposed a theoretical model should be credited analysis, Boehm titration, and solution calorimetry), and (2) having
because of their great contributions to the advancement of science. a good sense of the chemical nature of the adsorbate and adsorbent,
Therefore, the original work by Blanchard et al. (1984) should be adsorbent's surface, and chemical or physical interactions between

cited for the expression of the PSO equation (Ozacar, 2005; Kumar, the adsorbent and adsorbate (Volesky, 2007; Lima et al., 2015,
2006d; Kumar and Fa vere, 2006; Azizian, 2007; Plazinski et al., 2016). The use of analytical techniques together with adsorptive
2009; Wu, 2009; Utomo et al., 2010). thermodynamic data (i.e., changes in enthalpy and entropy) and
activation and adsorption energies, are necessary to confirm
whether the adsorption of contaminants in aqueous solution is a
5.4.2. Problems in the application of the PSO equation chemical or physical process (Lima et al., 2015; Tran et al., 2016).
First, many incorrect expressions of the linear form of PSO Finally, in most cases, experimental data for adsorption kinetics
equation are found in the literature, as shown in Eq. (30) (Fu et al., are well fitted by the linear form of the PSO equation. However,
2007; Kumar and Porkodi, 2007b), Eq. (31) (Ho, 2014f), Eqs. (32) caution should be taken in drawing conclusions based on fitting to
and (33) (Ho, 2014g), Eq. (34) (Ho, 2013), Eq. (35) (Lin and Wang, the linear form of the PSO equation. For example, Fig. 6 presents a
2009), Eq. (36) (Ho, 2014c), Eq. (37) (Ho, 2014d), Eq. (38) (Ho, plot of t/qt versus t. Clearly, the R2 values for the linear form of the
2014b), and Eq. (39) (Ho, 2015). PSO equation are very high (R2 > 0.99); however, the corresponding
R2 values for the nonlinear form of the PSO equation are signifi-
t 1 1
¼ þ (30) cantly lower (R2 ¼ 0.53e0.68). This finding means that the
qt k2 q2e qe t
adsorption process of methylene green 5 (MG5) onto synthesized
activated carbon and commercial activated carbon (porous mate-
t 1 1 rials) is not adequately described by the PSO equation. In fact, Tran
¼ þ (31)
qt 2k2 qe et al. (2017d) concluded that the process of MG5 adsorption onto
these porous materials involved p-p interactions and pore filling,
t 1 t and was not related to chemical adsorption. Obviously, the
¼ þ (32)
qt k2 qe qe nonlinear method can be applied to obtain kinetic model param-
eters that are more accurate than those obtained using the linear
  method.
t 1 2 1
¼ q þ t (33)
qt k2 e qe

t 1 1
¼ þ t (34)
qt k2  q2e qe
5.5. Elovich equation
 
t 1 1 5.5.1. Derivation of the Elovich equation
¼  t (35)
qt k2 q2e qe An empirical equation was firstly proposed by Roginsky and
Zeldovich (1934) for the adsorption of carbon monoxide onto
t 1 t manganese dioxide. However, this equation is now generally
¼ k q2 þ (36)
qt 2 2 e qe known as the Elovich equation and has been extensively applied to
chemisorption data (McLintock, 1967). This equation can be
expressed mathematically as follows:
1 1 1
¼ þ (37)
qt k2 q2e qe

t 1 1
¼ þ (38)
qt k2 q2e qe

t k t t
¼ 2 þ (39)
qt k2 q2e qe
Second, based on the best fit of kinetic experimental data using
the PSO model, many authors have drawn controversial conclu-
sions. For example, “the adsorption process is the chemisorption,
which involve valence forces through sharing electrons between
adsorbate and adsorbent”, “the best fit of experimental kinetic data in
pseudo-second-order kinetics suggests the chemisorption, which may
involve valency forces through sharing of electrons between dye anion
and adsorbent”, and other similar conclusions. The problem with
such conclusions has been discussed by several researchers (Kumar,
2006a; Lima et al., 2015; Lima et al., 2016), who claimed that
adsorption mechanisms cannot be directly assigned based on
observing simple kinetic experiments or by fitting kinetic models
(i.e., the PFO and PSO models).
Adsorption mechanisms can only be established by (1) using Fig. 6. Fits of kinetic adsorption data to the linear pseudo-second-order equation [the
study of methylene green 5 adsorption onto commercial activated carbon (CAC) and
several analytical techniques (i.e., FTIR, SEM, nitrogen adsorption- synthesized activated carbons prepared from golden shower through different
desorption isotherms, Raman spectroscopy, TGA/DTA, DSC, 29Si chemical activation methods (GSAC, GSBAC, GSHAC, and GSHBAC)] (Adapetd from Tran
and 13C solid-state NMR, XRD, XPS, pHPZC, pHIEP, CHN element et al., 2017d. Copyright (2017), with permission from Elsevier).
H.N. Tran et al. / Water Research 120 (2017) 88e116 97

Four steps associated with transport processes during adsorp-


dqt
¼ a expð  bqt Þ (40) tion by porous adsorbents were originally proposed in reference
dt (Walter, 1984). The first stage is transport in the solution phase
(known as “bulk transport”; occurs quickly), which can occur
where qe and qt are the amounts of adsorbate uptake per mass of
instantaneously after the adsorbent is transferred into the adsor-
adsorbent at equilibrium and at any time t (min), respectively; a
bate solution; therefore, it does not control engineering design. In
(mg/g  min) is the initial rate constant because dqt/dt / a when qt
most cases, this stage occurs too rapidly and its contribution is
/ 0; and b (mg/g) is the desorption constant during any one
considered negligible. The second stage is “film diffusion” (occurs
experiment.
slowly). In this stage, the adsorbate molecules are transported from
By applying the boundary conditions of qt ¼ 0 at t ¼ 0, the in-
the bulk liquid phase to the adsorbent's external surface through a
tegrated form of Eq. (40) will become Eq. (41) (nonlinear).
hydrodynamic boundary layer or film. The third stage involves
1 diffusion of the adsorbate molecules from the exterior of the
qt ¼ lnð1 þ abtÞ (41) adsorbent into the pores of the adsorbent, along pore-wall surfaces,
b
or both (known as “intraparticle diffusion”; occurs slowly). The last
To simplify the Elovich equation, Chien and Clayton (1980) stage, adsorptive attachment, often occurs very quickly; therefore,
assumed abt [ 1. Thus, a linear form (Eq. (42)) can be obtained: it is also not significant for design. These four steps are summarized
in Fig. 7 (Weber and Smith, 1987).
1 1
qt ¼ lnðt Þ þ lnðabÞ (42)
b b 6. Adsorption isotherms

and a plot of qt versus lnt should give a linear relationship with a 6.1. Adsorption equilibrium
slope of (1/b) and an intercept of (1/b)ln(ab).
Milonjic (2009a) and Milonjic (2010) suggested that the amount
5.5.2. Incorrect expression of the Elovich equation of adsorbate adsorbed onto an adsorbent depended on the equi-
Errors in the presentation of the nonlinear form of the Elovich librium concentrations of metal ions, equilibrium solution pH, and
model are common in the literature. These incorrect equations temperature. An adsorption isotherm should be given as the
(Eqs. (43)e(45)) can be found in several previous papers (Ho and equilibrium adsorbed amounts versus the equilibrium ion con-
McKay, 1998a; Lin and Wang, 2009; Alagumuthu et al., 2010; Ben centrations for a constant equilibrium solution pH and tempera-
Ali et al., 2016). ture. Thus, environmental parameters in the sorption system
(especially solution pH) must be carefully controlled at the given
qt ¼ lnðt Þ þ b lnðabÞ (43) value over the entire contact period until the sorption equilibrium
is reached. These comments were in good agreement with the
1 suggestion that metal-ion adsorption is pH dependent (Volesky,
qt ¼ lnðt Þ þ b lnðabÞ (44)
b 2003, 2007; Tien, 2007; Tien, 2008). However, the pH effect has
invariably been examined in terms of the initial pH of the aqueous
qt ¼ b lnðt Þ þ b lnðabÞ (45) solution. Not a single case has reported data demonstrating the
changes in the solution pH during the course of an adsorption
experiment.
In the scientific literature, adsorption equilibrium studies in
5.6. Intra-particle diffusion equation batch experiments have been conducted in different ways as fol-
lows: (1) solution pH values were not mentioned in the manuscript,
The linearized transformation of the intra-particle diffusion (2) initial pH values were neither measured nor adjusted and final
model (Weber and Morris, 1963) is presented as follows: pH values were measured, (3) initial and final pH values were
pffiffi measured, (4) initial pH values were adjusted and final pH values
qt ¼ kp t þ C (46) were not mentioned, (5) initial pH values were adjusted and final
pH values were measured, (6) initial and final pH values were
where kp (mg/g  min1/2) is the rate constant of the intra-particle adjusted, (7) pH was controlled several times during the course of
diffusion model and C (mg/g) is a constant associated with the the reaction, (8) a constant pH was maintained throughout the
thickness of the boundary layer, where a higher value of C corre- equilibration time by adding acidic or alkaline solutions, and (9)
sponds to a greater effect on the limiting boundary layer. experiments were conducted at a constant pH using a buffered
Generally, although the PSO model can adequately describe solution (Smiciklas et al., 2009).
adsorption kinetic experimental data, this model does not reveal Kumar (2006a) recommended that presenting a plot of qe versus
the adsorption mechanisms. Similarly, the Elovich equation seems Ce for the complete adsorption isotherm in an adsorption study
to describe various reaction mechanisms, such as bulk diffusion and plays a key role in identifying the regions (e.g., Henry, Freundlich,
surface diffusion. In contrast, the intra-particle diffusion model can Langmuir, and BET) in which the experimental data relating to
be useful for identifying the reaction pathways and adsorption adsorption equilibrium are actually located. In contrast, plots of the
mechanisms and predicting the rate-controlling step. In a solid- Freundlich, Langmuir, or other adsorption isotherm models do not
liquid sorption process, adsorbate transfer is often characterized help to identify these regions. Fig. 8a and b provide typical com-
by film diffusion (also known as external diffusion), surface diffu- plete adsorption isotherms that should be presented in the inves-
sion, and pore diffusion, or combined surface and pore diffusion. In tigation of liquid-phase adsorption. Vasanth Kumar and Sivanesan
short, if a plot of qt against t0.5 is linear and passes through the (2006) also recommended that using equilibrium data covering
origin, the adsorption is entirely governed by intra-particle diffu- the complete isotherm was the best way to obtain the parameters
sion. In contrast, if the intra-particle diffusion plot gives multiple in isotherm expressions; equilibrium data with a partial isotherm
linear regions, then the adsorption process is controlled by a was not sufficient, instead equilibrium data that covers the com-
multistep mechanism. plete isotherm was required.
98 H.N. Tran et al. / Water Research 120 (2017) 88e116

Fig. 7. Transport processes during adsorption by a porous adsorbent (Adapted with permission from Weber and Smith, 1987. Copyright (1987) American Chemical Society).

Furthermore, an inconsistent definition of the adsorption questionable conclusions when studying the effects of agitation
isotherm plot in the literature was recognized by Li and Pan (2007); rate or shaking speed on the adsorption capacity (qe) or maximum
the plot of qe against Ce/Wo (with Wo being a liquid/solid ratio) is adsorption capacity (Qomax) of an adsorbent toward an adsorbate.
invisible with the traditional definition of the complete adsorption For example, Chu et al. (2005) concluded that the adsorption ca-
isotherm (the plot of qe versus Ce). pacity of vitamin E onto silica varies as a function of the agitation
Recently, Senthil Kumar et al. (2014) presented a plot of % rate (120e180 rpm). The Qomax values exhibited the following or-
removal versus C0 in an adsorption equilibrium study instead of der: 43.71 mg/g at 180 rpm > 24.20 mg/g at 160 rpm > 24.20 mg/g
presenting the complete adsorption isotherm as shown in Fig. 8a. at 140 rpm > 17.62 mg/g at 120 rpm. However, Choong and Chuah
However, Hai (2017) made a critical note that it is clear that at every (2005) made a critical remark that the agitation rate only impacts
initial concentration of methylene blue (MB; 50, 100, 150, 200, on the speed at which a system reaches equilibrium, but not on the
250 mg/L), the removal percentage of MB by sulfuric acid-treated equilibrium itself. This means that when the adsorption process
orange peel (STOP) at 30  C was always overwhelmingly higher obtains a true equilibrium (sufficient contact time between adsor-
than the corresponding values at 40, 50, and 60  C. This result bent and adsorbate), the Qomax values at different agitation rates
revealed that increasing the temperature from 30 to 60  C resulted will be insignificantly different. This is because the decrease of the
in a decrease in the adsorption capacity (qe) of MB dye onto STOP boundary layer resistance a higher agitation rate will facilitate the
(Fig. 8a). However, the maximum adsorption capacities of STOP transport of the adsorbate to the adsorbent.
from the Langmuir model (Qomax) exhibited the following order: Although adsorption isotherms can insignificantly contribute to
60  C (83.333 mg/g) > 50  C (71.428 mg/g) > 40  C (62.5 mg/ elucidating adsorption mechanisms, they are less helpful in this
g) > 30  C (50 mg/g). This logic problem might be caused by the regard than data on adsorption kinetics and thermodynamics
adsorption process of MB onto STOP not reaching equilibrium at the (Volesky, 2003; Tien, 2007, 2008). However, collecting adsorption
different temperatures (see adsorption isotherms in Fig. 8a) owing isotherms is a useful strategy to both describe the relationship
to the low initial MB concentrations (50e250 mg/L) used at a high between the adsorbate concentration in solution (liquid phase) and
solid/liquid ratio of 4.0 g/L (Hai, 2017). Therefore, in adsorption the adsorbent (solid phase) at a constant temperature and design
equilibrium studies, presenting plots of %removal versus Co, % adsorption systems (Tran et al., 2016).
removal versus Ce, or qe versus Co should be avoided, and a plot of qe A wide variety of models of adsorption isotherms have been
versus Ce for the complete adsorption isotherm must be presented. applied in the literature. These models can be classified as follows:
Fig. 8b provides a typical example of the comparison of (1) irreversible isotherms and one-parameter isotherms (i.e., Henry
adsorption performance under the same operating conditions (i.e., isotherm); (2) two-parameter isotherms (i.e., Langmuir, Freundlich,
temperature and solution pH). Clearly, “adsorbent B” exhibits a DubinineRadushkevich, Temkin, FloryeHuggins, and Hill); (3)
higher affinity than “adsorbent A” at low initial adsorbate con- three-parameter isotherms (i.e., RedlichePeterson, Sips, Toth,
centrations (Co) and low equilibrium adsorbate concentrations (Ce), KobleeCorrigan, Khan, FritzeSchluender, VietheSladek, and Rad-
whereas the opposite order (“adsorbent A” > “adsorbent B”) was keePrausnitz); and (4) more than three-parameter isotherms (i.e.,
found at higher Co and Ce values (Volesky, 2007). Weberevan Vliet, FritzeSchlunder, and Baudu) (Kumar and
It was recently identified that many researchers have made Porkodi, 2006; Hamdaoui and Naffrechoux, 2007; Foo and
H.N. Tran et al. / Water Research 120 (2017) 88e116 99

 
1 1 1 1
¼ o K
þ o (49)
qe Qmax L Ce Qmax

EadieeHoffsiee linearization of the Langmuir model (Type 3):

 
1 qe o
qe ¼ þ Qmax (50)
KL Ce
Scatchard linearization of the Langmuir model (Type 4):

qe o
¼ KL qe þ Qmax KL (51)
Ce

where Qomax (mg/g) is the maximum saturated monolayer


adsorption capacity of an adsorbent, Ce (mg/L) is the adsorbate
concentration at equilibrium, qe (mg/g) is the amount of adsorbate
uptake at equilibrium, and KL (L/mg) is a constant related to the
affinity between an adsorbent and adsorbate. For a good adsorbent,
a high theoretical adsorption capacity Qomax and a steep initial
sorption isotherm slope (i.e., high KL) are generally desirable
(Kratochvil and Volesky, 1998; Wang, 2009).
The limitations of using the four linear forms of the Langmuir
model have been highlighted by Bolster and Hornberger (2007).
The transformation of data for linearization can result in modifi-
cations of error structure, introduction of error into the indepen-
dent variable, and alteration of the weight placed on each data
point, which often leads to differences in the fitted parameter
values between linear and nonlinear versions of the Langmuir
model. For the Type 1 linearization, because Ce and Ce/qe are not
independent, the correlation between Ce and Ce/qe is over-
estimated, i.e., this equation may provide good fits to data that do
Fig. 8. (a) Non-equilibrium adsorption isotherms of methylene blue onto sulfuric acid- not conform to the Langmuir model. For the Type 2 linearization,
treated orange peel at various temperatures (Adapted from Hai (2017). Copyright transformation leads to clustering of data points near the origin and
(2017) Taylor and Francis), and (b) comparison of adsorption isotherm curves: adsor- is extremely sensitive to variability at low values of qe (high values
bent B performs better (higher qe at q10) than adsorbent A at lower equilibrium con-
centrations (e.g., Ce ¼ 10 mg/L) (Adapted from Volesky 2007. Copyright (2007), with
of 1/qe). For the Type 3 linearization, the abscissa is not error free,
permission from Elsevier). and qe/Ce and qe are not independent. Thus, the correlation between
qe/Ce and qe is underestimated, i.e., this equation may provide poor
fits to data that do conform to the Langmuir model. Finally, the
Hameed, 2010). Of these, the Langmuir and Freundlich models are variables qe and qe/Ce are also not independent in the Type 4 line-
the most commonly used, followed by the DubinineRadushkevich arization. In this case, the correlation between qe and qe/Ce is
and RedlichePeterson models, because of the usefulness of their underestimated, i.e., this equation may provide a poor fit to data
model parameters, their simplicity, and their easy interpretability. that do conform to the Langmuir model.

6.2. Langmuir equation

6.2.1. Derivation of the Langmuir equation


6.2.2. Derivation of the separation factor
The theoretical Langmuir equation (Langmuir, 1918), which was
If the experimental data are adequately described by the Lang-
originally applied to the adsorption of gases on a solid surface, was
muir model, it is essential to calculate the separation factor. Hall
developed using the following assumptions: (1) a fixed number of
et al. (1966) originally proposed that the essential characteristics
accessible sites are available on the adsorbent surface and all active
of the Langmuir isotherm model can be expressed in terms of a
sites have the same energy; (2) adsorption is reversible; (3) once an
dimensionless constant called the separation factor or equilibrium
adsorbate occupies a site, no further adsorption can occur on that
parameter RL, which is defined as follows:
site; and (4) there is no interaction between adsorbate species. The
nonlinear form of the Langmuir model is described in Eq. (47) and
its four linearized forms are shown in Eqs. (48)e(51). Table 1
Relationship between isotherm parameters and isotherm shapes.
o K C
Qmax L e
qe ¼ (47) Freundlich exponent Separation factor Isotherm shapes Remarks
1 þ KL Ce
n ¼0 RL ¼0 Irreversible Horizontal
HaneseWoolf linearization of the Langmuir model (Type 1): n <1 RL <1 Favorable Concave
n ¼1 RL ¼1 Linear Linear
 
Ce 1 1 n >1 RL >1 Unfavorable Convex
¼ o
Ce þ o K
(48)
qe Qmax Qmax L Note: Information was adapted from Worch (2012). Copyright (2012), with
permission from De Gruyter. RL and n were calculated from Equations (52) and (56),
LineweavereBurk linearization of the Langmuir model (Type 2): respectively.
100 H.N. Tran et al. / Water Research 120 (2017) 88e116

1 1
RL ¼ (52) RL ¼ (55)
1 þ KL Co 1 þ KL Qmax
o C
o

where RL is a constant separation factor (dimensionless) of the Recently, Hai (2017) commented that “the authors' conclusion
solid-liquid adsorption system, KL is the Langmuir equilibrium about separation factor (RL) was not valid. It is noted that there is no
constant, and Co (mg/L) is the initial adsorbate concentration. information or value of RL mentioned in the paper's results and dis-
Notably, the isotherm shape can be predicted using the sepa- cussion. It is impossible for the authors to conclude that the values of RL
ration factor (RL) (Weber and Chakravorti, 1974) and the Freundlich were found to be between 0 and 1, indicating the favorable adsorption
exponent n (Worch, 2012). Table 1 summarizes the various of MB dye onto STOP.”
isotherm shapes. Clearly, the separation factor is related to the Langmuir equation
in the adsorption isotherm. Nevertheless, Goswami et al. (2016)
6.2.3. Mistakes concerning the Langmuir model and the separation recently described an incorrect relationship between the separa-
factor tion factor and adsorption kinetic models (i.e., PFO, PSO, and intra-
Wang et al. (2014) published a corrigendum because of a particle diffusion) at different initial cadmium concentrations. This
misapplication of the nonlinear form of the Langmuir equation (Eq. is a fundamental misunderstanding that should be avoided.
(53)). A similar mistake in presenting the nonlinear form of the
Langmuir equation (Eq. (54)) was observed by Kumar and Porkodi 6.3. Freundlich equation
(2007b).
6.3.1. Derivation of the Freundlich equation
Ce Ce K The Freundlich equation is one of the earliest empirical equa-
¼ o þ oL (53)
qe Qmax Qmax tions used to describe equilibrium data and adsorption character-
istics for a heterogeneous surface (Freundlich, 1906). Unlike the
Ce 1 1 Langmuir equation, the Freundlich isotherm can describe neither
¼ o
þ o (54) the (arithmetic) linearity range at very low concentrations nor the
qe KL Qmax Qmax Ce
saturation effect at very high concentrations. Hence, the Freundlich
Furthermore, an unexpected typographical error was made in isotherm does not describe the saturation behavior of an adsorbent
reference (Egirani, 2004). As a result, the sentence “a plot of Ce/qe (Wang, 2009). The nonlinear and linear forms of the Freundlich
versus Ce should indicate a straight line of slope 1/KLQomax and an equation can be expressed as shown in Eqs. (56) and (57),
intercept of 1/Qomax” was corrected to “a plot of Ce/qe versus Ce should respectively.
indicate a straight line of slope 1/Qomax and an intercept of 1/KLQomax”
by Ho (2004a). qe ¼ KF Cen (56)
The incorrect expression and inaccurate calculation of RL were
recognized and discussed by Kumar (2006a, 2006b). First, the log qe ¼ n log Ce þ log KF (57)
separation factor RL is not a Langmuir constant, and is only
employed for isotherms that obey the Langmuir model. Second, RL where qe (mg/g) is the amount of adsorbate uptake at equilibrium,
values will vary with the initial adsorbate concentration, irre- Ce (mg/L) is the adsorbate concentration at equilibrium, KF (mg/g)/
spective of the shape of the isotherm, and RL values at different (mg/L)n is the Freundlich constant, and n (dimensionless) is the
initial adsorbate concentrations cannot be calculated directly from Freundlich intensity parameter, which indicates the magnitude of
the Langmuir isotherm, but can be calculated from the Langmuir the adsorption driving force or the surface heterogeneity.
equilibrium constant KL. These misunderstandings and mistakes
are summarized in Table 2. 6.3.2. Mistakes concerning the Freundlich equation
Furthermore, a mistake related to the definition of RL compared Several investigators have identified an incorrect expression
with the original definition was stated by Fu et al. (2007). The (Eq. (58)) of the linear form of the Freundlich equation (Kumar and
incorrect presentation is shown in Eq. (55). Porkodi, 2007b; Alagumuthu et al., 2010).

1
Table 2 log qe ¼ þ log KF (58)
Mistakes in the presentation and calculation of the separation factor (RL).
n log Ce

Typical examples Remakes Furthermore, several discussions that are inconsistent with the
original concepts of the Freundlich theory have been identified in
Langmuir parameters The RL is not the Langmuir constant,
so it is mistake for presentation. the literature (Lu, 2008; Hai, 2017). Hai (2017) commented that the
Qomax KL RL exponent n of the Freundlich equation must be dimensionless.
Orange peel 54.5 0.101 0.01e0.17 Therefore, the original authors incorrectly reported the units of the
OP Biochar 115 0.019 0.08e0.57 exponent n (g/L); and there is no validation for the statement of the
Langmuir RL The RL values must be calculated at
significance of the n value for the adsorption of MB dye onto STOP is
parameters different range of initial adsorbate
concentration, so it is mistake for as follows: when n ¼ 1 the adsorption is linear, when n < 1 the
Qomax KL
calculation. adsorption is a chemical process, and when n > 1 the adsorption is a
Orange peel 54.5 0.101 0.17 physical process. According to the Freundlich theory, the adsorp-
OP Biochar 115 0.019 0.57 tion isotherm becomes linear when n ¼ 1, favorable when n < 1, and
Langmuir RL Correction presentation and
unfavorable when n > 1. Similarly, Harsha et al. (2015) also made a
parameters calculation.
basic mistake when they reported the units of the exponent n as g/
Qomax KL
L.
Orange peel 54.5 0.101 0.01e0.17 Lu (2008) highlighted that in many cases KF is not the maximum
OP Biochar 115 0.019 0.08e0.57 adsorption capacity and the KF value is only equal to the maximum
Data were modified from our previous publication (Tran et al., 2015). adsorption capacity when n approaches infinity. Although KF is not
H.N. Tran et al. / Water Research 120 (2017) 88e116 101

defined as the maximum adsorption capacity (Qomax), KF values and exponent g have been reported elsewhere (Inbaraj, 2006; Kumar
Qomax values should be of the same order. Such a mistake has been and Porkodi, 2007a; Kumar et al., 2007; Kumar et al., 2007;
discussed by Tran (2017), where the reported KF values for the Kumar and Porkodi, 2008). These mistakes involved values of the
adsorption of Cu2þ onto raw pomegranate peel biosorbent followed exponent g that were outside the range of 0e1 (Table 3). It is not
the order: 4.471 at 303 K > 3.439 at 313 and 323 K > 2.877 at 333 K, easy to predict the parameters of the RedlichePeterson equation
while the Qomax values were 30.12 mg/g at 313 K > 21.367 mg/g at because there are three unknown parameters. Although a new
323 K > 21.276 mg/g at 333 K > 20.492 mg/g at 303 K. This mistake linear form of the RedlichePeterson equation was published by Wu
might be attributable to a miscalculation. et al. (2010), the nonlinear form might be more appropriate for
Notably, Kumar (2006a) highlighted that values for the expo- accurately calculating the parameters of adsorption models with
nent n (Eq. (59)) in the range of 0e10 suggest favorable adsorption. more than two unknown parameters, such as the RedlichePeterson
If the experimental equilibrium data do not lie in the Henry region, model.
then the linear regression method will just check the hypothesis
instead of verifying the theory behind the Freundlich model;
therefore, it is not possible in practice to obtain n > 10 (Kumar, 6.5. DubinineRadushkevich equation
2006a). A corresponding error was also found in a published pa-
per (Huang et al., 2014), in which the n values reported for the 6.5.1. Derivation of the DubinineRadushkevich equation
removal of aniline and Cr(VI)/aniline þ Cr(VI) by an activated car- The DubinineRadushkevich equation was developed to account
bon/chitosan composite were 10.20 and 11.34, respectively. for the effect of the porous structure of an adsorbent (Dubinin and
Radushkevich, 1947), and is expressed as follows:
1=n
qe ¼ KF Ce (59)
qe ¼ qDR eKRD ε
2
(61)
The linear form of the DubinineRadushkevich equation is:
6.4. RedlichePeterson equation
ln qe ¼ KDR ε2 þ ln qDR (62)
6.4.1. Derivation of the RedlichePeterson equation  
A three-parameter RedlichePeterson isotherm was proposed 1
ε ¼ RT ln 1 þ (63)
upon considering the limitations of the Freundlich and Langmuir Ce
isotherms (Redlich and Peterson, 1959). This model incorporates
By inserting Eq. (63) into Eq. (62), Eq. (64) can be obtained:
the features of the Freundlich and Langmuir models and might be
 
applicable for demonstrating adsorption equilibrium over a wide 1
range of adsorbate concentrations. The nonlinear form of this ln qe ¼ KDR R2 T 2 ln2 1 þ þ ln qDR (64)
Ce
empirical model is given as follows:
The parameters qDR and KDR in Eq. (64) can be obtained as fol-
KRP Ce lows: (1) a plot of lnqe against ln2(1 þ 1/Ce) has a slope ¼ KDRR2T2
qe ¼ g (60)
1 þ aRP Ce and an intercept ¼ lnqRD, and the E value can be obtained using Eq.
(65), and (2) a plot of lnqe against R2T2ln2(1 þ 1/Ce) has a
where KRP (L/g) and aRP (mg/L)g are the RedlichePeterson con- slope ¼ KDR and an intercept ¼ lnqRD, and the E value can be
stants and g (dimensionless) is an exponent whose value must lie obtained using Eq. (66). Notably, the E values obtained from Eqs.
between 0 and 1. (65) and (66) are the same (Tran et al., 2016).
Eq. (60) becomes a linear isotherm (Henry's law equation) at
low surface coverage (g ¼ 0), reduces to the Langmuir isotherm 1 RT
E ¼ pffiffiffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (65)
when g ¼ 1, and transforms into the Freundlich isotherm when KRP 2KDR 2slope
and aKP [ 1 and g ¼ 1. Therefore, if the g value is outside the range
of 0e1, the data is not adequately explained by the RedlichePerson 1 1
equation. Accurate calculation of g can be helpful in expatiation of E ¼ pffiffiffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (66)
2KDR 2slope
where the isotherm presentations, neither in the Freundlich or
Langmuir (Vasanth Kumar and Sivanesan, 2006). where qRD (mg/g) is the adsorption capacity, KRD (mol2/kJ2) is a
constant related to the sorption energy, 3 is the Polanyi potential, E
6.4.2. Mistakes concerning the RedlichePeterson equation (kJ/mol) is the mean adsorption energy, R is the gas constant, T is
A series of comments related to mistakes in calculating the the temperature in Kelvin, and qe and Ce are obtained from Eq. (1).

Table 3
Incorrect values of the exponent g in the RedlichePeterson equation.

Adsorbent Adsorbate g Reference

Sawdust Reactive black 1.10 Inbaraj (2006)


Sawdust Reactive red 1.04 Inbaraj (2006)
Anilinepropylsilica xerogel Congo red 2.91 Kumar and Porkodi (2008)
Xad-4 resin 4-chlorophenol at 303 K 1.07 Kumar and Porkodi (2007a)
Xad-4 resin 4-chlorophenol at 318 K 1.19 Kumar and Porkodi (2007a)
Tectona grandis Cu2þ 0.35 Vasanth Kumar et al. (2007)
Syzygium cumini Pb2þ 1.23 Kumar et al. (2007)
Activated carbon/chitoshan composite Aniline/aniline þ Cr(VI) 1.04 Huang et al. (2014)
Activated carbon/chitoshan composite Cr(VI)/aniline þ Cr(VI) 1.04 Huang et al. (2014)
Activated carbon/chitoshan composite Cr(VI) 1.02 Huang et al. (2014)
102 H.N. Tran et al. / Water Research 120 (2017) 88e116

6.5.2. Mistakes concerning the DubinineRadushkevich equation the absolute temperature in Kelvin.
Some authors have highlighted erroneous expressions of the The Gibbs energy change (DG ) is directly calculated from Eq.
DubinineRadushkevich equation (Cavas, 2008; Fu et al., 2008). (71), whereas the enthalpy change (DH ) and entropy change (DS )
These erroneous forms are shown in Eqs. (67)e(70). are determined from the slope and intercept, respectively, of a plot
Incorrect nonlinear form of the DubinineRadushkevich of lnKC versus 1/T (Eq. (73)). The common units of DG , the gas
equation: constant, and temperature are J/mol, J/(mol  K), and K, respec-
 tively; therefore, the equilibrium constant KC in Eq. (71) must be
qe ¼ qDR expð  KDR ðRTðln 1 þ 1=Ce Þ2 (67) dimensionless (Milonji c, 2007, 2009b; Canzano et al., 2012;
Dawood and Sen, 2012; Zhou et al., 2012; Zhou and Zhou, 2014;
Incorrect linear form of the DubinineRadushkevich equation: Tran et al., 2016; Ghosal and Gupta, 2017; Hai, 2017; Rahmani-
  Sani et al., 2017; Zhou, 2017). This discussion is in accordance
1 with the assumption of the International Union of Pure and Applied
ln qe ¼ 2KDR RT ln 1 þ þ ln qDR (68)
Ce Chemistry.
Eq. (72) describes the relationship of DG with DH and DS , and
Incorrect expressions for the Polanyi potential (3 ):
it is useful for checking the logic of calculated values for these
ε ¼ RTe1=Ce (69) thermodynamic parameters. Unfortunately, associated mistakes
have been found in the literature. Fu et al. (2008) substituted the
  values of DH and DS into Eq. (72) to obtain DG , but the obtained
1
ε ¼ RT ln (70) DG values were significantly different from the DG values ob-
Ce tained using Eq. (71). Fu et al. (2008) highlighted that this mistake
Additionally, the magnitude of E may give useful information might result from incorrect application of the method for calcu-
about the type of adsorption process (physical or chemical) (Fu lating the equilibrium constant.
et al., 2008; Tran et al., 2016). Fu et al. (2008) commented that Clearly, accurate estimation of thermodynamic parameters is
the E values for the removal of phenolic compounds by organo- directly dependent on accurate determination of the equilibrium
philic bentonite can be categorized as corresponding to ion ex- constant between two phases (KC; dimensionless). In the literature,
change (E ¼ 8e16 kJ/mol), whereas the original authors made an the thermodynamic parameters can be calculated from KC values
inconsistent conclusion that the adsorption of two phenolic com- derived from adsorption-isotherm constants (i.e., Langmuir,
pounds could be regarded as physical adsorption. Freundlich, Frumkin, FloryeHuggins, and Henry) or the partition
Recently, Tran (2017) made the following comment on the coefficient (Liu, 2009; Doke and Khan, 2013; Tran et al., 2016).
conclusions of the original paper: “the obtained value of E in this However, to obtain an appropriate calculation method, several
work varies from 32.596 kJ/mol to 40.572 kJ/mol for studied temper- factors need to be considered thoroughly. First, the equilibrium
ature which is higher than the adsorption energy values previously constant (KC) must be dimensionless. Second, the linear regression
enumerated.” However, these adsorption energy (E) values might coefficient (R2) of the van't Hoff equation must be high. Third, the
not have been calculated correctly. Assuming the b values at 303 K concentration of adsorbate used in the adsorption equilibrium
(b ¼ 1.93  108 mol2/kJ2), 313 K (2.99  108), 323 K (2.00  108), study (adsorption isotherm) are low or high. In addition, the tem-
and 333 K (2.35  108) reported by these authors are correct, the peratures used for calculating the thermodynamic parameters
recalculated adsorption energy (E) values are 5,090 kJ/mol at 303 K, must have units of Kelvin (K), not degrees Celsius ( C). The incorrect
4,089 kJ/mol at 313 K, 5,000 kJ/mol at 323 K, and 4,613 kJ/mol at use of temperature units led to the publication of an erratum by Ho
333 K. Thus, the E values in this study ranged from 4,089 to 5,091 kJ/ (2004a), with accurate recalculations.
mol, which is impossible for a heavy-metal biosorption process. As the equilibrium constant KC can be derived using a number of
approaches, considerable variation in the thermodynamic param-
eters can be obtained; therefore, the most appropriate approach
7. Adsorption thermodynamics should be determined. Here, we thoroughly discuss the derivation
of the equilibrium constant KC from adsorption isotherms (i.e.,
7.1. Principles of adsorption thermodynamics Langmuir, Freundlich, and Henry constants) and the partition co-
efficient because of their popularity in the literature.
Thermodynamic studies are an indispensable component of
predicting adsorption mechanisms (e.g., physical and chemical).
The key distinctions between physical and chemical adsorption
were summarized in detail in a recent report (Tran et al., 2016). The 7.2. Equilibrium constant derived from the Langmuir constant (KL)
thermodynamic parameters can be computed according to the laws
of thermodynamics using the following equations: The Langmuir equation was initially derived from a kinetic study
and then subsequently from a thermodynamic study. The deriva-
DG ¼ RT ln KC (71) tion of the Langmuir equation from the thermodynamic perspec-
tive can be described as follows (Crittenden et al., 2012; Tran et al.,
The relationship between DG and DH and DS is described as
2016). The relationship between vacant surface sites on the surface
follows:
of an adsorbent (Sv; mmol/m2), adsorbate species in solution (A;
DG ¼ DH   T DS (72) mmol), and adsorbate species bound to surface sites (SA; mmol/m2)
can be described by the following reaction:
The well-known van't Hoff equation is obtained by substituting
Eq. (71) into Eq. (72) SV þ A#SA (74)

DH 1 DS On the basis of the Langmuir expression, it is assumed that the
ln KC ¼  þ (73) reaction has a constant Gibbs energy change (DG ; J/mol) for all
R T R
sites, so the thermodynamic equilibrium constant (KC; dimen-
where R is the universal gas constant (8.3144 J/(mol  K)) and T is sionless) can be expressed as:
H.N. Tran et al. / Water Research 120 (2017) 88e116 103

DG DH 1 DS


DG ¼ RT ln KC ⇔KC ¼ e RT (75) lnð55:5  1; 000  KL Þ ¼  þ (81)
R T R
Another Langmuir assumption is that each site is capable of
binding at most one adsorbate molecule (monolayer). According to where the factor 55.5 is the number of moles of pure water per liter
the equilibrium condition, the thermodynamic equilibrium con- (1,000 g/L divided by 18 g/mol) and the term 55.5  1,000  KL is
stant may be written as: dimensionless.
In the case of KL with units expressed in L/mg, Milonjic (2007
and 2009b) stated that KL could be obtained as a dimensionless
SA DG
KC ¼ ¼ e RT (76) parameter by multiplying KL by 106 (Eq. (82)). However, Zhou and
SV ½A
Zhou (2014) recommended that KC could be obtained as a dimen-
sionless parameter by multiplying KL by the molecular weight of
where [A] is the concentration of adsorbate A in solution at equi-
the adsorbate (Mw; g/mol), by 1,000, and then by 55.5 (Eq. (85)).
librium (mg/L), R is the universal gas constant (8.314 J/mol  K), and
T is absolute temperature K (273 þ  C).
KC ¼ 106 KL (82)
The critical problem with Eq. (76) is that there are two unknown
parameters (i.e., SV and [A]). However, this problem can be solved if  
the total sites available or monolayer coverage (ST; mol/m2) are DG ¼ RT ln 106 KL (83)
fixed:
  DH 1 DS
SA K ½AST ln 106 KL ¼  þ (84)
ST ¼ SV þ SA ¼ þ SA⇔SA ¼ C (77) R T R
KC ½A 1 þ KC ½A
where the factor 106 is the solution density (assuming the density
The expression of SA in units of mmol/m2 is not useful in mass
of pure water is 1.0 g/mL) and the term 106  K is dimensionless.
balance, and units of mass of adsorbate adsorbed per unit mass of
adsorbent (qe; mg/g) are much more useful. The mass loading (qe)
KC ¼ Mw  55:5  1; 000  KL (85)
can be obtained in units of mg/g (Eq. (1)) by multiplying both sides
of Eq. (71) by the surface area (SBET; m2/g) of the adsorbent and the
molecular weight of the adsorbate (Mw; g/mol). Thus, Eq. (77)
DG ¼ RT lnðMw  55:5  1; 000  KL Þ (86)
becomes:
DH 1 D S
lnðMw  55:5  1; 000  KL Þ ¼  þ (87)
KC ½AST SBET Mw Q o K Ce R T R
SA:SBET $Mw ¼ ⇔qe ¼ max L (78)
1 þ KC ½A 1 þ KL Ce
where the factor 55.5 is the number of moles of pure water per liter
and the term Mw  55.5  1,000  KL is dimensionless.
where qe (mg/g) ¼ SA  SBET  Mw is described by Eq. (1); Ce (mg/
In a recent paper, Tran et al. (2016) applied Eqs. (79)e(87) to
L) ¼ [A] is the concentration of adsorbate in solution at equilibrium;
compare thermodynamic parameters calculated from the Langmuir
Qomax (mg/g) ¼ ST  SBET  Mw is the maximum monolayer
constants derived from four linear forms of the Langmuir model.
adsorptive capacity of the adsorbent when the surface sites are
They concluded that (1) both the recommendations of Milonji c
saturated with adsorbate; the relationship between KL and KC is
(2007 and 2009b) and Zhou and Zhou (2014) provide the same
described in the following discussion.
methods for calculating the thermodynamic parameters, and (2)
However, the main problem is that the Langmuir constant KL is
there is some deviation between these methods for the results from
dimensional with common units of L/mmol or L/mg, while the
Types 1, 3, and 4; only Type 2 produces identical results.
equilibrium constant KC is dimensionless (without units). Thus, the
direct application of KL (L/mmol or L/mg) in the calculation of
thermodynamic parameters produces incorrect results, as dis- 7.3. Equilibrium constant derived from the Freundlich constant (KF)
cussed by many authors (Milonji c, 2007, 2009b; Canzano et al.,
2012; Dawood and Sen, 2012; Zhou et al., 2012; Zhou and Zhou, The Freundlich equation is consistent with the thermodynamics
2014; Anastopoulos and Kyzas, 2016; Tran et al., 2016; Ghosal and of heterogeneous adsorption (Crittenden et al., 2012). The
Gupta, 2017; Hai, 2017; Rahmani-Sani et al., 2017). To solve this Freundlich constant KF can be obtained as a dimensionless value
unit problem, several methods have been recommended (Milonjic, using Eq. (88) (Ghosal and Gupta, 2015; Tran et al., 2016).
2007; Zhou and Zhou, 2014; Tran et al., 2016). Depending on the  
units of KL, the equilibrium constant KC can easily be obtained as a   11n
KF r 106
dimensionless constant. KC ¼ (88)
When an adsorption study is conducted in aqueous solution and 1000 r
KL has units of L/mmol, KC can be easily obtained as a dimensionless  1
0
parameter by multiplying KL by 55.5 and then by 1,000 (Eq. (79)).  6  11n
This method was originally proposed by Milonji c (2007) and then B KF r 10 C
DG ¼ RT lnB
@1000
C
A (89)
developed by Zhou and Zhou (2014) and Tran et al. (2016). The r
values of the parameters DG , DH , and DS can be calculated using
the following equations:
0  1

 11n
KC ¼ 55:5  1; 000  KL (79) B KF r 106 C DH  1 DS
lnB
@1000

A x þ (90)
r R T R
DG ¼ RT lnð55:5  1; 000  KL Þ (80)
where r is the density of pure water (assumed as ~1.0 g/mL).
104 H.N. Tran et al. / Water Research 120 (2017) 88e116

It is noted that the units of KF rely on the units used for the Table 4
liquid-phase concentration (C) and solid-phase concentration (q). Comparison of incorrect and correct thermodynamic parameters for methylene blue
adsorption onto sulfuric acid-treated orange peel.
Units of mg/L or mmol/L for C and mg/g or mmol/g for q are used
most frequently to demonstrate adsorption from water solutions. Incorrect thermodynamic parameters reported by Senthil Kumar et al. (2014)
The differing units of KF can be reciprocally converted using Eq. (91) Co (mg/L) DH (kJ/mol) DS (J/mol/K)a DG (kJ/mol)
(Worch, 2012).
30  C 40  C 50  C 60  C

53.7 137 12.5 10.4 8.87 8.49


KF ¼ KF0 ðMw Þ11=n (91) 50
100 37.4 90.3 10.1 9.15 8.04 7.46
150 32.9 79.4 8.86 8.19 7.27 6.52
where the units of the Freundlich constant KF are (mg/g)/(mg/L)1/n, 200 27.5 64.2 8.03 7.54 6.64 6.19
with Co and Ce (mg/L) and qe (mg/g); the units of KF0 are (mmol/g)/ 250 23.7 54.4 7.29 6.61 6.17 5.64
(mmol/L)1/n, with Co and Ce (mmol/L) and qe (mmol/g); and Mw is Co (mg/L) KC values
the molecular weight of the adsorbate. The exponent 1/n is not
30  C 40  C 50  C 60  C
affected by the units of the liquid-phase concentration (C) and
solid-phase concentration (q). 50 142 54.9 27.2 21.5
100 54.8 33.6 19.9 14.8
Based on a comparison of the thermodynamic parameters 150 33.7 23.3 14.9 10.5
calculated from the Freundlich constant (dimensionless) and 200 24.2 18.1 11.9 9.35
Langmuir constant (dimensionless), Tran et al. (2016) concluded 250 18.1 12.7 9.97 7.66
that the signs and magnitudes of DG , DH , and DS calculated from Correct thermodynamic parameters suggested by Hai (2017)
KF were consistent with those calculated from KL. Of course, the
Temperature KC DG (kJ/ DH (kJ/ DS (J/mol  van't Hoff
experimental data in the adsorption isotherms must be fitted well (K) mol) mol) K) equation
by the Langmuir and Freundlich models, and the R2 value of the
303 5.25 4.18 13.65 31.39 y ¼ 1642e3.78
van't Hoff equation must be higher than 0.90 (Tran et al., 2016). It is
313 4.33 3.81 R2 ¼ 0.978
also evident that, to some extent, the application of KF to estimate 323 3.54 3.40
the thermodynamic parameters should be approached with 333 3.27 3.28
caution. a
The unit of J/mol/K was expressed as the original paper; the correction unit
might be J/(mol  K).

7.4. Equilibrium constant derived from the partition coefficient (Kp)

Changes in the thermodynamic partition coefficient with


changes in temperature were first reported by Biggar and Cheung calculation of the thermodynamic parameters for the adsorption of
(1973). The equilibrium constant can be defined as follows: MB onto STOP (Table 4). Hai (2017) highlighted that it is impossible
to obtain an equilibrium constant, Gibbs energy change (DG ),
as g Cs enthalpy change (DH ), and entropy change (DS ) for the adsorp-
Kp ¼ ¼ s (92)
ae ge Ce tion process at every initial MB concentration, as shown in Table 4.
Assuming that the experimental data reported by Senthil Kumar
where as is the activity of the adsorbate adsorbed onto the adsor- et al. (2014) are correct, Hai (2017) recalculated the thermody-
bent, ae is the activity of the adsorbate in solution at equilibrium, gs namic parameters based on the method of Biggar and Cheung
is the activity coefficient of the adsorbate adsorbed onto the (1973), and the corrected thermodynamic parameters are also lis-
adsorbent, ge is the activity coefficient of the adsorbate in solution ted in Table 4. Likewise, Lima et al. (2015) also determined the value
at equilibrium, Cs is the concentration of adsorbate adsorbed onto of the adsorption equilibrium constant (KC) from the value of the
the adsorbent at equilibrium (mg/L), and Ce is the concentration of best fit nonlinear isotherm equilibrium model instead of using the
adsorbate in solution at equilibrium (mg/L). Cs is defined by the equilibrium constant (Kp) calculated from the initial and equilib-
mass balance of adsorbate that disappears from the solution, which rium concentrations of the adsorbate.
should appear on the adsorbent. Importantly, according to the United States Environmental
When the concentration of adsorbate in the solution approaches Protection Agency (USEPA, 1999), the partition coefficient is only
zero, which results in Cs / 0 and Ce / 0, the activity of coefficient appropriate for calculating the thermodynamic parameters if the
g approaches unity, and Eq. (92) can be written as: initial used adsorbate concentration is low. Analogous conclusions
can be found in the literature (Liu, 2009; Salvestrini et al., 2014;
Cs as Lima et al., 2015; Tran et al., 2016; Hai, 2017). It has been high-
Lim ¼ ¼ Kp (93)
Cs/0 Ce ae lighted that the partition coefficient is only equal to the thermo-
dynamic equilibrium constant when the adsorbed concentrations
In this method, Kp values can be obtained plotting ln(Cs/Ce) versus
in the solutions are very low (infinite dilution). In this situation, the
Cs and extrapolating Cs to zero. If a straight line fits the data with a
partition coefficient can be used to calculate the thermodynamic
high regression coefficient (R2) and its intersection with the vertical
parameters. Recently, Tran et al. (2016), who compared the ther-
axis provides the value of Kp, the partition coefficient will be in
modynamic parameters calculated from the Langmuir constant and
unison with the equilibrium constant. The DG value can be directly
the partition coefficient, concluded that although the signs of the
calculated using Eq. (71), while the values of DH and DS are
calculated DG , DH , and DS parameters were the same, the
determined from the slope and intercept, respectively, of Eq. (73).
partition coefficient might not be appropriate for calculating the
This method has been widely applied in the literature. A sum-
thermodynamic parameters owing to the low regression coefficient
mary of the applications of this method can found in the critical
of the van't Hoff equation (R2 ¼ 0.53).
review in reference (Doke and Khan, 2013). However, incorrect
applications of Biggar and Cheung's definition have also been
identified in the literature. In a typical experiment, Senthil Kumar
et al. (2014) made a mistake in the application of Kp during
H.N. Tran et al. / Water Research 120 (2017) 88e116 105

7.5. Equilibrium constant derived from the distribution coefficient parameters must have a logical relationship with the relevant
(Kd) experimental data. For example, the adsorption capacity of adsor-
bate onto an adsorbent (i.e., qe or Qomax) increased with increasing
In 1987, Khan and Singh (1987) proposed a method for calcu- temperature, which means that the enthalpy change must be
lating thermodynamic parameters (Eq. (94)) based on the method positive (þDH ).
of Biggar and Cheung (1973). The distribution coefficient (Kd) can Notably, numerous recent studies have reported methods of
be defined as follows: calculating thermodynamic parameters from the partition coeffi-
cient (Kp) or distribution coefficient (Kd) that differed from the
qe traditional concepts defined in Sections 7.4 and 7.5, respectively. A
Kd ¼ (94)
Ce typical mistake is present in the work of Nekouei et al. (2016). They
studied the adsorption thermodynamics at only one initial adsor-
In this method, Kd values are obtained by plotting ln(qe/Ce)
bate concentration and different temperatures instead of using a
against Ce and extrapolating to zero Ce. If a straight line fits the data
range of adsorbate concentrations (i.e., from 100 mg/L to 1,000 mg/
with a high regression coefficient (R2), then its intersection with the
L) and different temperatures (i.e., 10  C, 20  C, 30  C, and 50  C).
vertical axis provides the value of Kd. Fundamentally, Eq. (94) can
This mistake led to invalid conclusions regarding the adsorption
be derived from the Freundlich and Langmuir equations. If n ¼ 1,
thermodynamics.
the Freundlich equation (Eq. (56)) will become Henry's equation for
a linearized isotherm (Eq. (95)), and the Freundlich constant (KF)
8. Others mistakes
will be equal to the Henry constant (KH). For very low concentra-
tions of adsorbate (KL and Ce ≪ 1), the Langmuir equation (Eq. (47))
8.1. Inconsistent data points in experimental data and model fitting
will become Henry's equation (Eq. (95)), and the Henry constant
(KH) will become QomaxKL. The Henry constant (KH) is also referred
Significant differences between the number of data points in an
to as the distribution coefficient (Kd).
experiment and those used for model fitting should be avoided. For
example, Azizian (2008a) made the following critical comments
qe ¼ KH Ce (95)
about the paper of Karadag et al. (2007): there were six experi-
This method has been widely applied to calculating thermody- mental data points regarding the adsorption kinetics (plots of qt
namic parameters in the literature. Unfortunately, the units of Kd in versus t), but only five experimental data points were used for
Eq. (94) are L/g as Ce has units of mg/L and qe has units of mg/g. fitting of the experimental data to the PSO model (plot of t/qt versus
Therefore, it is impossible to directly employ Kd as the equilibrium t). This omission resulted in inaccurate calculations and question-
constant for the calculation of DG , DH , and DS . This problem able conclusions. The R2 values of the PSO model were very high
associated with the units has been discussed by several scientists (0.992e1.000), but the difference between the experimental (qe,exp)
(Patrickios and Yamasaki, 1997; Milonjic, 2007; Canzano et al., and calculated (qe,cal) values was more than 100% (for example
2012; Dawood and Sen, 2012; Tran et al., 2016; Zhou, 2017). qe,exp ¼ 6.720 mg/g and qe,cal ¼ 15.526 mg/g). Clearly, the qe values
Canzano, Iovino et al. (2012) suggested that the partition coefficient calculated from the experimental data and the model were signif-
Kd (L/g) could be converted to KC (dimensionless) by multiplying Kd icantly different, but the authors (Karadag et al., 2007) concluded
by a factor of 1,000, as originally proposed by Milonji c (2007 and that “The pseudo-second-order kinetic model agrees very well with
2009b). the dynamic behavior for the adsorption of dyes RR239 and RB5 onto
Similar to the partition coefficient, the distribution coefficient CTAB-zeolite under several different dye concentrations, temperatures,
may be appropriate for estimating the thermodynamic parameters and pH values”; the validity of this conclusion is very questionable.
when the initial concentration of the adsorbate is very low Lima et al. (2015) observed a change in kinetic parameters when
(Salvestrini et al., 2014; Tran et al., 2016). Tran et al. (2016) five experimental points were deleted to achieve a good R2 value.
concluded that the signs of DG , DH , and DS estimated from the They concluded that the strategy of deleting some data points to
distribution coefficient and the Langmuir constant are the same. improve the R2 value caused serious errors in the values of qe and k1
However, the units of Kd (L/g) are different from those of KL (L/mg), in the PFO equation or k2 in the PSO equation.
and the adsorption equilibrium data do not fit Henry's adsorption Azizian (2008a,b) pointed out that in reference (Karadag et al.,
isotherm. As a result, application of the distribution coefficient was 2007), a comparison was made between the results of fitting four
not appropriate for calculating the thermodynamic parameters in experimental equilibrium data points to the Langmuir model and
their study; the low regression coefficient of the van't Hoff equation fitting five experimental equilibrium data points to the Freundlich
(R2 ¼ 0.42) is further evidence of the inappropriateness of this model. Therefore, it is impossible for the authors (Karadag et al.,
approach. 2007) to conclude that “the Freundlich model exhibited a slightly
In summary, to determine accurate thermodynamic parameters, better fit than the Langmuir model.” To compare the fits of two
several points should be thoroughly considered. First, the equilib- different models, the same experimental data (including the same
rium constant (KC) must be dimensionless. Second, if the concen- number of experimental data points) should be used (Azizian,
trations of adsorbate used to obtain the adsorption isotherm are 2008a). Likewise, Harsha et al. (2015) examined adsorption iso-
very low (infinite dilution), the partition (Kp) or distribution co- therms using seven experimental data points. However, in the plots
efficients (Kd) are appropriate for calculation of the thermodynamic for calculating the Langmuir, Freundlich, and Dubi-
parameters; otherwise, the Langmuir (KL) or Freundlich (KF) con- nineRadushkevich models, only three experimental points were
stant models may be more appropriate. Third, the plot of lnKC used. It is impossible to obtain the Langmuir, Freundlich, and
against 1/T corresponding to the van't Hoff equation must be linear DubinineRadushkevich parameters from only three experimental
with a high regression coefficient (R2). Additionally, it is necessary points. A similar mistake was found in reference (Goswami et al.,
to consider whether the adsorption process can reach equilibrium 2016).
and consideration of the adsorption isotherm shapes and the
adsorption model fit are also recommended. Thus, presentation of 8.2. Oxidation state of chromium
the complete adsorption isotherm (plot of qe versus Ce) is strongly
recommended. Finally, the determined thermodynamic Many investigators (Aoyama, 2003; Park et al., 2006a, 2006b,
106 H.N. Tran et al. / Water Research 120 (2017) 88e116

2006c) have commented that mistakes in analyzing chromium comprised three steps: (1) binding of anionic Cr(VI) ion species to
species in aqueous solution have resulted in the incorrect eluci- positively charged groups on the biomass surface, (2) reduction of
dation of hexavalent chromium adsorption, attributing Cr(VI) Cr(VI) to Cr(III) by adjacent electron-donor groups, and (3) release
removal from an aqueous solution to “anionic adsorption”. Gener- of the Cr(III) ions into the aqueous phase, owing to electronic
ally, when Cr(VI) comes into contact with organic substances or repulsion between the positively charged groups and the Cr(III)
reducing agents, especially in an acidic medium, it is easily or ions, or complexation of Cr(III) with adjacent groups capable of
spontaneously reduced to Cr(III), as Cr(VI) has a high redox po- binding chromium. If there are a small number of electron-donor
tential (above þ1.3 V under standard conditions). Therefore, there groups in the biomass or a low concentration of protons in the
is a strong possibility that the mechanism of Cr(VI) removal by aqueous phase, chromium bound on the biomass can remain in the
biomaterials or biomaterial-based activated carbon is not “anionic hexavalent state. Therefore, the extent to which mechanisms I and
adsorption” but “adsorption-coupled reduction”. Various authors II operate depends on the nature of the biosorption system,
(Aoyama, 2003; Park et al., 2006a, 2006b, 2006c) also suggested including solution pH, temperature, functional groups on the
that in Cr(VI) adsorption studies, it is necessary to analyze both the biomass, and biomass and Cr(VI) concentrations (Park et al., 2006a,
Cr(VI) and total Cr in aqueous solution, as well as determining the 2006b). These two main mechanisms are summarized in Fig. 9.
oxidation state of chromium bound to the biomaterial or activated
carbon. 8.3. Incorrect labels
The authors (Aoyama, 2003; Park et al., 2006a, 2006b, 2006c)
pointed out that in most papers, only Cr(VI) in aqueous solutions Incorrect labeling of compounds should also be avoided. A
was analyzed by the standard colorimetrical method; the pink publication entitled “Selenite adsorption and desorption in main
colored complex formed from between 1,5-diphenycarbazide and Chinese soils with their characteristics and physicochemical proper-
Cr(VI) in an acidic solution can be spectrophotometrically analyzed ties” included such a mistake, as pointed out by Goldberg (2016).
at 540 nm. Similarly, the total Cr in aqueous solutions has been The purpose of this work was to study the adsorption of Se(IV)
analyzed using atomic absorption spectroscopy (AAS) or induc- (selenite), not Se(VI) (selenate); however, the paper reported that
tively coupled plasma-atomic emission spectroscopy (ICP-AES), the authors prepared Se solutions using Na2SeO4 (hexavalent
which cannot distinguish Cr(VI) from the total Cr. Additionally, Se(VI)). This mistake was explained by the incorrect labeling of
Aoyama (2003) underlined that the total Cr present in a solution Na2SeO3 as Na2SeO4 in the Materials and Methods section.
can be determined by oxidizing any Cr(III) formed with KMnO4, Similarly, the author of the paper entitled “Adsorption sequence
followed by the same procedures described for the determination of toxic inorganic anions on a soil” prepared solutions from
of Cr(VI). The adsorption density can be calculated from the dif- Na2HAsO4$7H2O as As(V) species. However, they mislabeled
ference between the initial concentration of Cr(VI) and the final Na2HAsO4$7H2O as NaAsO2, a trivalent As(III) species (Goldberg,
concentration of total Cr. In contrast, the Cr(III) concentration is 2009).
calculated from the difference between the final total Cr and final
Cr(VI) concentrations. 8.4. BET specific surface area of an adsorbent
Several techniques have been applied in the literature to
ascertain the oxidation state of chromium on an adsorbent, such as Using the BrunauereEmmetteTeller (BET) surface area of
X-ray photoelectron spectroscopy (XPS) and X-ray absorption pahokee peat (0.88 m2/g) and assuming that the atrazine molecule
spectroscopy (XAS). For example, Dambies et al. (2001) investi- is a sphere with a radius (r) of 4.16  1010 m, Borisover and Graber
gated the characterization of Cr(VI) interactions with chitosan us- (1997) estimated the maximum adsorption capacity of pahokee
ing XPS. On the basis of their XPS data, they concluded that Cr(VI) peat for atrazine. First, the cross-section area (A) of atrazine can be
was entirely reduced to Cr(III) on glutaraldehyde crosslinked beads, obtained as 3.14  (4.16  1010)2 ¼ 54.4  1020 m2; subsequently,
while only 60% of Cr(VI) was found in its reduced form on raw the surface covered by 1 mol of atrazine (MA) can be calculated as
beads. Additionally, Cr(VI) binding and reduction to Cr(III) by (54.4  1020 m2)  (6.023  1023/mol) ¼ 3.3  105 m2/mol. Sec-
agricultural byproducts of Avena monida (oat) biomass was char- ond, the maximum possible concentration of atrazine on the sur-
acterized by Gardea-Torresdey et al. (2000). Their XAS studies face area (Qomax,covered; maximum adsorption capacity) was
further corroborated that, although Cr(VI) could bind to oat obtained by dividing the BET surface area (0.88 m2/g) by the surface
biomass, it was easily reduced to Cr(III) by positively charged covered by 1 mol of atrazine (3.3  105 m2/mol) to give
functional groups, and Cr(III) was subsequently adsorbed by avail- 2.7  106 mol/g. Finally, Qomax,covered was converted into weight
able carboxyl groups. units using (2.7  106 mol/g)  (215.7 g/mol) to give 582 mg/g. This
A study on using fermentation waste for detoxification of Cr(VI) value (582 mg/g) is substantially higher than the maximum
contaminated aqueous solutions was published by Park et al. adsorption capacity obtained from the Langmuir equation (161 mg/
(2008). In this study, the colorimetric method was used in combi- g). Although these calculations are correct, these inconsistent
nation with excess potassium permanganate to analyze both the values might originate from a misunderstanding by the authors
Cr(VI) and total Cr in aqueous solution, and XPS was used to (Borisover and Graber, 1997) who state that “N2 BET measured
ascertain the oxidation state of the chromium bound on the external surface area”, and thus attributed the value of 582 mg/g to
biomass. Park et al. (2008) concluded that the mechanism of Cr(VI) the maximal possible atrazine concentration on the external sur-
removal by the fermentation waste involved reduction of Cr(VI) to face of pahokee peat.
Cr(III) (redox reaction). As a result, kinetic and equilibrium models According to the 2011 guide for the accelerated surface area and
based on the “reduction” mechanism were used to describe the porosimetry (ASAP) system, the definition of the BET specific sur-
behavior of Cr(VI) and total Cr in aqueous solution. face area (m2/g) includes both the t-Plot external surface area (m2/
Two main mechanisms for removal of Cr(VI) from aqueous so- g) and the t-Plot micropore area (m2/g). Table 5 gives a typical
lution by nonliving biomass were proposed by Park et al. (2005). In example of correct and incorrect use of the Micromeritics report
the first mechanism (direct reduction), Cr(VI) was directly reduced (experimental data measured using a Micromeritics ASAP 2020
to Cr(III) in the aqueous phase by contact with electron-donor sorptometer at 77 K). Although there were no problems found in
groups of the biomass, i.e., groups with lower reduction poten- the N2 adsorption/desorption isotherms, the report results for BET
tials than that of Cr(VI) (þ1.3 V). However, the second mechanism analysis of the prepared activated carbon in Table 5 contain
H.N. Tran et al. / Water Research 120 (2017) 88e116 107

Fig. 9. Proposed mechanisms for Cr(VI) biosorption by nonliving biomass (Adapted from Park et al., 2005. Copyright (2005), with permission from Elsevier).

Table 5 Biosorbents that have not undergone any treatment have never
A typical example of correct and incorrect use of a Micromeritics report. been defined as porous materials; thus, the porosity of a biosorbent
Incorrect data Correct data is negligible.
Therefore, the iodine number cannot be applied for the deter-
Prepared activated carbon Prepared Biochar
mination of porosity for any biosorbent. Moreover, it is impossible
BET surface area 613 536 to obtain the very high iodine number reported for PGP (602 mg/g).
Langmuir surface area 1684 660
Micropore area NA 348
The high iodine number for PGP can be attributed to the high sulfur
External surface area 698 188 content (0.89%) and prolonged contact time (>30 s) (Tran, 2017).
According to the ASTM (D4607-14 2014), after transferring acti-
Note: NA means that the micropore area is not reported because either the micro-
pore volume is negative or the calculation of external surface area is larger than the vated carbon (g) into hydrochloric acid solution (5 wt%), the
total surface area (BET surface area). The unit is m2/g. Data are not published. mixture should be boiled gently for 30 s to remove any sulfur that
may interfere with the test results. Therefore, it is clear that the
iodine number is greatly affected by high sulfur contents. Notably,
the contact time between activated carbon and iodine solution
discrepancies, as the BET specific surface area value is less than the
(0.1 N) strongly impacts the determined iodine number. The con-
external surface area. This error resulted from calculation of the
tact time recommended by the ASTM was approximately 30 s,
specific surface area (SBET) in the range 0.38 < p/p0 < 0.57, instead of
while the contact time used by Ben Ali et al. was approximately
0.05 < p/p0 < 0.3.
4 min. Recently, Tran et al. (2017c) prepared activated carbons from
Recently, Ben Ali et al. (2016) used the “iodine number” to
golden shower using different chemical activation methods with
determine the surface area of a biosorbent without any treatment
K2CO3 (GSAC, GSBAC, and GSHAC). Their results indicated that the
(pomegranate peel; PGP) and concluded that “the specific surface
iodine numbers (mg/g) of the prepared activated carbons at a
area obtained is equal to 598.78 m2/g. Iodine number is generally used
contact time of 30 s were significantly lower than those at 5 min;
as an approximation for surface area and microporosity of active
GSAC (2,604 mg/g < 2,883 mg/g), GSBAC (1,568 mg/g < 2,296 mg/
carbons with good precision”. Tran (2017) commented that there are
g), GSHAC (2,695 mg/g < 4,842 mg/g).
two serious misconceptions in this work that need to be discussed.
Second, the textural properties of an adsorbent (i.e., specific
First, the iodine number of activated carbon is often determined
surface area, total pore volume, and micropore volume) can be
following the internal method, ASTM D4607-14 (D4607-14, 2014).
obtained from conventional analysis of nitrogen adsorption-
According to the ASTM, the iodine number (mg/g; amount of iodine
desorption isotherms, which are measured at 77 K using a sorp-
adsorbed (mg) by 1.0 g of activated carbon) is a relative indicator of
tometer (e.g., Micromeritics ASAP 2020). The BET method is widely
porosity in activated carbon. Iodine molecules (z0.27 nm) can be
employed to compute specific surface area (SBET) (Marsh and
adsorbed into the micropores (pore width > 1 nm) of porous ma-
Reinoso, 2006). Therefore, it is impossible to measure the surface
terials. However, the iodine number does not necessarily provide a
area of an adsorbent using the iodine number method. This
measure of the ability of activated carbon to adsorb other species.
mistaken assumption will lead to incorrect result, such as the
Although the iodine number may be used to approximate the
extremely high surface area of PGP (598.78 m2/g). The BET surface
surface area for several types of activated carbon, it must be real-
areas of various biosorbents are approximately 40 m2/g for yellow
ized that the relationship between surface area and iodine number
passion-fruit shell, 20 m2/g for orange peel, 8.17 m2/g for wheat
cannot be generalized, as it varies with changes in the raw carbon
straw, 8.13 m2/g for sargassum, 4.01 m2/g for Moringa oleifera
material, processing conditions, and pore volume distribution.
108 H.N. Tran et al. / Water Research 120 (2017) 88e116

8.6. Cp-cation and p-p interactions

8.6.1. Cp-cation interactions


Recently, Morosanu et al. (2016) investigated the biosorption of
lead ions onto rapeseed biomass that was used as a biosorbent
without any treatments. On the basis of FTIR measurements before
and after adsorption, they concluded that the disappearance of the
peak at 1710 cm1, which is characteristic of C]O stretching,
indicated surface complexion through Cp-cation interactions.
Similarly, Medellin-Castillo et al. (2017) examined the biosorption
of Pb2þ and Cd2þ onto a biosorbent derived from industrial chili
seeds. They also found changes in the intensity of the absorbance
peak at 1657 cm1 in heavy-metal-loaded chili seeds, suggesting
that the adsorption of Cd2þ and Pb2þ involved p-cation interactions
between the aromatic rings of the lignin and the Cd2þ and Pb2þ
cations in the solution.
In general, Cp-cation interactions are attributed to electrostatic
interactions between the aromatic rings of basic carbonaceous
materials (i.e., biochar, carbon foam, carbon nanotubes, activated
carbon, and graphene) and metallic cations. For example, Tran et al.
(2015) used orange-peel-derived biochar, which was produced
through a carbonization process under limited-oxygen conditions
(also known as pyrolysis) at a high temperature (>400  C). Biochar
is known as a carbon-enriched porous material, similar to activated
carbon. Thus, biochar also possesses a graphite structure (C]C
bonds; p-electrons). As a result, Cp-cation interactions played a
primary role in the adsorption of heavy metals onto biochar.
Analogous conclusions were reached by other scholars
(Swiatkowski et al., 2004; Uchimiya et al., 2010; Rivera-Utrilla and
Sanchez-Polo, 2011).
The contribution of Cp-cation interactions to the mechanism of
heavy-metal adsorption onto carbonaceous material cannot be
denied. However, unlike biochar and activated carbon, biosorbents
Fig. 10. (a) Chemical structure of Congo Red and (b) UVevis spectra of Congo Red (commonly derived from lignocellulose materials) do not possess a
solutions at different solution pH values (Adapted from Zhou et al., 2011. Copyright
graphite structure. In addition, although hydrochar prepared
(2011), with permission from Elsevier).
through hydrothermal carbonization (i.e., 150e250  C controlled
temperature for 2e72 h at a specific pressure) exhibits an aromatic
lamarck seed powder, 1.31 m2/g Spirogyra species, 1.21 m2/g for carbon network, its properties are similar to those of biosorbents
waste pomace from an olive oil factory, 0.76 m2/g for soy meal (Tran et al., 2017c). Therefore, the presence of Cp-cation in-
shells, 0.48 m2/g for rubber tree leaves, and 0.48 m2/g for rice bran teractions in heavy-metal adsorption onto biosorbents and hydro-
(Farooq et al., 2010; Tran et al., 2016). Thus, the discussions and char is still controversial and needs to be further investigated.
conclusions regarding the surface area of PGP made by Ben Ali et al. Bui and Choi (2010) considered the role of cation-p bonding
(2016) are not valid, as they are inconsistent with the ASTM defi- between the protonated amino group of oxytetracycline (OTC) and
nitions and fundamental understanding of the porosities of the graphene p electrons of multiwalled carbon nanotubes
materials. (MWCNTs). Essentially, the cation-p bonding is dominated by the
electrostatic forces between the cation and the permanent quad-
rupole of the p-electron-rich aromatic structure and cation-
8.5. Maximum absorption wavelength in dye adsorption studies induced polarization (Ji et al., 2009). Like p-p interactions,
cation-p bonding would be suppressed with increasing pH. This is
A comment on the method of using the maximum absorption because at a higher pH, deprotonation of the charged amino group
wavelength (lmax) to calculate Congo red (CR) solution concentra- and protonation of the enol groups will be enhanced and so the
tions was published by Zhou et al. (2011). They pointed out that CR electron-acceptor ability of the oxytetracycline molecule is weaker
is very sensitive to pH and changes from red to blue owing to a p-p* (Ji et al., 2009). Bui and Choi (2010) opined that Oleszczuk et al.
transition of the azo group following protonation (Fig. 10a). At (2010) ignored the important roles of Ca2þ in the solution; for
lower pH values, CR is protonated and cationic CR shows two example, Ca2þ possibly forms a complex with OTC in the solution or
tautomeric forms: ammonium rich variety and azonium variety. Ca2þ can act as a cation bridge between the negatively charged OTC
Therefore, the maximum absorption wavelength used to calculate and the negative charges of the MWCNTs. Therefore, Bui and Choi
the concentration of CR in solution is strongly dependent on the (2010) suggested that Ca2þ may simultaneously bind with the
solution pH (both initial and final) (Fig. 10b); for example, lmax was negatively charged OTC and interact either with the negative
576 nm at pH 2.18e3.16, 567 nm at pH 3.86, and 496 nm at charges or the graphene p electrons of the MWCNTs via cation-p
pH  4.71. bonding. As pH is increased, both MWCNTs and OTC become more
Tien (2007) and Tien (2008) highlighted that commercial dyes negatively charged and, consequently, probably interact more
are often mixtures of active ingredients and filler materials are strongly with Ca2þ, leading to lower desorption of OTC. This hy-
often not recognized. Therefore, the possibility that experiments pothesis could be further verified by studying the desorption of OTC
may involve bi-solute adsorption is often not considered. in the presence of NaCl, instead of CaCl2.
H.N. Tran et al. / Water Research 120 (2017) 88e116 109

Fig. 11. An example of a graphene layer and proposed mechanisms of methylene adsorption green 5 onto biochar, synthesized activated carbon, and commercial activated charcoal
(Adapetd from Tran et al., 2017 b,d. Copyright (2017), with permission from Elsevier).

8.6.2. p-p interactions aromatic carbon network; therefore, p-p interactions might be
p-p interactions (also known as p-p electron donor-acceptor expected between the p-electrons of the aromatic C]C bonds in
interactions) between the p-electrons in a carbonaceous adsor- hydrochar and the p-electrons of an adsorbate. However, a recent
bent and the p-electron in the aromatic ring of an adsorbate were study (Tran et al., 2017e) demonstrated that oxygenation of a
initially proposed by Coughlin and Ezra (1968). In short, the addi- hydrochar surface through a hydrothermal process with acrylic
tion of electron-withdrawing groups (i.e., oxygen-containing and acid contributed to increasing MG5 adsorption and indicated the
nitrogen-containing functional groups) at the edges of the indi- negligible role of p-p interaction in the adsorption process. FTIR
vidual graphene layers within a carbonaceous solid causes a analysis demonstrated that the aromatic C]C peak did not signif-
considerable drop in the p-electron density. Positive holes are icantly decrease in intensity or shift toward higher/lower wave-
consequently created in the conduction p-band of the individual numbers after adsorption, which further confirms the insignificant
graphene layers, and the interactions between the p-electrons of a contribution of p-p interactions. Electrostatic attraction played a
carbonaceous adsorbent and the p-electrons of the adsorbate ar- major role in the adsorption mechanism of MG5 onto hydrochar,
omatic rings become weaker. while hydrogen bonding and n-p interactions were minor con-
Recently, Tran et al. (2017b, 2017d) investigated the mechanisms tributors. The primary adsorption mechanisms of MG5 onto
of methylene green 5 (MG5) adsorption by commercial activated hydrochar were similar to those on biosorbents, but dissimilar to
charcoal, synthesized activated carbons, and prepared biochars. biochar and activated carbon (i.e., p-p interaction and pore filling).
They proposed that the primary mechanisms in MG5 adsorption An identical result was obtained in the investigation of phenol,
were p-p interactions and pore filling, while hydrogen bonding and MG5, and acid read 1 adsorption onto a commercial glucose-
n-p interactions were minor contributors (Fig. 11). To identify the derived spherical hydrochar functionalized with triethylenetetr-
existence of p-p interactions, they used two pieces of experimental amine (Tran et al., 2017a). Therefore, it can be concluded that the
evidence. First, FTIR analysis showed that a peak corresponding to addition of oxygen- and nitrogen-containing functional groups
the skeletal vibration of aromatic C]C bonds decreased in intensity (electron-withdrawing groups) on the surface of hydrochar does
and upshifted after MG5 adsorption. Secondly, oxygenation of the not cause a considerable drop in the p-electron density, but this
surface of the carbonaceous solids (i.e., biochar and activated car- process provides abundant adsorption sites on the hydrochar
bon) through a hydrothermal process with acrylic acid resulted in a surface.
decrease in MG5 adsorption and indicated the importance of p-p Jiang et al. (2009) and Chen et al. (2009) argued over the pri-
interactions to the adsorption process. mary adsorption mechanisms of hydroxyl- and amino-substituted
As discussed in Section 8.6.1, hydrochar also possesses an aromatics on carbon nanotubes and graphite (CNTs/graphite).
110 H.N. Tran et al. / Water Research 120 (2017) 88e116

Chen et al. (2009) proposed that p-p interactions played a major (9.81 ± 0.07) of CAC was insignificantly dependent on the operation
role in the adsorption mechanism. However, Jiang et al. (2009) conditions.
remarked that the experiments of Chen et al. (2009) were insuffi- Geckeis and Rabung (2004) pointed out that photon correlation
cient to support the presence of p-p electron-donoracceptor spectroscopy is a method for colloid size determination, not for
(EDA) interactions between hydroxyl-substituted aromatics and zeta-potential analysis. Methods that have been correctly applied
CNTs/graphite. According to Jiang et al. (2009), phenolic com- for zeta-potential analysis include electrophoretic light scattering
pounds were not appropriate candidates to assess the presence of and laser Doppler velocimetry, and the units for the measured zeta
p-p EDA interactions between p-electron-donating aromatics and potentials are mV, not eV.
CNTs because of the significant effects of oxygen. Therefore, the Lee (2017) considered the potential effects of flow velocity on
oxygen effect should be considered in more detail in order to gain the concentration and distribution of triphenyltin chloride (TPT) on
in-depth insights into the adsorption mechanisms of environ- nano zinc oxide (nZnO). The author remarked that TPT was possibly
mentally relevant phenolic compounds onto carbonaceous re-suspended by the hydraulic flow that occurs during hydrant
materials. flushing in a reactor because TPT can be loosely deposited on the
nZnO surface. As reported in the literature, the flow velocity in a
8.7. Other miscellaneous errors wastewater treatment system is one of the factors affecting the
removal efficiency for a tested pollutant. The distribution of TPT on
Azizian (2007) commented that the theoretical development of nZnO might vary during the adsorption process according to the
empirical PFO and PSO kinetic adsorption models using a more time of hydrant flushing; therefore, the process of TPT adsorption
general rate theory was first reported by Azizian (2004), not by onto nZnO might be affected by the operating conditions of a
Rudzinski and Plazinski (2007). Furthermore, the idea that the PFO reactor. However, the authors of the commented paper replied that
model can only be derived theoretically at a nearly constant bulk the application of nZnO for the removal of TPT from dockyard
concentration was first proposed by Azizian (2004). In addition, the wastewater was examined using a batch adsorption technique. As a
conclusion that the PFO and PSO models are simplified forms of a result, the effects of flow velocity on the adsorption process con-
more general equation was proposed independently by Azizian ducted in the batch experiments were negligible compared to that
(2004) (who relied on the Langmuir kinetic model) and by in pilot studies and/or column experiments.
Rudzinski and Plazinski (2007) (who relied on the statistical rate Kamagate  et al. (2016) and Koma
rek et al. (2016) considered the
theory). Similarly, mistakes related to either no citation of the influence of Si species originating from the partial dissolution of
original paper or incorrect citation were pointed out by Azizian quartz on the competitive adsorption of Cd(II), Cr(VI), and Pb(II)
(2008b). onto nanomaghemite- and nanomaghemite-coated quartz. They
A series of mistakes related to incorrectly digitized data, typos, recommended that to accurately assess the interfacial reaction
inaccurate calculations, incorrect use of conditional equilibrium mechanisms occurring at the Fe-oxide/water interface in the
constants, incorrect units, incorrect references, incorrect applica- presence of Si-complex mineral assemblages or Fe-coated sand
tion of modeling procedures, and other miscellaneous errors have systems, more attention should be paid to the possible release of
been highlighted by Gustafsson and Lumsdon (2014). silicate from Si-bearing minerals and its subsequent adsorption on
The distinction between heat of adsorption, adsorption energy, reactive phases.
and activation energy in adsorption was reported elsewhere A critical paper entitled “comments on “zirconium-carbon
(Inglezakis and Zorpas, 2012). The authors concluded that it is hybrid sorbent for removal of fluoride from water: oxalic acid
important to distinguish the differential enthalpy at zero coverage mediated Zr(IV) assembly and adsorption mechanism”” was pub-
(DHdo) from the differential enthalpy or isosteric heat (DHd) and the lished by Zhao (2015). The authors of the original paper concluded
standard heat of complete coverage or integral enthalpy (DH ), that the fluoride adsorption occurs on the surface sites of ZrOx-AC
which expresses the total heat generation for the adsorption and is (commercial activated carbon modified with Zr together with
related to the thermodynamic equilibrium constant. oxalic acid) with eOH displacement and/or interacts with the
Li et al. (2010) pointed out that (1) although the cited documents positive charge of zirconium to form eCOOH groups in the oxalate
were solid and sound in their scientific aspects, the reviewers acid; however, Zhao (2015) suggested that if the fluoride concen-
seemed to fail to take full advantage of the intrinsically contained tration is high enough, it would completely desorb the oxalate acid
information; (2) there were some critical errors in the main text; from the ZrOx-AC surface. According to the results of previous
and (3) the writing style of the paper was misleading and, to some experiments, Zhao (2015) suggested that it is necessary to further
extent, lacked cautiousness. They focused on 11 comments and consider or add the exchangeable role of chloride and oxalic acid in
concluded that the original paper's conclusions were not opti- the proposed fluoride-adsorption mechanism.
mistic. However, most of these comments were rejected by the Fan et al. (2007) additionally discussed the mechanisms of
original authors (Mohan and Pittman, 2011). fluoride adsorption by calcined Mg-Al-CO3 layered double hy-
Regarding the correct citation for determining pHPZC using the droxides (CLDH). Based on the experimental data published in the
batch equilibration method, Milonji c (2009a) commented that this original paper and analysis of other literature results, they
method was originally proposed by Milonji c et al. (1975). Further- concluded that XRD measurements indicated that mixed Mg-Al
more, Milonjic (2009a) noted that using correct and updated cita- oxides, as well as partially restored LDH, was present in fluoride-
tions were very important for researchers to find relevant loaded CLDH after freeze-drying treatment. Furthermore, fluoride
information, pioneer ideas, and make progress in a particular can be effectively adsorbed onto MgO, demonstrating that fluoride
subject. Additionally, from a scientific point of view, it is always adsorption over the mixed oxide along with a memory effect ac-
necessary to give credit to the authors who first proposed a method counts for the effective removal of fluoride by CLDH.
or theoretical model. Recently, Tran et al. (2017b) applied the “drift An analytical article on the magnitude of estimated maximum
method” to determine the pHPZC of commercial activated charcoal surface-area-normalized adsorption capacities (Qmax*) for the
(CAC). The effects of various operation conditions (i.e., different adsorption of polycyclic aromatic hydrocarbons and poly-
degassing times with N2, background electrolytes, concentrations chlorobiphenyls on soot and soot-like carbon materials was pub-
of an electrolyte, solid/liquid ratios, and contact times) on the pHPZC lished by Werner and Karapanagioti (2005). The original authors
were investigated. The results demonstrated that the pHPZC (van Noort et al., 2005) used the Langmuir isotherm to extrapolate
H.N. Tran et al. / Water Research 120 (2017) 88e116 111

Qmax* at the solubility limit from solid-water distribution co- removal of different pollutants (e.g. methyl orange, safranin, Cr3 þ,
efficients (Ks) measured in the pg/L to ng/L range along with esti- and Pb2 þ). After a discussion, Müller et al. (2017) concluded that
mated Langmuir affinities for adsorption (KL) on a carbonaceous because “any experimental study stands and falls on the reliability of
surface. After comparison with estimated surface-area-normalized the investigated specimen the lack of any significance therefore holds
monolayer adsorption capacities (Qmono*) and with empirical for this study as a whole”.
sorption data from the literature, Werner and Karapanagioti (2005) Interestingly, a series of recommendations and suggestions
concluded that “the Langmuir isotherm cannot be used to extrapolate were written by the editor of Carbon journal for improving
maximum sorption capacities (Qmax*) near the aqueous-sorbate manuscript quality and avoiding mistakes (Thrower, 2007, 2008a,
saturation limit from distribution coefficients measured at extremely 2008b, 2010, 2011). The author focused on five main topics: titles
low aqueous-sorbate concentrations. Adsorption sites in carbon ma- and abstracts, introduction and references, experimental, results
terials are not uniform and other processes, such as multilayer and discussion, and language.
adsorption, condensation in capillary pores, and absorption into the
polymeric matrix, may be relevant near the solubility limit”. 9. Nonlinear-optimization technique
Müller et al. (2017) analyzed strong conflicts between the XPS
and XRD data and their interpretation by the original author in a To calculate the parameters of kinetic and isotherm models
study on the synthesis of Yb doped CuFe2O4 nanoferrite adsorbents accurately in both batch and column experiments, application of
(i.e., CuYb0.5Fe1.5O4) and the application of such adsorbents in the the nonlinear method instead of the linear method has been

Fig. 12. A simple guide to the nonlinear method of calculating the Langmuir parameters using (a) “Solver add-in” in Microsoft office Excel (RSS: residual sum of squares, TSS: total
sum of squares) and (b) Origin program.
112 H.N. Tran et al. / Water Research 120 (2017) 88e116

recommended by many researchers (Porter, 1985; Ho, 2004b; Ho ii. The use of accurate citations and correct mathematical ex-
et al., 2005; Bolster and Hornberger, 2007; Han et al., 2007; pressions in original works is also recommended.
Kumar, 2007; El-Khaiary et al., 2010; Chowdhury and Das Saha, iii. Measurements of adsorption kinetics should be started at an
2011; Lima et al., 2015; Tran et al., 2015, 2017d). initial time of less than 2 min and finish when the adsorption
Furthermore, to identify the best-fit model, calculation of the process reaches true equilibrium.
chi-squared (c2, Eq. (96)) value is recommended in addition to iv. A complete adsorption isotherm (plotting qe versus Ce)
calculating the coefficient of determination (R2, Eq. (97)) for the should be presented when investigating liquid-phase
nonlinear method. In the chi-squared test, the squares of the dif- adsorption to identify the regions in which the experi-
ferences between the experimental data and data calculated using mental data relating to adsorption equilibrium are actually
the models are divided by the corresponding data obtained and located.
then summed. If the data obtained using a model are similar to the v. Differences between the number of data points in an
experimental data, c2 is close to zero. High c2 values indicate high experiment and those used for model fitting (i.e., the pseudo-
bias between the experiment and model. Therefore, analyzing the second-order or Langmuir models) should be avoided.
data set using the chi-squared test is necessary to confirm the best- vi. For adsorption thermodynamics, the equilibrium constant
fit isotherm for a given sorption system (Ho, 2004b; Ho et al., 2005; (KC) must be dimensionless. The optimal method for calcu-
Chowdhury and Das Saha, 2011; Lima et al., 2015; Tran et al., 2015, lating thermodynamic parameters should be appropriately
2017b, 2017d, 2017e). selected according to the data from the adsorption equilib-
rium experiment. Notably, it is impossible to calculate ther-
 2
modynamic parameters from an experiment using one initial
X qe;exp  qe;cal
c2 ¼ (96) adsorbate concentration at different temperatures.
qe;cal vii. To accurately estimate the parameters of adsorption kinetic
and isotherm models, the nonlinear optimization technique
P 2 should be applied to decrease the bias between the qe values
qe;exp  qe;cal calculated from the experimental data and those estimated
2
R ¼1 (97)
P 2 from the models. The chi-squared (c2) test should be used
qe;exp  qe;mean alongside the nonlinear determination coefficient (R2) to
obtain the best-fit models for adsorption kinetics and
where qe,exp (mg/g) is the amount of adsorbate uptake at equilib- isotherms.
rium obtained from Eq. (1), qe,cal (mg/g) is the amount of adsorbate viii. An in-depth understanding of the models applied to
uptake achieved from the model using the Solver add-in, and adsorption kinetics and isotherms is necessary to avoid
qe,mean (mg/g) is the mean of the qe,exp values. misapplication of these models, as well as inaccurate dis-
Lima et al. (2015) confirmed that the use of nonlinear equilib- cussions, calculations, and conclusions.
rium and kinetic adsorption models gives values that are reliable ix. More attention should be paid to various common mis-
and statistically relevant for modeling the isotherm and kinetics of understandings about fundamental adsorption and chemis-
adsorption. Moreover, they did not suggest using linearized equi- try concepts. Understanding the unique properties of
librium and kinetic adsorption models, even in the cases of the adsorbents and target adsorbates can help avoid mistakes
linearized Langmuir Type 1 and linearized PSO Type 1 equations made when explaining adsorption mechanisms.
that provide R2 values close to unity. The linearization of equilib-
rium and kinetic adsorption models could make the parameters Although the comments and recommendations of some authors
determined from these models meaningless. are not always correct, their observations and suggestions should
Owing to the wide application of the nonlinear method in the be acknowledged because of their great contributions in trans-
study of adsorption processes, we provide here a short introduction ferring scientific knowledge.
to the application of this method using the “Solver add-in” in
Microsoft Office Excel (2013 version). First, it is necessary to load
Acknowledgements
the Solver add-in in Microsoft Excel: File / Excel options / Add-
ins (select Excel Add-ins in Manage box) / select the Solver Add-in
The first author would like to thank Chung Yuan Christian
check box. Second, all Ce and qe values obtained experimentally are
University for the Distinguished International Graduate Students
considered as input data, and the qe values from the Langmuir
(DIGS) scholarship to pursue his doctoral studies. The authors
equation are calculated based on two variables (Qomax and KL s 0).
gratefully acknowledge the editor and anonymous reviewers for
Using the Solver Add-in (Fig. 12a), the Qomax, KL, and R2 values can
their invaluable insight and helpful suggestion.
be obtained. Notably, other programs (i.e., Origin; Fig. 12b) also
provide similar results and can accurately calculate the parameters
of kinetic and isotherm models. References

Alagumuthu, G., Veeraputhiran, V., Rajan, M., 2010. Comments on “Fluoride removal
from water using activated and MnO2-coated Tamarind Fruit (Tamarindus
10. Conclusions indica) shell: batch and column studies”. J. Hazard. Mater. 183 (1e3), 956e957.
Anastopoulos, I., Kyzas, G.Z., 2016. Are the thermodynamic parameters correctly
estimated in liquid-phase adsorption phenomena? J. Mol. Liq. 218, 174e185.
In scientific publications, the correct use of technical terms and Aoyama, M., 2003. Comment on “Biosorption of chromium(VI) from aqueous so-
accurate calculations are essential. According to the viewpoints lution by cone biomass of Pinus sylvestris”. Bioresour. Technol. 89 (3), 317e318.
Arica, M.Y., 2003. In reaction to the comment by Dr. Y.-S. Ho on our publication
discussed in this review, we put forward the following key con-
“Affinity dye-ligand poly(hydroxyethylmethacrylate/chitosan composite mem-
clusions and perspectives: brane for adsorption lysozyme and kinetic properties, Biochemical Engineering
Journal 13(2003) 35e45. Biochem. Eng. J. 15 (1), 79e80.
i. Adsorption performance should be expressed as qe (mg/g), Azizian, S., 2004. Kinetic models of sorption: a theoretical analysis. J. Colloid
Interface Sci. 276 (1), 47e52.
while the use of the unit of %removal should be avoided or Azizian, S., 2006. A novel and simple method for finding the heterogeneity of ad-
used cautiously. sorbents on the basis of adsorption kinetic data. J. Colloid Interface Sci. 302 (1),
H.N. Tran et al. / Water Research 120 (2017) 88e116 113

76e81. systems. Chem. Eng. J. 156 (1), 2e10.


Azizian, S., 2007. Comment on “Kinetics of solute adsorption at solid/solution in- Freundlich, H., 1906. Über die Adsorption in Lo €sungen. Z Phys. Chem. 57, 385e471.
terfaces: a theoretical development of the empirical pseudo-first and pseudo- Fu, L., Huang, X., Wang, J., 2007. Comment on “Removal of supranol yellow 4GL by
second order kinetic rate equations, based on applying the statistical rate adsorption onto Cr-intercalated montmorillonite”. J. Hazard. Mater. 148 (3),
theory of interfacial transport”. J. Phys. Chem. B 111 (1), 318. 771e772.
Azizian, S., 2008a. Comments on “adsorption equilibrium and kinetics of reactive Fu, L., Wang, J., Lu, H., Su, Y., Ren, A., 2008. Comment on “The removal of phenolic
Black 5 and reactive red 239 in aqueous solution onto surfactant-modified compounds from aqueous solutions by organophilic bentonite”. J. Hazard.
zeolite” (Karadag, D.; Turan, M.; Akgul, E.; Tok, S.; Faki, A. J. Chem. Eng. Data Mater. 151 (2e3), 851e854.
2007, 52, 16151620). J. Chem. Eng. Data 53 (1), 322e323. Gardea-Torresdey, J.L., Tiemann, K.J., Armendariz, V., Bess-Oberto, L., Chianelli, R.R.,
Azizian, S., 2008b. Comments on “Biosorption isotherms, kinetics and thermody- Rios, J., Parsons, J.G., Gamez, G., 2000. Characterization of Cr(VI) binding and
namics” review. Sep. Purif. Technol. 63 (2), 249e250. reduction to Cr(III) by the agricultural byproducts of Avena monida (Oat)
Ben Ali, S., Jaouali, I., Souissi-Najar, S., Ouederni, A., 2016. Characterization and biomass. J. Hazard. Mater. 80 (1e3), 175e188.
adsorption capacity of raw pomegranate peel biosorbent for copper removal. Geckeis, H., Rabung, T., 2004. Comment on the paper: X. Wang, Th. Rabung, H.
J. Clean. Prod. 142, 3809e3821. Geckeis, Effect of pH and humic acid on the adsorption of cesium onto g-Al2O3.
Biggar, J.W., Cheung, M.W., 1973. Adsorption of picloram (4-amino-3,5,6- J. Radioanal. Nucl. Chem. 260 (2), 437e438.
trichloropicolinic acid) on panoche, ephrata, and palouse soils: a thermody- Ghosal, P.S., Gupta, A.K., 2015. An insight into thermodynamics of adsorptive
namic approach to the adsorption mechanism1. Soil Sci. Soc. Am. J. 37 (6). removal of fluoride by calcined Ca-Al-(NO3) layered double hydroxide. RSC Adv.
Blanchard, G., Maunaye, M., Martin, G., 1984. Removal of heavy metals from waters 5 (128), 105889e105900.
by means of natural zeolites. Water Res. 18 (12), 1501e1507. Ghosal, P.S., Gupta, A.K., 2017. Determination of thermodynamic parameters from
Bolster, C.H., Hornberger, G.M., 2007. On the use of Linearized Langmuir equation. Langmuir isotherm constant-revisited. J. Mol. Liq. 225, 137e146.
Soil Sci. Soc. Am. J. 71 (6), 1796e1806. Goldberg, S., 2009. Comments on “adsorption sequence of toxic inorganic anions on
Borisover, M.D., Graber, E.R., 1997. Comment on “Competitive sorption between a soil” by K. Saeki Bull Environ Contam Toxicol (2008) 81:508e512. Bull. En-
atrazine and other organic compounds in soils and model sorbents”. Environ. viron. Contam. Toxicol. 83 (1), 1e2.
Sci. Technol. 31 (5), 1577. Goldberg, S., 2016. A comment on “Selenite adsorption and desorption in main
Bui, T.X., Choi, H., 2010. Comment on “Adsorption and desorption of oxytetracycline Chinese soils with their characteristics and physicochemical properties. J Soils
and carbamazepine by multiwalled carbon nanotubes”. Environ. Sci. Technol. Sediments 15(5):1150e1158(2015), doi:10.1007/s11368-015-1085-7. J. Soils
44 (12), 4828. Sediments 16 (1), 324.
Canzano, S., Iovino, P., Salvestrini, S., Capasso, S., 2012. Comment on “Removal of Goswami, R., Shim, J., Deka, S., Kumari, D., Kataki, R., Kumar, M., 2016. Character-
anionic dye Congo red from aqueous solution by raw pine and acid-treated pine ization of cadmium removal from aqueous solution by biochar produced from
cone powder as adsorbent: equilibrium, thermodynamic, kinetics, mechanism Ipomoea fistulosa at different pyrolytic temperatures. Ecol. Eng. 97, 444e451.
and process design”. Water Res. 46 (13), 4314e4315. Guo, H., Zhang, S., Kou, Z., Zhai, S., Ma, W., Yang, Y., 2015. Removal of cadmium(II)
Cavas, L., 2008. Comment on Equilibrium Sorption Isotherm Studies of Cd(II), Pb(II) from aqueous solutions by chemically modified maize straw. Carbohydr. Polym.
and Zn (II) Ions Detoxification from Waste Water Using Unmodified and EDTA- 115 (0), 177e185.
modified Maize Husk. Gustafsson, J.P., Lumsdon, D.G., 2014. Comment on “Citrate adsorption can decrease
Chen, W., Duan, L., Wang, L., Zhu, D., 2009. Response to comment on “Adsorption of soluble phosphate concentration in soils: results of theoretical modelling” by
hydroxyl- and amino-substituted aromatics to carbon nanotubes”. Environ. Sci. Marek Duputel, Nicolas Devau, Michel Brossard, Benoît Jaillard, Davey L. Jones,
Technol. 43 (9), 3400e3401. Philippe Hinsinger and Fre deric Gerard (2013). Appl. Geochem. 46, 85e89.
Chien, S.H., Clayton, W.R., 1980. Application of Elovich equation to the kinetics of Haerifar, M., Azizian, S., 2013. Mixed surface reaction and diffusion-controlled ki-
phosphate release and sorption in soils. Soil Sci. Soc. Am. J. 44 (2), 265e268. netic model for aAdsorption at the solid/solution interface. J. Phys. Chem. C 117
Choong, T.S.Y., Chuah, T.G., 2005. Comment on “Separation of vitamin E from palm (16), 8310e8317.
fatty acid distillate using silica: I. Equilibrium of batch adsorption by B.S. Chu Hai, T.N., 2017. Comments on “Effect of temperature on the adsorption of methylene
et al. [Journal of Food Engineering 62(2004) 97e103]”. J. Food Eng. 67 (3), 379. blue dye onto sulfuric acidetreated orange peel”. Chem. Eng. Commun. 204 (1),
Chowdhury, S., Das Saha, P., 2011. Comparative analysis of linear and nonlinear 134e139.
methods of estimating the pseudo-second-order kinetic parameters for sorp- Halder, G., Khan, A.A., Dhawane, S., 2016. Fluoride sorption onto a steam-activated
tion of malachite green onto pretreated rice husk. Bioremed. J. 15 (4), 181e188. biochar derived from cocos nucifera shell. CLEAN Soil Air Water 44 (2),
Chu, B.S., Baharin, B.S., Man, Y.B.C., Quek, S.Y., 2005. Reply from the authors to the 124e133.
comments by Choong and Chuah in a recent Letter to the Editor on “Separation Hall, K.R., Eagleton, L.C., Acrivos, A., Vermeulen, T., 1966. Pore- and solid-diffusion
of vitamin E from palm fatty acid distillate using silica: I. Equilibrium of batch kinetics in fixed-bed adsorption under constant-pattern conditions. Ind. Eng.
adsorption by B.S. Chu et al. [Journal of Food Engineering 62(2004) 97e103]”. Chem. Fundam. 5 (2), 212e223.
J. Food Eng. 69 (1), 131e132. Hamdaoui, O., Naffrechoux, E., 2007. Modeling of adsorption isotherms of phenol
Coughlin, R.W., Ezra, F.S., 1968. Role of surface acidity in the adsorption of organic and chlorophenols onto granular activated carbon: Part II. Models with more
pollutants on the surface of carbon. Environ. Sci. Technol. 2 (4), 291e297. than two parameters. J. Hazard. Mater. 147 (1e2), 401e411.
Crittenden, J.C., Trussell, R.R., Hand, D.W., Howe, K.J., Tchobanoglous, G., 2012. Han, R., Wang, Y., Zou, W., Wang, Y., Shi, J., 2007. Comparison of linear and nonlinear
Adsorption. MWH's Water Treatment: Principles and Design, third ed. John analysis in estimating the Thomas model parameters for methylene blue
Wiley & Sons, Inc., pp. 1117e1262 adsorption onto natural zeolite in fixed-bed column. J. Hazard. Mater. 145 (1e2),
D4607-14, A., 2014. Standard Test Method for Determination of Iodine Number of 331e335.
Activated Carbon. American Society for Testing and Materials, Philadelphia, Harsha, N., Krishna, K.V.S., Renuka, N.K., Shukla, S., 2015. Facile synthesis of
U.S.A. [gamma]-Fe2O3 nanoparticles integrated H2Ti3O7 nanotubes structure as a
Dambies, L., Guimon, C., Yiacoumi, S., Guibal, E., 2001. Characterization of metal ion magnetically recyclable dye-removal catalyst. RSC Adv. 5 (38), 30354e30362.
interactions with chitosan by X-ray photoelectron spectroscopy. Colloids Surf. A Ho, Y.-S., 1995. Adsorption of Heavy Metals from Waste Streams by Peat (Ph.D.
Physicochem. Eng. Asp. 177 (2e3), 203e214. Thesis). The University of Birmingham, Birmingham, U.K.
Dawood, S., Sen, T.K., 2012. Author's Responses to the comment by Canzano et al Ho, Y.-S., 2004a. Erratum to “Removal of copper ions from aqueous solution by tree
and also corrigendum to “Removal of anionic dye Congo red from aqueous fern”: [Water Res. 37(2003) 2323e2330]. Water Res. 38 (20), 4536e4537.
solution by raw pine and acid-treated pine cone powder as adsorbent: equi- Ho, Y.-S., 2004b. Selection of optimum sorption isotherm. Carbon 42 (10),
librium, thermodynamic, kinetics, mechanism and process design” published in 2115e2116.
Water Research, Vol. 46, pp. 1933e1946, 2012. Water Res. 46 (13), 4316e4317. Ho, Y.-S., 2005. Comment on “Two-stage batch sorber design using second-order
Doke, K.M., Khan, E.M., 2013. Adsorption thermodynamics to clean up wastewater; kinetic model for the sorption of metal complex dyes onto pine sawdust” by
critical review. Rev. Environ. Sci. Bio Technol. 12 (1), 25e44. €
Ozacar, _
M. and Şengýl, I.A. Biochem. Eng. J. 23 (3), 291e292.
Dubinin, M., Radushkevich, L., 1947. Equation of the characteristic curve of activated Ho, Y.-S., 2013. Comments on “Removal of zirconium(IV) from aqueous solution by
charcoal. Chem. Zentr. 1 (1), 875. Coriolus versicolor: equilibrium and thermodynamic study”. Ecol. Eng. 58, 1e2.
Egirani, D., 2004. Comments on: removal of copper ions from aqueous solution by Ho, Y.-S., 2014a. Comments on “Adsorption of 2-mercaptobenzothiazole from
tree ferns. Water Res. 38 (20), 4535. aqueous solution by organo-bentonite” by P. Jing, M.H. Hou, P. Zhao, X.Y. Tang,
El-Khaiary, M.I., Malash, G.F., Ho, Y.-S., 2010. On the use of linearized pseudo- H.F. Wan. J. Environ. Sci. 26 (12), 2571e2572.
second-order kinetic equations for modeling adsorption systems. Desalination Ho, Y.-S., 2014b. Comment on “New Calix[4]arene appended amberlite XAD-4 resin
257 (1e3), 93e101. with versatile perchlorate removal efficiency”. J. Chem. Eng. Data 59 (6), 2131.
Fan, J., Xu, Z., Zheng, S., 2007. Comment on “Factors influencing the removal of Ho, Y.-S., 2014c. Comments on “Adsorption characteristics and behaviors of gra-
fluoride from aqueous solution by calcined MgeAleCO3 layered double hy- phene oxide for Zn(II) removal from aqueous solution”. Appl. Surf. Sci. 301, 584.
droxides”. J. Hazard. Mater. 139 (1), 175e177. Ho, Y.-S., 2014d. Comments on “Simultaneous adsorption of aniline and Cr(VI) Ion
Farooq, U., Kozinski, J.A., Khan, M.A., Athar, M., 2010. Biosorption of heavy metal by activated carbon/chitosan composite”. J. Appl. Polym. Sci. 131 (22) (n/a-n/a).
ions using wheat based biosorbents e a review of the recent literature. Bio- Ho, Y.-S., 2014e. Comments on the paper “Removal of Cu (II) by loofah fibers as a
resour. Technol. 101 (14), 5043e5053. natural and low-cost adsorbent from aqueous solutions”. J. Mol. Liq. 198, 322.
Febrianto, J., Kosasih, A.N., Sunarso, J., Ju, Y.-H., Indraswati, N., Ismadji, S., 2009. Ho, Y.-S., 2014f. The real pseudo-second-order rate equation. Ind. Crops Prod. 52, 17.
Equilibrium and kinetic studies in adsorption of heavy metals using biosorbent: Ho, Y.-S., 2014g. Using of “pseudo-second-order model” in adsorption. Environ. Sci.
a summary of recent studies. J. Hazard. Mater. 162 (2e3), 616e645. Pollut. Res. 21 (11), 7234e7235.
Foo, K.Y., Hameed, B.H., 2010. Insights into the modeling of adsorption isotherm Ho, Y.-S., 2015. Comment on “Genetic characterization, nickel tolerance,
114 H.N. Tran et al. / Water Research 120 (2017) 88e116

biosorption, kinetics, and uptake mechanism of a bacterium isolated from Kumar, K.V., Porkodi, K., 2006. Relation between some two- and three-parameter
electroplating industrial effluent”. Can. J. Microbiol. 61 (11), 881e882. isotherm models for the sorption of methylene blue onto lemon peel.
Ho, Y.-S., 2016a. Comments on the paper ‘Synthesis and application of ion- J. Hazard. Mater. 138 (3), 633e635.
imprinted resin based on modified melamine-thiourea for selective removal Kumar, K.V., Porkodi, K., 2007a. Comments on “adsorption of 4-chlorophenol from
of Hg(II)’. Polym. Int. 65 (7), 845e846. aqueous solutions by xad-4 resin: isotherm, kinetic, and thermodynamic
Ho, Y.-S., 2016b. Comments on the paper “Removal of aqueous Hg(II) and Cr(VI) analysis”. J. Hazard. Mater. 143 (1e2), 598e599.
using phytic acid doped polyaniline/cellulose acetate composite membrane”. Kumar, K.V., Porkodi, K., 2007b. Comments on “Biosorption of nickel from pro-
J. Hazard. Mater. 311, 273e274. tonated rice bran”. J. Hazard. Mater. 147 (1e2), 679.
Ho, Y.-S., 2016c. Comments on the paper “Simultaneous removal of mixed con- Kumar, K.V., Porkodi, K., 2008. Comments on “removal of Congo red from aqueous
taminants by organoclays d amoxicillin and Cu(II) from aqueous solution”. solution by anilinepropylsilica xerogel” by Pavan FA, Dias SLP, Lima EC, Ben-
Appl. Clay Sci. 126, 297e298. venutti EV. Dyes and Pigments 2008;76:64e9. Dyes Pigments 77 (2), 481e482.
Ho, Y.-S., 2016d. Comments on using of “pseudo-first-order model” [J. Taiwan Inst. Kumar, K.V., Porkodi, K., Rocha, F., 2007. Comments on “Removal of lead from
Chem. Eng. Vol. 59]. J. Taiwan Inst. Chem. Eng. 68, 136e137. aqueous solution using Syzygium cumini L.: equilibrium and kinetic studies”.
Ho, Y.-S., 2016e. Comments on using of “pseudo-first-order model” in adsorption J. Hazard. Mater. 147 (1e2), 677e678.
[Int. J. Biol. Macromol., vol. 81]. Int. J. Biol. Macromol. 88, 505e506. Lagergren, S., 1898. About the theory of so-called adsorption of soluble substances.
Ho, Y.-S., 2017. Comments on the paper “Application of Mn/MCM-41 as an adsor- K. Sven. Vetensk. Handl. 24 (4), 1e39.
bent to remove Methyl Blue from aqueous solution”. J. Colloid Interface Sci. 498, Langmuir, I., 1918. The adsorption of gases on plane surfaces of glass, mica and
461. platinum. J. Am. Chem. Soc. 40 (9), 1361e1403.
Ho, Y.-S., Chiu, W.-T., Wang, C.-C., 2005. Regression analysis for the sorption iso- Lee, D.-G., 2017. Comment on: Olushola S, Olalekan S, Folahan A, Leslie F and Ximba
therms of basic dyes on sugarcane dust. Bioresour. Technol. 96 (11), 1285e1291. BJ (2014) Application of nano zinc oxide (nZnO) for the removal of triphenyltin
Ho, Y.S., McKay, G., 1998a. A comparison of chemisorption kinetic models applied to chloride (TPT) from dockyard wastewater. Water SA (Online) 43 (1), 175.
pollutant removal on various sorbents. Process Saf. Environ. Prot. 76 (4), Li, W., Pan, G., 2007. Comments on “Ion adsorption components in liquid/solid
332e340. systems”. J. Environ. Sci. 19 (6), 768.
Ho, Y.S., McKay, G., 1998b. Sorption of dye from aqueous solution by peat. Chem. Li, Z., Imaizumi, S., Katsumi, T., Tang, X., Chen, Y., 2010. Comment on JHM 142 (2007)
Eng. J. 70 (2), 115e124. 1e53 ‘Arsenic removal from waterewastewater using adsorbentsda critical
Ho, Y.S., Wase, D.A.J., Forster, C.F., 1996. Kinetic studies of competitive heavy metal review’ by D Mohan and CU Pittman Jr. J. Hazard. Mater. 175 (1e3), 1116e1117.
adsorption by sphagnum moss peat. Environ. Technol. 17 (1), 71e77. 
Lima, E.C., Adebayo, M.A., Machado, F.M., 2015. Kinetic and Equilibrium Models of
Huang, R., Yang, B., Liu, Q., Liu, Y., 2014. Simultaneous adsorption of aniline and Adsorption. In: Carbon Nanomaterials as Adsorbents for Environmental and
Cr(VI) ion by activated carbon/chitosan composite. J. Appl. Polym. Sci. 131 (4) (n/ Biological Applications. Springer, pp. 33e69.
a-n/a). Lima, E.C., Cestari, A.R., Adebayo, M.A., 2016. Comments on the paper: a critical
Hung, H.-W., Lin, T.-F., 2006. Adsorption of MTBE from contaminated water by review of the applicability of Avrami fractional kinetic equation in adsorption-
carbonaceous resins and mordenite zeolite. J. Hazard. Mater. 135 (1e3), based water treatment studies. Desalination Water Treat. 57 (41), 19566e19571.
210e217. Lin, J.X., Wang, L., 2009. Comment on “Adsorption of Supranol Yellow 4 GL from
Inbaraj, B.S., 2006. Comment on “Adsorption of reactive dyes from a textile effluent aqueous solution by surfactant-treated aluminum/chromium-intercalated
using sawdust as the adsorbent”. Ind. Eng. Chem. Res. 45 (21), 7362. bentonite”. J. Hazard. Mater. 172 (1), 516e517.
Inglezakis, V.J., Zorpas, A.A., 2012. Heat of adsorption, adsorption energy and acti- Liu, Y., 2009. Is the free energy change of adsorption correctly calculated? J. Chem.
vation energy in adsorption and ion exchange systems. Desalination Water Eng. Data 54 (7), 1981e1985.
Treat. 39 (1e3), 149e157. Lu, X., 2008. Comment on “Thermodynamic and isotherm studies of the biosorption
Ji, L., Chen, W., Duan, L., Zhu, D., 2009. Mechanisms for strong adsorption of of Cu(II), Pb(II), and Zn(II) by leaves of saltbush (Atriplex canescens)”. J. Chem.
tetracycline to carbon nanotubes: a comparative study using activated carbon Thermodyn. 40 (4), 739e740.
and graphite as adsorbents. Environ. Sci. Technol. 43 (7), 2322e2327. Marsh, H., Reinoso, F.R., 2006. Activated Carbon. Elsevier.
Jiang, J., Pang, S.-Y., Ma, J., 2009. Comment on “Adsorption of hydroxyl- and amino- Masson, S., Gineys, M., Delpeux-Ouldriane, S., Reinert, L., Guittonneau, S., Be guin, F.,
substituted aromatics to carbon nanotubes”. Environ. Sci. Technol. 43 (9), Duclaux, L., 2016. Single, binary, and mixture adsorption of nine organic con-
3398e3399. taminants onto a microporous and a microporous/mesoporous activated carbon
Jing, P., Hou, M., Zhao, P., Tang, X., Wan, H., 2014. Reply to comments on “Adsorption cloth. Microporous Mesoporous Mater. 234, 24e34.
of 2-mercaptobenzothiazole from aqueous solution by organo-bentonite” by McKay, G., Ho, Y.S., Ng, J.C.Y., 1999. Biosorption of copper from waste waters: a
Yuhshan Ho. J. Environ. Sci. 26 (12), 2573e2574. review. Sep. Purif. Rev. 28 (1), 87e125.
Kamagate , M., Lützenkirchen, J., Huber, F., Hanna, K., 2016. Comment on McLintock, I., 1967. The Elovich equation in chemisorption kinetics. Nature 216,
“Competitive adsorption of Cd(II), Cr(VI), and Pb(II) onto nanomaghemite: a 1204e1205.
spectroscopic and modeling approach”. Environ. Sci. Technol. 50 (3), Medellin-Castillo, N.A., Padilla-Ortega, E., Regules-Martínez, M.C., Leyva-Ramos, R.,
1632e1633. Ocampo-Pe rez, R., Carranza-Alvarez, C., 2017. Single and competitive adsorption
Karadag, D., Turan, M., Akgul, E., Tok, S., Faki, A., 2007. Adsorption equilibrium and of Cd(II) and Pb(II) ions from aqueous solutions onto industrial chili seeds
kinetics of reactive black 5 and reactive red 239 in aqueous solution onto (Capsicum annuum) waste. Sustain. Environ. Res. 27 (2), 61e69.
surfactant-modified zeolite. J. Chem. Eng. Data 52 (5), 1615e1620. Milonji c, S.K., 2007. A consideration of the correct calculation of thermodynamic
Khan, A.A., Singh, R.P., 1987. Adsorption thermodynamics of carbofuran on Sn (IV) parameters of adsorption. J. Serb. Chem. Soc. 72 (12), 1363e1367.
arsenosilicate in Hþ, Naþ and Ca2þ forms. Colloids Surf. 24 (1), 33e42. Milonji c, S.K., 2009a. Comments on “Factors influencing the removal of divalent
Kom arek, M., Koretsky, C.M., Stephen, K.J., Alessi, D.S., Chrastný, V., 2016. Response cations by hydroxyapatite”. J. Hazard. Mater. 162 (2e3), 1588e1589.
to Comment on “Competitive adsorption of Cd(II), Cr(VI), and Pb(II) onto Milonji c, S.K., 2009b. Comments on “Removal of uranium (VI) from aqueous solu-
nanomaghemite: a spectroscopic and modeling approach”. Environ. Sci. Tech- tion by adsorption of hematite”, by X. Shuibo, Z. Chun, Z. Xinghuo, Y. Jing, Z.
nol. 50 (3), 1634e1635. Xiaojian, W. Jingsong. J. Environ. Radioact. 100 (10), 921e922.
Kim, Y., Bae, J., Park, H., Suh, J.-K., You, Y.-W., Choi, H., 2016. Adsorption dynamics of Milonji c, S.K., 2010. Comments on the authors' response to the comments on
methyl violet onto granulated mesoporous carbon: facile synthesis and “Factors influencing the removal of divalent cations by hydroxyapatite”, by
adsorption kinetics. Water Res. 101, 187e194. Smiciklas et al. J. Hazard. Mater. 176 (1e3), 1126e1127.
Kosmulski, M., 2009. Surface Charging and Points of Zero Charge. CRC Press. Milonji  si
c, S.K., Ruvarac, A.L., Su c, M.V., 1975. The heat of immersion of natural
Kratochvil, D., Volesky, B., 1998. Advances in the biosorption of heavy metals. magnetite in aqueous solutions. Thermochim. Acta 11 (3), 261e266.
Trends Biotechnol. 16 (7), 291e300. Mohan, D., Pittman Jr., C.U., 2011. Reply to the comments on HAZMAT 142 (2007)
Kumar, K.V., 2006a. Comments on “Adsorption of acid dye onto organobentonite”. 1e53 ‘Arsenic removal from water/wastewater using adsorbents e a critical
J. Hazard. Mater. 137 (1), 638e639. review’ by D. Mohan and C.U. Pittman Jr. made by Zhenze Li et al. [Hazmat 175
Kumar, K.V., 2006b. Comments on “Equilibrium studies for the adsorption of Acid (2010) 1116e1117]. J. Hazard. Mater. 185 (2e3), 1614e1617.
dye onto modified hectorite”. J. Hazard. Mater. 137 (2), 1252e1253. Morosanu, I., Teodosiu, C., Paduraru, C., Ibanescu, D., Tofan, L., 2016. Biosorption of
Kumar, K.V., 2006c. A note on the comments by Dr. Y.S. Ho on “Nitrate removal from lead ions from aqueous effluents by rapeseed biomass. New Biotechnol. http://
aqueous solution by adsorption onto various materials”. J. Hazard. Mater. 136 dx.doi.org/10.1016/j.nbt.2016.08.002.
(3), 995e996. Müller, G., Janoskov r, T., Cakl, J., Jir
a, K., Bakala , H., 2012. Removal of Zn(II)
ankova
Kumar, K.V., 2006d. A note on the comments by Dr. Y.S. Ho on “Remediation of soil from aqueous solutions using Lewatit S1468. Desalination Water Treat. 37
contaminated with the heavy metal (Cd2þ)”. J. Hazard. Mater. 136 (3), (1e3), 146e151.
993e994. Müller, F., Faidt, T., Grandthyll, S., Natter, H., 2017. Comment on “CuYb0.5Fe1.5O4
Kumar, K.V., 2007. Optimum sorption isotherm by linear and non-linear methods nanoferrite adsorbent structural, morphological and functionalization charac-
for malachite green onto lemon peel. Dyes Pigments 74 (3), 595e597. teristics for multiple pollutant removal by response surface methodology” by
Kumar, K.V., F avere, V.T., 2006. Reply to Comments on ‘Chitosan functionalized with M.A. Rehman et al., J. Mol. Liq. 224(2016) 1256e1265. J. Mol. Liq. 233, 52e54.
2[-bis-(pyridylmethyl) aminomethyl]4-methyl-6-formyl-phenol: equilibrium Nekouei, F., Nekouei, S., 2017. Comments on the paper “Adsorptive removal of
and kinetics of copper(II) adsorption’ by Yuh-Shan Ho: discussion on pseudo methylene blue by rhamnolipid-functionalized graphene oxide from waste-
second order kinetic expression. Polymer 47 (6), 1772e1773. water”. Water Res. 108, 462e463.
Kumar, K.V., Guha, B., 2006. Reply to the comments on “Study on biosorption of Cr Nekouei, F., Kargarzadeh, H., Nekouei, S., Tyagi, I., Agarwal, S., Kumar Gupta, V.,
(VI) by Mucor hiemalis” by Y.-S. Ho, Biochem. Eng. J. 26(2005) 82e83. Biochem. 2016. Preparation of Nickel hydroxide nanoplates modified activated carbon for
Eng. J. 30 (2), 222e223. Malachite Green removal from solutions: kinetic, thermodynamic, isotherm
H.N. Tran et al. / Water Research 120 (2017) 88e116 115

and antibacterial studies. Process Saf. Environ. Prot. 102, 85e97. Thrower, P.A., 2010. Writing a scientific paper: IV. Results and discussion. Carbon 48
Oleszczuk, P., Pan, B., Xing, B., 2010. Response to comment on “adsorption and (10), 2675e2676.
desorption of oxytetracycline and carbamazepine by multiwalled carbon Tien, C., 2007. Remarks on adsorption manuscripts received and declined: an
nanotubes”. Environ. Sci. Technol. 44 (12), 4829. editorial. Sep. Purif. Technol. 54 (3), 277e278.
Onjia, A.E., Milonjic, S.K., 2002. Influence of the background electrolyte on the point Tien, C., 2008. Remarks on adsorption manuscripts revised and declined: an
of zero charge of chromium(III)-oxide. Mater. Sci. Forum 413, 87e92. editorial. J. Hazard. Mater. 150 (1), 2e3.

Ozacar, M., 2005. Response to comment on “Two-stage batch sorber design using Tran, H.N., 2017. Comments on “Characterization and adsorption capacity of raw
second-order kinetic model for the sorption of metal complex dyes onto pine pomegranate peel biosorbent for copper removal”. J. Clean. Prod. 144, 553e558.
sawdust” by Dr. Y.-S. Ho, Biochem. Eng. J. 21(2004) 39e45. Biochem. Eng. J. 23 Tran, H.N., Huang, F.-C., Lee, C.-K., Chao, H.-P., 2017a. Activated carbon derived from
(3), 293. spherical hydrochar functionalized with Triethylenetetramine: synthesis,
Park, D., Yun, Y.-S., Kim, J.Y., Park, J.M., 2008. How to study Cr(VI) biosorption: use of characterizations, and adsorption application. Green Process. Synth. http://
fermentation waste for detoxifying Cr(VI) in aqueous solution. Chem. Eng. J. 136 dx.doi.org/10.1515/gps-2016-0178.
(2e3), 173e179. Tran, H.N., Wang, Y.-F., You, S.-J., Chao, H.-P., 2017b. Insights into the mechanism of
Park, D., Yun, Y.-S., Park, J.M., 2005. Studies on hexavalent chromium biosorption by cationic dye adsorption on activated charcoal: the importance of p-p in-
chemically-treated biomass of Ecklonia sp. Chemosphere 60 (10), 1356e1364. teractions. Process Saf. Environ. Prot. 107, 168e180.
Park, D., Yun, Y.-S., Park, J.M., 2006a. Comment on “Chromate ion adsorption by Tran, H.N., You, S.-J., Chao, H.-P., 2015. Effect of pyrolysis temperatures and times on
agricultural by-products modified with dimethyloldihydroxylethylene urea and the adsorption of cadmium onto orange peel derived biochar. Waste Manag.
choline chloride” by Wartelle and Marshall. Water Res. 40 (7), 1501e1504. Res. 34 (2), 129e138.
Park, D., Yun, Y.-S., Park, J.M., 2006b. Comment on the removal mechanism of Tran, H.N., You, S.-J., Chao, H.-P., 2016. Thermodynamic parameters of cadmium
hexavalent chromium by biomaterials or biomaterial-based activated carbons. adsorption onto orange peel calculated from various methods: a comparison
Ind. Eng. Chem. Res. 45 (7), 2405e2407. study. J. Environ. Chem. Eng. 4 (3), 2671e2682.
Park, D., Yun, Y.-S., Park, J.M., 2006c. Mechanisms of the removal of hexavalent Tran, H.N., You, S.-J., Chao, H.-P., 2017c. Activated carbons from golden shower upon
chromium by biomaterials or biomaterial-based activated carbons. J. Hazard. different chemical activation methods: synthesis and characterizations.
Mater. 137 (2), 1254e1257. Adsorpt. Sci. Technol. http://dx.doi.org/10.1177/0263617416684837.
Patrickios, C.S., Yamasaki, E.N., 1997. A correction to the calculation of the Gibbs free Tran, H.N., You, S.-J., Chao, H.-P., 2017d. Fast and efficient adsorption of methylene
energy of adsorption for biomolecules in ion-exchange systems. Biophys. Chem. green 5 on activated carbon prepared from new chemical activation method.
69 (2e3), 219e220. J. Environ. Manag. 188, 322e336.
Plazinski, W., Rudzinski, W., Plazinska, A., 2009. Theoretical models of sorption Tran, H.N., You, S.-J., Chao, H.-P., 2017e. Insight into adsorption mechanism of
kinetics including a surface reaction mechanism: a review. Adv. Colloid Inter- cationic dye onto agricultural residues-derived hyrochars: negligible role of p-p
face Sci. 152 (1e2), 2e13. interaction. Korean J. Chem. Eng. http://dx.doi.org/10.1007/s11814-017-0056-7.
Porter, W.R., 1985. Comment on: “Equilibrium adsorption of polycyclic aromatic Uchimiya, M., Lima, I.M., Thomas Klasson, K., Chang, S., Wartelle, L.H., Rodgers, J.E.,
hydrocarbons from water onto activated carbon”. Environ. Sci. Technol. 19 (9), 2010. Immobilization of heavy metal ions (CuII, CdII, NiII, and PbII) by Broiler
869e870. litter-derived biochars in water and soil. J. Agric. Food Chem. 58 (9),
Radovic, L.R., 1999. Surface chemistry of activated carbon materials: state of the art 5538e5544.
and implications for adsorption. Surfactant Sci. Ser. 78, 529e565. USEPA, 1999. Understanding Variation in Partition Coefficient, Kd, Values. United
Rahmani-Sani, A., Shan, R.-r., Yan, L.-g., Hosseini-Bandegharaei, A., 2017. Response States Office of Air and Radiation. EPA 402-R-99e004A. Environmental Pro-
to “Letter to Editor: minor correction to the thermodynamic calculation using tection Agency, Washington, USA, p. 1.
the distribution constant by Shan et al. and Rahmani-Sani, et al.” J. Hazard. Utomo, R.P., Koswojo, R., Ju, Y.-H., Ayucitra, A., Soetaredjo, F.E., Sunarso, J.,
Mater. 325 (5). Ismadji, S., 2010. Reply to the comment on “Acid green 25 removal from
Rayne, S., 2013. Comment on “Adsorption of ionizable organic contaminants on wastewater by organo-bentonite from Pacitan” by R. Koswojo, R. P. Utomo, Y.-H.
multi-walled carbon nanotubes with different oxygen contents [Li et al., J. Ju, A. Ayucitra, F. E. Soetaredjo, J. Sunarso, S. Ismadji [Applied Clay Science
Hazard. Mater. 186(2011) 407e415]”. J. Hazard. Mater. 252e253, 1. 48(2010) 81e86]. Appl. Clay Sci. 50 (1), 165e166.
Redlich, O., Peterson, D.L., 1959. A useful adsorption isotherm. J. Phys. Chem. 63 (6), van Noort, P.C.M., Jonker, M.T.O., Koelmans, A.A., 2005. Response to comment on
1024. “Modeling maximum adsorption capacities of soot and soot-like materials for
Rivera-Utrilla, J., Sanchez-Polo, M., 2011. Adsorbent-adsorbate interactions in the PAHs and PCBs”. Environ. Sci. Technol. 39 (1), 383e384.
adsorption of organic and inorganic species on ozonized activated carbons: a Vasanth Kumar, K., Porkodi, K., Rocha, F., 2007. Comments on “Equilibrium and
short review. Adsorption 17 (3), 611e620. kinetic studies for the biosorption system of copper(II) ion from aqueous so-
Robalds, A., Naja, G.M., Klavins, M., 2016. Highlighting inconsistencies regarding lution using Tectona grandis L.f. leaves powder”. J. Hazard. Mater. 146 (1e2),
metal biosorption. J. Hazard. Mater. 304, 553e556. 428e429.
Roginsky, S., Zeldovich, Y.B., 1934. The catalytic oxidation of carbon monoxide on Vasanth Kumar, K., Rattanaphani, S., 2007. Reply to ‘Comments on “An adsorption
manganese dioxide. Acta Phys. Chem. USSR 1, 554. and kinetic study of lac dyeing on silk”’ by Yuh-Shan Ho: discussion on pseudo
Rudzinski, W., Plazinski, W., 2007. Reply to “Comment on “kinetics of solute second order kinetic expression. Dyes Pigments 75 (1), 253e254.
adsorption at solid/solution Interfaces: a theoretical development of the Vasanth Kumar, K., Sivanesan, S., 2006. Equilibrium data, isotherm parameters and
empirical pseudo-first and pseudo-second order kinetic rate equations, based process design for partial and complete isotherm of methylene blue onto
on applying the statistical rate theory of interfacial transport”. J. Phys. Chem. B activated carbon. J. Hazard. Mater. 134 (1e3), 237e244.
111 (1), 319. Volesky, B., 2003. Sorption and Biosorption. BV Sorbex.
Salvestrini, S., Leone, V., Iovino, P., Canzano, S., Capasso, S., 2014. Considerations Volesky, B., 2007. Biosorption and me. Water Res. 41 (18), 4017e4029.
about the correct evaluation of sorption thermodynamic parameters from Walter, W.J., 1984. Evolution of a technology. J. Environ. Eng. 110 (5), 899e917.
equilibrium isotherms. J. Chem. Thermodyn. 68, 310e316. Wang, H., Yuan, X., Wu, Y., Huang, H., Zeng, G., Liu, Y., Wang, X., Lin, N., Qi, Y., 2014.
Senthil Kumar, P., Fernando, P.S.A., Ahmed, R.T., Srinath, R., Priyadharshini, M., Corrigendum to “Adsorption characteristics and behaviors of graphene oxide
Vignesh, A.M., Thanjiappan, A., 2014. Effect of temperature on the adsorption of for Zn(II) removal from aqueous solution” [Appl. Surf. Sci. 279(2013) 432e440].
methylene blue dye onto sulfuric acidetreated orange peel. Chem. Eng. Com- Appl. Surf. Sci. 301, 585.
mun. 201 (11), 1526e1547. Wang, L.K., 2009. Heavy Metals in the Environment. CRC Press, Boca Raton.
Sharma, P., Hussain, N., Borah, D.J., Das, M.R., 2013. Kinetics and adsorption behavior Weber, T.W., Chakravorti, R.K., 1974. Pore and solid diffusion models for fixed-bed
of the methyl blue at the graphene oxide/reduced graphene oxide nano- adsorbers. AIChE J. 20 (2), 228e238.
sheetewater interface: a comparative study. J. Chem. Eng. Data 58 (12), Weber, W.J., Morris, J.C., 1963. Kinetics of adsorption on carbon from solution.
3477e3488. J. Sanit. Eng. Div. 89 (2), 31e60.
Sharma, Y.C., Gupta, G.S., Prasad, G., Rupainwar, D.C., 1990. Use of wollastonite in the Weber, W.J., Smith, E.H., 1987. Simulation and design models for adsorption pro-
removal of Ni(II) from aqueous solutions. Water Air Soil Pollut. 49 (1), 69e79. cesses. Environ. Sci. Technol. 21 (11), 1040e1050.
Smiciklas, I., Onjia, A., Rai cevi
c, S., Jana
ckovi
c, D., 2009. Authors' response to Werner, D., Karapanagioti, H.K., 2005. Comment on “Modeling maximum adsorp-
comments on “Factors influencing the removal of divalent cations by hy- tion capacities of soot and soot-like materials for PAHs and PCBs”. Environ. Sci.
droxyapatite”. J. Hazard. Mater. 168 (1), 560e562. Technol. 39 (1), 381e382.
Somasundaran, P., 1968. Zeta potential of apatite in aqueous solutions and its Worch, E., 2012. Adsorption Technology in Water Treatment: Fundamentals, Pro-
change during equilibration. J. Colloid Interface Sci. 27 (4), 659e666. cesses, and Modeling. Walter de Gruyter.
Swiatkowski, A., Pakula, M., Biniak, S., Walczyk, M., 2004. Influence of the surface Wu, C.-H., 2009. Response to comment on “Adsorption of direct dyes from aqueous
chemistry of modified activated carbon on its electrochemical behaviour in the solutions by carbon nanotubes: determination of equilibrium, kinetics and
presence of lead(II) ions. Carbon 42 (15), 3057e3069. thermodynamics parameter” by Dr. Ho. J. Colloid Interface Sci. 333 (1), 413e414.
Thrower, P., 2011. Writing a scientific paper V e language. Carbon 49 (15), Wu, F.-C., Liu, B.-L., Wu, K.-T., Tseng, R.-L., 2010. A new linear form analysis of
4957e4960. RedlichePeterson isotherm equation for the adsorptions of dyes. Chem. Eng. J.
Thrower, P.A., 2007. Writing a scientific paper: I. Titles and abstracts. Carbon 45 (11), 162 (1), 21e27.
2143e2144. Wu, Z., Zhong, H., Yuan, X., Wang, H., Wang, L., Chen, X., Zeng, G., Wu, Y., 2017. Reply
Thrower, P.A., 2008a. Writing a scientific paper: II. Introduction and references. for comment on “Adsorptive removal of methylene blue by rhamnolipid-
Carbon 46 (2), 183e184. functionalized graphene oxide from wastewater”. Water Res. 108, 464e465.
Thrower, P.A., 2008b. Writing a scientific paper: III. Experimental. Carbon 46 (8), Zhao, Y., 2015. Comment on “ZirconiumeCarbon hybrid sorbent for removal of
1113e1114. fluoride from water: oxalic acid mediated Zr(IV) assembly and adsorption
116 H.N. Tran et al. / Water Research 120 (2017) 88e116

mechanism”. Environ. Sci. Technol. 49 (19), 11982e11983. (Part B), 735e736.


Zhou, Q., Xie, C., Gong, W., Xu, N., Zhou, W., 2011. Comments on the method of using Zhou, X., Liu, H., Hao, J., 2012. Letters to the editor. Adsorpt. Sci. Technol. 30 (7),
maximum absorption wavelength to calculate Congo red solution concentration 647e649.
published in J. Hazard. Mater. J. Hazard. Mater. 198, 381e382. Zhou, X., Zhou, X., 2014. The unit problem in the thermodynamic calculation of
Zhou, X., 2017. Minor correction to the thermodynamic calculation using the dis- adsorption using the Langmuir equation. Chem. Eng. Commun. 201 (11),
tribution constant by Shan et al. and Rahmani-Sani, et al. J. Hazard. Mater. 323 1459e1467.

You might also like