Content uploaded by Zhigang Wei
Author content
All content in this area was uploaded by Zhigang Wei on Jun 10, 2023
Content may be subject to copyright.
Content uploaded by Diego Cortés-Arriagada
Author content
All content in this area was uploaded by Diego Cortés-Arriagada on Jul 31, 2020
Content may be subject to copyright.
The effect of pH on the adsorption of arsenic(III) and arsenic(V) at the
TiO
2
anatase [101] surface
Zhigang Wei
a,
⇑
, Kai Liang
a
, Yang Wu
b
, Yandi Zou
a
, Junhui Zuo
a
, Diego Cortés Arriagada
c
,
Zhanchang Pan
a
, Guanghui Hu
a
a
School of Chemical Engineering and Light Industry, Guangdong University of Technology, Guangzhou 510006, PR China
b
College of Chemistry, Liaoning University, Shenyang 110036, PR China
c
Laboratorio de Química Teórica-Computacional, Departamento de Química-Física, Pontificia Universidad Católica de Chile, Chile
graphical abstract
article info
Article history:
Received 27 August 2015
Accepted 6 October 2015
Keywords:
Arsenate
Arsenite
TiO
2
Density functional theory
Adsorption
abstract
Octahedral TiO
2
nanocrystals (OTNs) have been prepared by a hydrothermal method with the main sur-
face of (101). Then the arsenic adsorption behavior on OTNs is investigated in a broad experimental pH
range from 1.0 to 13.5. The maximum adsorptions of arsenite (As(III)) and arsenate (As(V)) appear at pH
values 8 and 4, respectively. It is interesting to see that the minimum adsorptions of As(III) and As(V) are
both at pH 12 and then their adsorptions increase again at higher pH values such as 13.0 and 13.5. To our
best knowledge, it is quite new to report the arsenic adsorption on the controlled TiO
2
surface especially
at very high pH values. These results might be helpful to understand the adsorption mechanism. On the
other hand, periodic slab models of TiO
2
anatase (101) surface with some H
+
cations, some water mole-
cules or some OH
ions are suggested to simulate the pH effect. Using these models, the adsorptions of As
(III) and As(V) are simulated by the density functional theory (DFT) method. Qualitatively, the adsorption
abilities of arsenic species, water and OH
follow the order of AsO
3
3
>OH
> HAsO
3
2
>
H
2
AsO
3
>H
2
O>H
3
AsO
3
for As(III) and AsO
4
3
>OH
> HAsO
4
2
>H
2
AsO
4
>H
3
AsO
4
>H
2
O for As(V). It
implies that H
2
AsO
3
should be the major As(III) species at pH 8 and H
2
AsO
4
should be the major As
(V) species at pH 4, and the most negative charged ions AsO
3
3
and AsO
4
3
should correspond to the
adsorptions at the high pH values 13 and 13.5.
Ó2015 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcis.2015.10.018
0021-9797/Ó2015 Elsevier Inc. All rights reserved.
⇑
Corresponding author.
E-mail address: weizg2003@126.com (Z. Wei).
Journal of Colloid and Interface Science 462 (2016) 252–259
Contents lists available at ScienceDirect
Journal of Colloid and Interface Science
journal homepage: www.elsevier.com/locate/jcis
1. Introduction
As a cheap and nontoxic material, TiO
2
has been extensively
used for arsenic removal due to the strong sorption of As(III) and
As(V) on it [1,2]. Some investigations were carried out by different
groups and similar conclusions were obtained that the max
adsorption pH for As(III) was near 9 and the max adsorption pH
for As(V) was near 4 [3–8]. Two kinds of adsorption mechanisms
have been suggested based on the models of electrostatic factors
and surface complexes [3]. These results and conclusions are very
useful. However, some improvements are still needed both from
experiments and from the adsorption models to deeply understand
the adsorption mechanism.
It is known that the (101) surface is the most abundant surface
for anatase in nature. During previous experiments, when they
could not control the surface to be (10 1), it was difficult to exclude
or evaluate the interferences from other surfaces. In addition, due
to the neutral pH of the usually used water, most investigations
were focused on the pH range from 4 to 9. However, the minimum
adsorption at high pH value is important for TiO
2
regeneration pro-
cess. In the present paper, the (101) surface has been controlled
during the powder synthesis and a broad experimental pH range
of 1.0–13.5 is tested to show whether the surface control has a pro-
found influence on the pH effect.
The adsorption mechanism in some sense is more complex to
discuss because different distributions of As(III) and As(V) exist
in water solution as a function of pH such as in Eqs. (1)–(3) [3,9]
and Eqs. (4)–(6) [3,9,10] at 25 °C.
H
3
AsO
3
$H
þ
þH
2
AsO
3
pK
1
¼9:23 ð1Þ
H
2
AsO
3
$H
þ
þHAsO
2
3
pK
2
¼12:10 ð2Þ
HAsO
2
3
$H
þ
þAsO
3
3
pK
3
¼13:41 ð3Þ
H
3
AsO
4
$H
þ
þH
2
AsO
4
pK
1
¼2:3ð4Þ
H
2
AsO
4
$H
þ
þHAsO
2
4
pK
2
¼6:8ð5Þ
HAsO
2
4
$H
þ
þAsO
3
4
pK
3
¼11:6ð6Þ
Therefore, different kinds of arsenic species might appear on the
surface at different pH range. The main question is whether the
main species in the solution would be the main one on the surface,
and how the surface complex changes with pH. In the present
paper, a new attempt is made to explain the adsorption mechanism
based on the DFT calculations with periodic slab models. The next
section gives details about the experimental and computational
methods. The results section examines the adsorption behavior of
arsenic on TiO
2
anatase (101) surface at pH range of 1.0–13.5 and
a comparison between previous experimental results is made. Then
the DFT calculation results are shown to explain the adsorption
mechanism and the most favorable adsorption species and struc-
tures are suggested. At last, a brief summary is made and some con-
clusions are given.
2. Experimental and computational methods
2.1. Sample preparation and characterization
Octahedral TiO
2
nanocrystals (OTNs) are prepared by a
hydrothermal method [11]. 1 g of P25 is added into 10 M KOH
solution. After 1 h ultrasonic dispersion uniformity, the turbid liq-
uid is transferred into a 60 mL Teflon-lined autoclave, heated at
200 °C for 24 h, and then is naturally cooled to room temperature.
The producing white precipitates are centrifuged and washed thor-
oughly with deionized water. Subsequently, the white precipitates
are immerged into a 0.1 M NH
4
NO
3
solution for 12 h, washed with
deionized water and then dried at 60 °C for 10 h. The proper
obtained precursors with 0.05 M hexamethylenetetramine are
placed into a 60 mL Teflon-lined autoclave and heated at 200 °C
for 24 h. Finally, the OTNs are isolated from solution by centrifuga-
tion and dried at 60 °C for 10 h. The obtained OTNs are removed
organic compounds by heated at 350 °C.
OTNs are determined by a powder X-ray diffractometry (XRD,
Rigaku-Ultima III with Cu K
a
radiation). Scanning electron micro-
scopy (SEM) is recorded on a field-emission scanning electron
microscopy (FE-SEM, Hitachi S-4800). Specific surface areas (S
BET
)
and pore distribution of adsorbents are measured by Brunauere–
Emmette–Teller N
2
adsorption–desorption on a Micromeritics
ASAP 2020 Analyzer.
Stock solutions containing 100 mg L
1
of arsenic are prepared
by dissolving As
2
O
3
in redistilled water containing 0.1% (w/w)
NaOH and Na
3
AsO
4
12H
2
O in redistilled water. The arsenic concen-
trations are determined with a rapid colorimetric method [12]. All
chemicals used in the experiments are analytical grade and all the
solutions are prepared with the redistilled water. It should be
noted that all adsorption experiments are carried out for three
times.
2.2. Computational method
The calculations are performed with DFT in periodic slab mod-
els [13]. Atomic basis sets are applied numerically in terms of a
double numerical plus polarization function [14] and a global orbi-
tal cutoff of 5.2 Å is employed. The exchange–correlation interac-
tion is treated within the generalized gradient approximation
(GGA) with the functional parameterized by Perdew, Burke and
Enzerhof (PBE) [15]. The spin-polarization effects are also included
in the calculations for the open-shell systems. The system charges
are used for the charged systems [16,17]. The conductor-like
screening model (COSMO) [18,19] is applied to simulate the water
solvent environment using a dielectric constant of 78.54 appropri-
ate to water at 25 °C. All electron DFT calculations are performed
using a DMol3 package [20–22] in Materials Studio.
3. Results and discussion
3.1. Structure and morphology
Fig. 1 shows the XRD patterns of obtained precursors and OTNs.
The sharp diffraction peaks of OTNs match clearly with the
crystal structure of the anatase TiO
2
phase (tetragonal, I4
1
/amd,
Fig. 1. X-ray diffraction patterns: (a) OTNs precursors and (b) the OTNs products.
Z. Wei et al. / Journal of Colloid and Interface Science 462 (2016) 252–259 253
JCPDS 21-1272), indicating pure and perfect crystallographic
structure is observed. Fig. 2(a) shows a higher SEM image of OTNs
with octahedral bipyramid morphology. All the OTNs perform
well-defined lateral crystal face (predominantly anatase 101 facet)
with sharp edges and adjacent crystal faces are vertical as well as
with the same width (inset of Fig. 2(a)). Fig. 2(b) shows the particle
diameter statistical histogram of OTNs, and the average size
96.34 nm is calculated from measuring the diameter of stochastic
100 OTNs. The specific surface area of OTNs measured by BET is
17.6 m
2
g
1
.
3.2. Adsorption equilibrium time
The experiments are performed in air-dark systems and the sus-
pensions are prepared in 0.5-L glass beakers. Aliquots of As(III) and
As(V) stock solutions are added to make 0.2 mg L
1
of As(III) or As
(V) concentration. After adjusting the pH of solutions to the desired
values such as 4, 7 and 9 by adding hydrochloric acid and sodium
hydroxide, TiO
2
is added to attain a 0.1 g L
1
suspension. Then the
0.5 L suspensions are sealed and stirred by magnetic stirrer at
room temperature. As shown in Fig. 3, As(III) and As(V) adsorption
equilibria at various pH are established in approximately 90 and
150 min, respectively. We have kept the adsorption overnight but
the results do not change (see Fig. 1s in the appendix).
3.3. Effect of pH on arsenic adsorption
To a deeper understanding of the pH effect, a systemic study has
been carried out for all the possible pH values such as 1, 2, 3, ... 13,
13.5. The instruments and operating steps are the same as
described in Section 3.2. The removals of As(III) and As(V) by
TiO
2
adsorbent are shown in Fig. 4. It shows that the adsorption
of As(III) increases from 11% to 70% when the solution pH increases
from 1 to 8, then decreases from 70% to 8% for pH from 8 to 12,
finally increases to 11% and 20% for pH values 13 and 13.5, respec-
tively. Fig. 4 also illustrates that the maximum adsorption of As(V)
appeares at pH 4 and the maximum adsorption ratio is 68%, then
there is a decrease from pH 4 to pH 1. On the other side, there is
a large decrease from pH 4 to pH 12, and the adsorption ratio at
the minimum point is only about 5%. Then, there is a rise from
pH 12 to 13.5 with an adsorption ratio increase from 5% to 22%.
It should be noted here that the adsorption increase from pH 12
to 13.5 for both As(III) and As(V) should correspond to some
changes for the adsorption mechanism that has not yet been
clearly observed and understood.
3.4. Comparing with previous experimental results
There have been some experiments about arsenic adsorption on
TiO
2
anatase surface [3–8], as collected in Table 1. It is clear that all
Fig. 2. (a) Higher SEM image of OTNs and (b) statistical analyses of OTNs particles diameters. The inset of (a) is the top view of a single OTNs.
Fig. 3. Adsorption equilibrium of arsenic removal by OTNs adsorbent. Initial As(III)
= As(V) = 0.200 mg L
1
; OTNs content = 0.1 g L
1
; solution pH values were 4, 7 and
9.
Fig. 4. Removal of arsenic as a function of solution pH in a suspension containing
0.2 mg L
1
of arsenic and 0.1 g L
1
of OTNs.
254 Z. Wei et al. / Journal of Colloid and Interface Science 462 (2016) 252–259
the experiments were carried out at similar processes and condi-
tions except for different TiO
2
particle size, S
BET
, arsenic and TiO
2
concentrations, etc. It should be noted that the (10 1) surface,
which is the most thermodynamically stable surface, should also
be the most abundant surface in the previous papers. As shown
in Fig. 4, our results are in agreement with the previous results
during pH 4–9 for As(V) and during pH 3–11 for As(III). Especially,
the maximum and minimum adsorption pH values for arsenic can
be clearly observed in Fig. 4. The maximum adsorption pH value
for As(III) is at pH 8, which is only a little lower than the previous
results (pH 8.5 or 9) as listed in Table 1. All the previous reports
showed that the adsorption ability of As(III) became lower on both
sides of the maximum adsorption pH value. The same trend can be
found in Fig. 4. The most notable difference with respect to our
data is that the minimum value at high pH of 12. To our knowl-
edge, only Pena et al. [5] and Nabi et al. [8] reported about the
adsorption of As(III) at pH values higher than 12. For Pena et al.
[5], an adsorption minimum for As(III) at pH 12 was given for
the dark system in their challenge water, whereas in their deion-
ized water experiments the adsorption minimum disappeared.
Nabi et al. [8] reported that the adsorption decreased from pH 7
to pH 14. For a detail comparison, the arsenic concentrations of
Pena et al. [5] and Nabi et al. [8] were 5 times and 25 times higher
respectively than that in the present paper, and at the same time
the concentrations of TiO
2
were 2 and 10 times higher respectively
than that in the present paper. Furthermore, the surface area
17.6 m
2
g
1
in the present paper is much lower than that of
330 m
2
g
1
from Pena et al. [5]. Therefore, there should be strong
particle size effect and arsenic concentration effect between our
results and their results.
Compared to As(III), the effect of pH on As(V) sorption is more
pronounced. Although many researches did not obtain the maxi-
mum adsorption pH values for As(V), they all agreed that acidic
condition were helpful for the adsorption of As(V). Dutta et al.
[3] suggested that the maximum and minimum adsorption pH val-
ues for As(V) were 3 and 9, although the pH range in their paper
was only from 3 to 9. Nabi et al. [8] reported that the adsorption
decreased from pH 7 to pH 14. In the present paper, the maximum
and minimum adsorption pH values are 4 and 12 as shown in
Fig. 4. Further work about the size, concentration and temperature
effects is still desirable for this system.
3.5. The adsorption mechanism
In 2004, Dutta et al. suggested two models [3]. The first one was
the electrostatic factors based on crystal pH
pzc
. For examples, at pH
4 the crystal possesses positive charge and As(V) (H
2
AsO
4
) pos-
sesses negative charge, thus the adsorption is good; at pH 9 the
crystal possesses negative charge and As(V) (HAsO
4
2
) also
possesses negative charge, thus the adsorption is bad. Whereas this
model did not fit well with As(III). Because the maximum adsorp-
tion is near pH 9 when the crystal possesses negative charge and
the As(III) possesses approximately equimolar mixture of H
3
AsO
3
and H
2
AsO
3
in the solution. Therefore, Dutta et al. suggested the
formation of surface complexes, which might vary with pH.
The electrostatic factors between the TiO
2
surface and arsenic
species should be the basic reason for the strong adsorption of
arsenic. However, pH
PZC
(the pH for the point of zero charge) is cor-
responding to the whole crystal, while the adsorption process is a
microcosmic process mainly between arsenic and surface, i.e., the
surface Ti
5c
(the 5 coordinates Ti) with positive charge would
attract O atom of arsenic species and surface O
2c
(the 2 coordinates
O) with negative charge would attract H atom of arsenic species.
So, microcosmic models are more desirable to describe arsenic
adsorption. Basically there are two kinds of models used such as
the cluster model [23,24] and the periodic slab model [25]. Simple
sketches for the cluster model and the periodic slab model are
shown in Figs. 2s and 3s in the appendix.
In 2009, Pan et al. [26] simulated the pH effect of As(V) through
changing the number of H
+
with a cluster model similar to Fig. 2s.
Based on the results of the adsorption energy, the bidentate binu-
clear (BB) surface complex was the most thermodynamically favor-
able mode (244.5 kJ mol
1
, i.e., 58.4 Kcal mol
1
) at low pH, but MM
surface complex was the most thermodynamically favorable mode
(135.6–27.5 kJ mol
1
, i.e., 32.4–6.6 Kcal mol
1
) at intermediate and
high pH. In 2011, Pan et al. [27] simulated the pathway from the
reactant complex (0.0 Kcal mol
1
) to the monodentate mononu-
clear (MM) surface configuration (9.2 Kcal mol
1
) and then to
the BB surface configuration (16.7 Kcal mol
1
)inFig. 4 of that
paper. However, all these calculations only considered H
2
AsO
4
.
For broad pH range and from Eqs. (4)–(6), all the As(V) species such
as H
3
AsO
4
,H
2
AsO
4
, HAsO
4
2
, AsO
4
3
might appear and change to
each other both in the solution and on the surface, and the adsorp-
tion and desorption might be reversible. In the models of Pan et al.
[26,27] the total charge of the clusters were all positive such as
H
2
AsO
4
(H
2
O)
12
+ [Ti
2
(OH)
4
(H
2
O)
6
]
4+
for the low pH, H
2
AsO
4
(H
2
O)
12
+ [Ti
2
(OH)
5
(H
2
O)
5
]
3+
for intermediate pH and H
2
AsO
4
(H
2
O)
12
+ [Ti
2
(OH)
6
(H
2
O)
6
]
2+
for the high pH. In our point of view,
all these models belong to low pH range. In addition, one weakness
of the cluster model is that it cannot simulate the spatial extent of
the real surface when Fig. 2s compares with Fig. 3s. As a result, the
cluster model can only simulate the Ti
5c
on the surface, but it can-
not simulate the O
2c
on the surface, therefore it cannot simulate
the surface functional groups and surface complexes accurately.
In the recent years, we tried to simulate the arsenic-TiO
2
system
by the periodic slab model [13,28,29]. In our previous papers
[28,29], all the species of As(III) and As(V) with the BB and MM
structures were considered and calculated. The adsorption orders
Table 1
A comparison of the physicochemical properties of TiO
2
anatase samples and their adsorption properties.
This work Ref. [3] Ref. [5,6] Ref. [4] Ref. [7] Ref. [8]
Particle size (
l
m) 0.08–0.11 <0.010 0.5–2 150–600 – 0.108
S
BET
(m
2
g
1
) 17.6 334 330 250.7 98.3 –
C
As(III)
(mg L
1
) 0.200 10 1.0 0.300 1 5
C
As(V)
(mg L
1
) 0.200 10 1.0 0.300 1 5
C
TiO2
(g L
1
) 0.1 – 0.2 0.3; 0.6; 1.0 0.2 1
V
suspension
(L) 0.5 – 0.05 0.1 – 0.25
Mixed time (h) 4 2 22 2 for As(V) 72 24
5 for As(III)
pH rang 1–13.5 3–9 4–13 4–11 3–11 1–14
Water resource Redistilled water Ultra pure water Deionized water Ground water – Deionized water
Max adsorption pH for As(III) 8 – 9 8.5 9 7
Min adsorption pH for As(III) 12 – – – – 14
Max adsorption pH for As(V) 4 3 5 4 6 1–5
Min adsorption pH for As(V) 12 9 – – – 14
pH
PZC
– 6.2 5.8 – 4.8 –
Z. Wei et al. / Journal of Colloid and Interface Science 462 (2016) 252–259 255
of As(III) and As(V) were OH
> AsO
3
3
> HAsO
3
2
>H
2
AsO
3
>
H
2
O>H
3
AsO
3
and OH
> AsO
4
3
> HAsO
4
2
>H
2
AsO
4
>H
3
AsO
4
>H
2
-
O, respectively. Thus, at high pH values OH
would be more favor-
able on the surface than As(III) and As(V) species. But this
conclusion cannot agree with the experimental results in the pre-
sent paper such that there are minimum adsorption points at about
pH 12 for both As(III) and As(V).
In the present paper, some improvements have been made in
that model (Ti
32
O
64
super cell) as shown in Fig. 3s. The arsenic
complexes are adsorbed on the upper surface only, and the adsorp-
tion energy (in the supercell) is calculated as follows:
E
ads
¼E
arsenic
þE
surface
E
arsenic=surface
ð7Þ
where E
arsenic
and E
surface
are the energies of an isolated arsenic spe-
cies and the surface respectively, and E
arsenic/surface
is the total
energy of the same arsenic species adsorbed on the same surface.
It should be mentioned that other E
ads
are calculated by the same
way such as water, H
+
, and OH
.
After these changes, a series of calculations are performed in
order to ascertain whether the model could simulate the pH effect.
Fig. 5. The E
ads
values of H
+
cations or OH
ions on the surface, i.e., 8 H
+
,7H
+
,6H
+
,
5H
+
,4H
+
,3H
+
,2H
+
,H
+
,OH
,2OH
,3OH
,4OH
,5OH
,6OH
,7OH
,8OH
.
Fig. 6. The optimized As(III) geometries on TiO
2
anatase (101) surface (A–C at acidic condition; D–F at neutral condition; G–I at alkaline condition). (A) TiO
2
-H
2
AsO
3
-1; (B)
TiO
2
-H
2
AsO
3
-2; (C) TiO
2
-H
2
AsO
3
-3; (D) TiO
2
-H
2
AsO
3
-1; (E) TiO
2
-H
2
AsO
3
-2; (F) TiO
2
-H
2
AsO
3
-3; (G) TiO
2
-AsO
3
3
-1; (H) TiO
2
-AsO
3
3
-2; and (I) TiO
2
-AsO
3
3
-3.
256 Z. Wei et al. / Journal of Colloid and Interface Science 462 (2016) 252–259
In Fig. 5, the calculation E
ads
results from the lowest pH (from eight
H
+
cations on the surface to one H
+
cation on the surface) to the
highest pH (from one OH
ion to eight OH
ions) are given. The lin-
ear relation of E
ads
in Fig. 5 implies that this simple model, though
heavily approximated, may give us useful information to explain
the experimental results for the pH effect. Although it is difficult
to give a clear function, there should be some relations between
E
ads
in Fig. 5 and the zeta potential of the TiO
2
particle such as in
Fig. 1 from Dutta et al. [3]. From Fig. 5, it is clear that the E
ads
of
a single H
+
decreases from one H
+
to eight H
+
, whereas the total E
ads
of all the H
+
cations would increases. Therefore, there should be a
large positive zeta potential for eight H
+
cations, i.e., at low pH
value. On the other hand, the changing of the zeta potential should
be slower at low pH values because the decrease of E
ads
for a single
H
+
. The similar trend can be found for OH
such as the large
negative zeta potentials at high pH values and the slower changing
of the zeta potential at high pH values. From Fig. 5 the larger E
ads
of
H
+
(133.5 kcal mol
1
) than that of OH
(73.6 kcal mol
1
) agrees
with the pH
pzc
at 6.2 in Fig. 1 from Dutta et al. [3].
Using this model and Eq. (7), six water molecules, six H
+
cations
and six OH
ions are put on the surface to qualitatively simulate
the neutral pH, the acidic pH and the alkaline pH conditions,
respectively. The adsorption structures and energies are shown in
Fig. 7. The optimized As(V) geometries on TiO
2
anatase (101) surface (A–C at acidic condition; D–E at neutral condition; F–G at alkaline condition). (A) TiO
2
-H
2
AsO
4
-1; (B)
TiO
2
-H
2
AsO
4
-2; (C) TiO
2
-H
2
AsO
4
-3; (D) TiO
2
-HAsO
4
2
-1; (E) TiO
2
-HAsO
4
2
-2; (F) TiO
2
-AsO
4
3
-1; and (G) TiO
2
-AsO
4
3
-2.
Table 2
The optimized geometries (Å) and E
ads
of As(III) adsorption configurations on TiO
2
anatase (101) surface at acidic, neutral and alkaline pH conditions.
Species As-O1 As-O2 As-O3 As-Ti1 As-Ti2 E
ads
TiO
2
–H
2
AsO
3
-1
a
1.776 1.996 1.856 3.507 3.814 80.2
TiO
2
–H
2
AsO
3
-2
a
1.784 2.048 1.821 3.395 3.734 76.1
TiO
2
–H
2
AsO
3
-3
a
1.804 1.803 1.952 3.671 – 69.9
TiO
2
–H
2
AsO
3
-1
b
1.810 1.973 1.772 3.528 3.888 31.1
TiO
2
–H
2
AsO
3
-2
b
1.831 1.996 1.766 3.410 3.778 37.8
TiO
2
–H
2
AsO
3
-3
b
1.846 1.849 1.788 3.623 – 41.2
TiO
2
–AsO
3
3
-1
c
1.751 1.868 1.865 3.322 3.319 90.9
TiO
2
–AsO
3
3
-2
c
1.748 1.859 1.866 3.504 3.507 90.7
TiO
2
–AsO
3
3
-3
c
1.781 1.762 1.940 3.333 – 82.2
a
At acidic pH condition.
b
At neutral pH condition.
c
At alkaline pH condition.
Z. Wei et al. / Journal of Colloid and Interface Science 462 (2016) 252–259 257
Figs. 6 and 7 and Tables 2 and 3 (A more detail report about the cal-
culation results are in the appendix such as Figs. 4s–9s and Tables
1s–6s). In addition, the adsorptions of other relative species are
also simulated such as water (24.6, 12.3 and 15.4 kcal mol
1
,
respectively for the acidic, the neutral and the alkaline conditions),
OH
ion (66.1 and 49.3 kcal mol
1
, respectively for the neutral and
the alkaline conditions), etc.
It is clear that the adsorption energies can fit with some results
from the cluster model, such as in Table 3 the E
ads
69.8 and
68.9 kcal mol
1
of TiO
2
–HAsO
4
2
-1
a
and TiO
2
–HAsO
4
2
-2
a
qualita-
tively agree with 58.4 Kcal mol
1
of the As(V) BB complex at low
pH [26], and the energy difference from 69.8 and 68.9 kcal mol
1
to 47.1 kcal mol
1
possesses the same trend from 16.7 to
9.2 Kcal mol
1
[27] for the BB and MM complexes. At the same
time, our calculation results qualitatively agree with the
experimental results from Jing et al. [6,30], Jegadeesan et al. [7]
and Pan et al. [31] (see Table 7s in the appendix for detail). It
should be noted that the calculation method used in the present
paper such as the GGA exchange–correlation function possesses a
trend of a little over estimate the bond distance which has been
discussed in the paper of Zhang et al. [32].
Now we can explain the adsorption mechanisms. Based on the
adsorption energies, the adsorption orders of As(III) and As(V)
are AsO
3
3
>OH
> HAsO
3
2
>H
2
AsO
3
>H
2
O>H
3
AsO
3
and AsO
4
3
>
OH
>HAsO
4
2
>H
2
AsO
4
>H
3
AsO
4
>H
2
O, respectively. Under acidic
and neutral conditions, Eqs. (1)–(3) shows that H
3
AsO
3
is the
dominant species in aqueous solution whereas the adsorption
ability of H
3
AsO
3
is weaker than that of water, therefore the
adsorption of As(III) is low at low pH range; the adsorption of As
(III) increased from acidic to neutral and weak alkaline conditions,
because H
2
AsO
3
should appear on the surface, then the maximum
adsorption appear at pH 8–9; H
2
AsO
3
and HAsO
3
2
cannot compet-
itive with OH
ion at high pH conditions, and only AsO
3
3
could
compete with OH
ion, therefore there is a minimum adsorption
point at about pH 12 for As(III). Under acidic condition, Eqs. (4)
and (5) show that H
2
AsO
4
is the dominant As(V) species and its
adsorption ability is stronger than that of water, therefore H
2
AsO
4
-
should be the predominant species on the surface. The MM sur-
face complex TiO
2
–H
2
AsO
4
-3
a
can enhance the concentration of As
(V) on the surface because it occupies only one surface Ti
5c
site. At
neutral and weak alkaline conditions, HAsO
4
2
should present on
the surface. It is interesting that even if many attempts have been
made, we could not get a stable MM structure for HAsO
4
2
. One rea-
son is that there would be only one H-bond between HAsO
4
2
and
surface O
2c
, while two H-bonds can make the MM structure stable.
The other reason is that the E
ads
values of TiO
2
–HAsO
4
2
-1
b
and
TiO
2
–HAsO
4
2
-2
b
are much higher than that of two waters. So
HAsO
4
2
should change to the BB types, even if its initial adsorption
structure is a MM type. As a result, more BB type HAsO
4
2
would
make the adsorption density of As(V) lower. HAsO
4
2
cannot com-
petitive with OH
ion at high pH conditions, and only AsO
4
3
can
compete with OH
ion, therefore there is a minimum adsorption
point at about pH 12 for As(V). It is also significant to point out
here that based on our calculation results there should be some
negative charged arsenic species adsorped on the TiO
2
surface to
make the total charge of the surface more negative. This conclusion
agrees with the experimental results in Fig. 1 from Pena et al. [6]
that the adsorption of As(V) and As(III) decreased the zeta potential
of the TiO
2
particle, suggesting the formation of negatively charged
inner-sphere surface complexes for both arsenic species.
4. Conclusion
To a deeper understanding of the pH effect for As(III) and As(V)
adsorption behaviors on the TiO
2
anatase (10 1) surface, a broad pH
range has been explored such as from pH 1 to pH 13.5. It is shown
that the maximum adsorption of As(III) presents at pH 8 and its
minimum adsorption presents at pH 12. The maximum and mini-
mum adsorption pH values for As(V) are 4 and 12, respectively. The
whole pH effect is crucial for arsenic adsorption and TiO
2
regener-
ation to improve arsenic remediation technology. Using DFT
method, all the As(III) and As(V) solution species such as H
3
AsO
3
,
H
2
AsO
3
, HAsO
3
2
, AsO
3
3
, and H
3
AsO
4
,H
2
AsO
4
, HAsO
4
2
, AsO
4
3
as
well as water and OH
ion are put onto the surface with various
different possible attitudes to obtain their adsorption geometries
and adsorption energies. Based on the calculated adsorption ener-
gies, H
2
AsO
3
should be the predominant As(III) species on the sur-
face from acidic to weak basic pH conditions, AsO
3
3
would be the
predominant species at pH higher than 12. For As(V), H
2
AsO
4
would present at acidic pH conditions, HAsO
4
2
would present at
neutral and weak basic pH conditions, AsO
4
3
would present at high
pH conditions.
Acknowledgement
We thank GDUT high performance center for calculation sup-
port, and this work is supported by the National Natural Science
Foundation of China (20803014, 21173022, 21373104) and the
211 funding program of Guangdong Province.
Appendix A. Supplementary material
The long time adsorption curve at pH 1 and pH 13.5; Simple
sketches for the cluster model and the periodic slab model; The
detail calculated results of As(III) and As(V) on TiO
2
anatase
(101) surfaces; A comparison of the arsenic structures on the
TiO
2
surfaces by EXAFS method. This material is available free of
charge via the Internet at http://www.sciencedirect.com/.
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jcis.2015.10.018.
Table 3
The optimized geometries (Å) and E
ads
(kcal mol
1
) of As(V) adsorption configurations on TiO
2
anatase (101) surface at acidic, neutral and alkaline pH conditions.
Species As-O1 As-O2 As-O3 As-O4 As-Ti1 As-Ti2 E
ads
TiO
2
–H
2
AsO
4
-1
a
1.749 1.718 1.747 1.737 3.364 3.396 69.8
TiO
2
–H
2
AsO
4
-2
a
1.712 1.724 1.776 1.740 3.376 3.377 68.9
TiO
2
–H
2
AsO
4
-3
a
1.839 1.769 1.768 1.641 3.599 – 47.1
TiO
2
–HAsO
4
2
-1
b
1.755 1.757 1.796 1.672 3.454 3.450 60.6
TiO
2
–HAsO
4
2
-2
b
1.750 1.754 1.673 1.817 3.368 3.309 60.0
TiO
2
–AsO
4
3
-1
c
1.774 1.774 1.719 1.711 3.437 3.434 65.3
TiO
2
–AsO
4
3
-2
c
1.789 1.730 1.729 1.714 3.464 – 58.5
a
At acidic pH condition.
b
At neutral pH condition.
c
At alkaline pH condition.
258 Z. Wei et al. / Journal of Colloid and Interface Science 462 (2016) 252–259
References
[1] X. Meng, M. Dadachov, G.P. Korfiatis, C. Christodoulatos, Methods of preparing
a surface-activated titanium oxide product and of using same in water
treatment processes, U.S. patent, 2005,6,919,029.
[2] X. Guan, J. Du, X. Meng, Y. Sun, B. Sun, Q. Hu, Application of titanium dioxide in
arsenic removal from water: a review, J. Hazard. Mater. 215–216 (2012) 1–16.
[3] P.K. Dutta, A.K. Ray, V.K. Sharma, F.J. Millero, Adsorption of arsenate and
arsenite on titanium dioxide suspensions, J. Colloid Interface Sci. 278 (2004)
270–275.
[4] S. Bang, M. Patel, L. Lippincott, X. Meng, Removal of arsenic from groundwater
by granular titanium dioxide adsorbent, Chemosphere 60 (2005) 389–397.
[5] M.E. Pena, G.P. Korfiatis, M. Patel, L. Lippincott, X. Meng, Adsorption of As(V)
and As(III) by nanocrystalline titanium dioxide, Water Res. 39 (2005) 2327–
2337.
[6] M. Pena, X. Meng, G.P. Korfiatis, C. Jing, Adsorption mechanism of arsenic on
nanocrystalline titanium dioxide, Environ. Sci. Technol. 40 (2006) 1257–1262.
[7] G. Jegadeesan, S.R. Al-Abed, V. Sundaram, H. Choi, K.G. Scheckel, D.D.
Dionysiou, Arsenic sorption on TiO
2
nanoparticles: size and crystallinity
effects, Water Res. 44 (2010) 965–973.
[8] D. Nabi, I. Aslam, I.A. Qazi, Evaluation of the adsorption potential of titanium
dioxide nanoparticles for arsenic removal, J. Environ. Sci. 21 (2009) 402–408.
[9] V.K. Sharma, M. Sohn, Aquatic arsenic: toxicity, speciation, transformations,
and remediation, Environ. Int. 35 (2009) 743–759.
[10] S. Goldberg, C.T. Johnston, Mechanisms of arsenic adsorption on amorphous
oxides evaluated using macroscopic measurements, vibrational spectroscopy,
and surface complexation modeling, J. Colloid Interface Sci. 234 (2001) 204–
216.
[11] J. Li, Y. Yu, Q. Chen, J. Li, D. Xu, Controllable synthesis of TiO
2
single crystals
with tunable shapes using ammonium-exchanged titanate nanowires as
precursors, Cryst. Growth Des. 10 (2010) 2111–2115.
[12] R.K. Dhar, Y. Zheng, J. Rubenstone, A. van Geen, A rapid colorimetric method
for measuring arsenic concentrations in groundwater, Anal. Chim. Acta 526
(2004) 203–209.
[13] Z. Wei, S. Zhang, Z. Pan, Y. Liu, Theoretical studies of arsenite adsorption and its
oxidation mechanism on a perfect TiO
2
anatase (101) surface, Appl. Surf. Sci.
258 (2011) 1192–1198.
[14] N.A. Benedek, I.K. Snook, K. Latham, I. Yarovsky, Application of numerical basis
sets to hydrogen bonded systems: a density functional theory study, J. Chem.
Phys. 122 (2005) 144102.
[15] J.P. Perdew, Y. Wang, Accurate and simple analytic representation of the
electron-gas correlation energy, Phys. Rev. B 45 (1992) 13244–13249.
[16] V. Milman, M.C. Payne, V. Heine, R.J. Needs, J.S. Lin, M.H. Lee, Free energy and
entropy of diffusion by ab initio molecular dynamics: alkali ions in silicon,
Phys. Rev. Lett. 70 (1993) 2928–2931.
[17] M. Hakala, M.J. Puska, R.M. Nieminen, First-principles calculations of
interstitial boron in silicon, Phys. Rev. B 61 (2000) 8155–8161.
[18] B. Delley, The conductor-like screening model for polymers and surfaces, Mol.
Simul. 32 (2006) 117–123.
[19] A. Klamt, G. Schüürmann, COSMO: a new approach to dielectric screening in
solvents with explicit expressions for the screening energy and its gradient, J.
Chem. Soc., Perkin Trans. 2 (1993) 799–805.
[20] B. Delley, An all-electron numerical method for solving the local density
functional for polyatomic molecules, J. Chem. Phys. 92 (1990) 508–517.
[21] B. Delley, Fast calculation of electrostatics in crystals and large molecules, J.
Phys. Chem. 100 (1996) 6107–6110.
[22] B. Delley, From molecules to solids with the DMol3 approach, J. Chem. Phys.
113 (2000) 7756–7764.
[23] C.I. Sainz-Diaz, V. Timon, V. Botella, A. Hernandez-Laguna, Isomorphous
substitution effect on the vibration frequencies of hydroxyl groups in
molecular cluster models of the clay octahedral sheet, Am. Miner. 85 (2000)
1038–1045.
[24] A.C.Q. Ladeira, V.S.T. Ciminelli, H.A. Duarte, M.C.M. Alves, A.Y. Ramos,
Mechanism of anion retention from EXAFS and density functional
calculations: arsenic(V) adsorbed on gibbsite, Geochim. Cosmochim. Acta 65
(2001) 1211–1217.
[25] A. Vittadini, A. Selloni, F.P. Rotzinger, M. Gratzel, Structure and energetics of
water adsorbed at TiO
2
anatase (10 1) and (0 0 1) surfaces, Phys. Rev. Lett. 81
(1998) 2954–2957.
[26] G. He, M. Zhang, G. Pan, Influence of pH on initial concentration effect of
arsenate adsorption on TiO
2
surfaces: thermodynamic, DFT, and EXAFS
interpretations, J. Phys. Chem. C 113 (2009) 21679–21686.
[27] G. He, G. Pan, M. Zhang, Studies on the reaction pathway of arsenate
adsorption at water-TiO
2
interfaces using density functional theory, J. Colloid
Interface Sci. 364 (2011) 476–481.
[28] L. Yu, Y. Liu, Z.G. Wei, G.Q. Diao, M. Sun, Q. Yu, Density functional theory
calculations of arsenic(III) structures on perfect TiO
2
anatase (101) surface,
Adv. Mater. Res. 233–235 (2011) 491–494.
[29] Z.G. Wei, Y.D. Zou, H.X. Zeng, X.C. Zhong, Z.J. Cheng, S.G. Xie, Density functional
theory calculations of arsenic(V) structures on perfect TiO
2
anatase (10 1)
surface, Adv. Mater. Res. 233–235 (2011) 495–498.
[30] C. Jing, S. Liu, M. Patel, X. Meng, Arsenic leachability in water treatment
adsorbents, Environ. Sci. Technol. 39 (2005) 5481–5487.
[31] G. He, G. Pan, M. Zhang, XANES analysis of spectral properties and structures of
arsenate adsorption on TiO(2) surfaces, J. Synchrotron Radiat. 19 (2012) 394–
399.
[32] N. Zhang, P. Blowers, J. Farrell, Evaluation of density functional theory methods
for studying chemisorption of arsenite on ferric hydroxides, Environ. Sci.
Technol. 39 (2005) 4816–4822.
Z. Wei et al. / Journal of Colloid and Interface Science 462 (2016) 252–259 259
