Enhanced sorption of perfluorooctane sulfonate (PFOS) on carbon
Chemosphere 93 (2013) 1593–1599
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/chemosphere
Enhanced sorption of perﬂuorooctane sulfonate (PFOS) on carbon
nanotube-ﬁlled electrospun nanoﬁbrous membranes
Yunrong Dai, Junfeng Niu ⇑, Lifeng Yin, Jiangjie Xu, Kang Sun
State Key Laboratory of Water Environment Simulation, School of Environment, Beijing Normal University, Beijing 100875, PR China
h i g h l i g h t s
g r a p h i c a l a b s t r a c t
Multiwalled carbon nanotubes are
immobilized in electrospun
Addition of MWCNTs increases
speciﬁc surface area and tensile
strength of ENFMs.
MWCNTs-ENFMs show faster
sorption rate and higher sorption
capacity for PFOS.
Solution pH affects surface property
and sorption efﬁciency of MWCNTsENFMs.
Sorption mechanisms mainly include
hydrophobic and electrostatic
a r t i c l e
i n f o
Received 26 March 2013
Received in revised form 26 July 2013
Accepted 5 August 2013
Available online 31 August 2013
Electrospun nanoﬁbrous membranes
a b s t r a c t
Multi-walled carbon nanotube-ﬁlled electrospun nanoﬁbrous membranes (MWCNT-ENFMs) were prepared by electrospinning. The addition of MWCNTs (0.5 wt.% vs. ENFMs) doubled the speciﬁc surface area
and tensile strength of the ENFMs. The MWCNT-ENFMs were used to adsorb perﬂuorooctane sulfonate
(PFOS) in aqueous solutions. The sorption kinetics results showed that the sorption rate of PFOS onto
the MWCNT-ENFMs was much higher than the sorption rate of PFOS onto the pure ENFMs control,
and the pseudo-second-order model (PSOM) described the sorption kinetics well. The sorption isotherms
indicated that the sorption capacity of the MWCNT-ENFMs for PFOS (16.29 ± 0.26 lmol g1) increased
approximately 18 times, compared with the pure ENFMs (0.92 ± 0.06 lmol g1). Moreover, the solution
pH signiﬁcantly affected the sorption efﬁciency and sorption mechanism. The MWCNT-ENFMs were negatively charged from pH 2.0–10.0, but the electrostatic repulsion between the MWCNT-ENFMs and PFOS
was overcome by the hydrophobic interactions between PFOS and the MWCNTs or nanoﬁbers. The strong
hydrophobic interactions between PFOS and the MWCNTs played a dominant role in the sorption process.
For the pure ENFMs, the electrostatic repulsion was conquered by the hydrophobic interactions between
PFOS and the nanoﬁbers at pH > 3.1. In addition to the hydrophobic interactions, an electrostatic attraction between PFOS and the pure ENFMs was involved in the sorption process at pH < 3.1.
Ó 2013 Elsevier Ltd. All rights reserved.
Perﬂuorooctane sulfonate (PFOS) is a type of fully ﬂuorinated
organic anion that possesses high-energy carbon–ﬂuorine (C–F)
⇑ Corresponding author. Tel./fax: +86 10 5880 7612.
E-mail address: [email protected] (J.F. Niu).
0045-6535/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
bonds. It has been used extensively in surfactants, refrigerants,
adhesives, ﬁre retardants, lubricants, and medicines (Jin et al.,
2009; Wang et al., 2009). Due to its persistence, bioaccumulation,
toxic effects, and long-range transportation throughout the environment, PFOS was categorized as one of the new persistent organic pollutants (POPs) during the Stockholm Convention in May 2009
(Hagenaars et al., 2008; Wang et al., 2009; Kunacheva et al., 2012).
Y. Dai et al. / Chemosphere 93 (2013) 1593–1599
Compared with other POPs, PFOS is highly water soluble and easy
to transport in aquatic environments. To date, PFOS has been
widely detected in municipal wastewater, surface water, groundwater, and even tap water (Lien et al., 2008; Tsuda et al., 2010;
Hu et al., 2011; Thompson et al., 2011). Meanwhile, the hydrophobic chain and hydrophilic functional groups of PFOS may provide it
opportunities to adsorb onto many solid environmental matrices.
Eventually, PFOS will become widely present in the environment,
which represents a considerable ecosystem and human health hazard (Huang et al., 2010; Pan and You, 2010). Therefore, investigations into the effective removal of PFOS are urgently needed.
Due to the complete substitution of ﬂuorine (C–F bond) for
hydrogen (C–H bond), PFOS is recalcitrant towards oxidation, and
it is also difﬁcult to eliminate PFOS using other conventional chemical and microbial techniques. Several alternative treatment techniques have been proposed for the removal of PFOS from aquatic
environments, including a mechanochemical treatment (Shintani
et al., 2008), nanoﬁltration (Vecitis et al., 2009), ultraviolet irradiation (Yamada et al., 2008), sorption (Senevirathna et al., 2010a;
Wang et al., 2012), and sonochemical degradation (Cheng et al.,
2008). Among these methods, sorption treatments are widely applied because sorption provides an effective and economical approach to remove PFOS from water. Various types of sorbents,
including activated carbons (Yu et al., 2009), ion- or non-ion-exchange polymers (Senevirathna et al., 2010b), carbon nanotubes
(CNTs) (Deng et al., 2012) and alumina (Wang and Shih, 2011),
have been reported to be effective for PFOS removal. However,
these sorbents that generally exist in the form of powders or particles are difﬁcult to recycle from water. Although the sorption efﬁciency of CNTs for PFOS is excellent, their nanotoxicity may cause
health and environmental risks once they are released into the
water environment (Yang et al., 2006). Thus, the practical applications of CNTs in water treatment are restricted.
Electrospun nanoﬁbrous membranes (ENFMs) are composed of
non-woven ﬁbers with diameters ranging from several hundreds to
tens of nanometers. They are fabricated using a versatile and lowcost electrospinning technique (Li and Xia, 2004; Greiner and
Wendorff, 2007). Due to the small diameters of the electrospun ﬁbers, ENFMs exhibit many amazing properties, such as high surface-to-volume ratios, porous structures, and superior
mechanical properties (Burger et al., 2006). Beneﬁtting from these
advantages, ENFMs have attracted considerable attention and have
been extensively used in tissue engineering, biotechnology, and
environmental remediation (Burger et al., 2006; Thavasi et al.,
2008). In our previous studies, several types of ENFMs constructed
from different polymers were used to remove hydrophobic organic
pollutants, including polycyclic aromatic hydrocarbons (PAHs) and
pentachlorophenol, from water, and the ENFMs exhibited excellent
sorption properties (Dai et al., 2011; Niu et al., 2013). Furthermore,
compared with other powdered or granular sorbents, the membrane sorbents could be used directly and then easily separated
from the reaction solution after sorption. Therefore, ENFMs were
viewed as a type of potential sorbent and were used to remove
PFOS from water in our experiments. However, the ENFMs showed
poor sorption efﬁciency due to the hydrophobic and oleophobic
properties of PFOS.
Considering the extraordinary sorption properties of CNTs and
the highly tunable structures of ENFMs, it would be beneﬁcial if
these advantages were combined to achieve enhanced sorption
efﬁciencies of ENFMs for PFOS that would also inhibit the release
of the CNTs into water. Thus, in this study, we prepared multiwalled carbon nanotube (MWCNT)-ﬁlled nanoﬁbrous membranes
by electrospinning the mixture of polymer and MWCNT solution.
Meanwhile, the morphology, structure and physicochemical properties of the MWCNT-ﬁlled electrospun nanoﬁbrous membranes
(MWCNT-ENFMs) were investigated. The sorption kinetics and iso-
therms of PFOS onto the MWCNT-ENFMs were further assessed,
and the inﬂuences of solution pH on the sorption behaviors of PFOS
were also explored. Finally, the sorption mechanisms and the possible interactions between the sorbents and sorbates were
2. Materials and methods
Poly(D,L-lactide) (PDLLA, MW 100 000) was purchased from Jinan Daigang biomaterials Co., Ltd. (Shandong, China). Methylene
dichloride and methanol (HPLC, 99.9%) were bought from J.T.Baker
(USA). PFOS (>98.0%) was obtained commercially from Sigma–Aldrich (USA) and was used as received without further puriﬁcation.
The MWCNTs were provided by Chengdu Organic Chemistry Co.,
Ltd., Chinese Academy of Sciences (Sichuan, China). According to
the manufacturer, the MWCNTs were synthesized using the chemical vapor deposition (CVD) method and had a purity of >95%, outer
diameter of 30–50 nm. All other reagents and solvents were analytical grade and used without further puriﬁcation. All solutions
were prepared using high-purity water that was obtained from a
Milli-Q Plus/Millipore puriﬁcation system (USA).
Electrospinning was performed on a self-made electrospinning
apparatus in our laboratory. The procedure for the MWCNT-ENFMs
was as follows: ﬁrst, 3 g of PDLLA was dissolved in 20 g methylene
dichloride with gentle stirring for 3 h at ambient temperature to
form a homogeneous solution. Then, 0.5 mL of the MWCNT solution (treated by using sulfuric and nitric acid solution (Bang
et al., 2012) and dispersed in methylene dichloride, 10 mg mL1)
was added to the PDLLA solution and mixed fully via vortexing
and ultrasonication. The mixture was then loaded into a 10 mL
spinning solution cartridge, with 12 needles (0.8 mm inner diameter) attached. A stainless steel needle was inserted into the cartridge and connected to a high-voltage power supply (HB-Z5032AC, Tianjin Hengbo High-Voltage Power Supply Plant, China). A
syringe pump (RWD Life Science Co., Ltd, China) was set to inject
the solution at a ﬂow rate of 1.5 mL min1. Electrospinning was
conducted at a voltage of 11 kV, and a grounded iron plate covered
with aluminum foil was placed at a distance of 15 cm from the
needle tip as a ﬁber collector. It usually took 0.5–1 h to obtain sufﬁciently thick and integrated MWCNT-ENFMs. The samples were
ﬁrst washed three times with high-purity water to elute the
MWCNTs onto the surfaces of the nanoﬁbrous membranes and
then dried in a vacuum for 10 h before use. Pure ENFMs prepared
with only the PDLLA polymer solution were used as the control.
The electrospinning voltage for the pure ENFMs was enhanced to
20 kV to obtain nanoﬁbers that were similar to the MWCNTENFMs. All experiments were conducted at room temperature
(25 ± 1 °C) with a relative humidity of approximately 20 ± 2%.
The morphologies of the MWCNT-ENFMs and pure ENFMs were
observed with a ﬁeld emission scanning electron microscope (FESEM S-4800; Hitachi, Japan). The speciﬁc surface area was determined using a fully automatic speciﬁc surface area analyzer
(ASAP 2020; Micromeritics, USA). To measure the hydrophilic–
hydrophobic properties of the polymer, PDLLA was dissolved in
methylene dichloride and prepared into a cast ﬁlm, and the contact
angle was tested with a contact angle measurement system
(OCA20; Dataphysics, Germany). The tensile mechanical properties
Y. Dai et al. / Chemosphere 93 (2013) 1593–1599
of the MWCNT-ENFMs and the pure ENFMs were measured with a
material testing machine (Instron 3366; USA). The zeta potentials
of the MWCNT-ENFMs and the pure ENFMs at different pH solutions (2.0–10.0) were measured using a solid surface zeta potential
analyzer (SurPASS; Anton Paar, Austria).
where qt is the sorption amount of PFOS at time t (lmol g1), C0 is
the PFOS concentration (lmol L1), Ct is the initial PFOS concentration at time t (lmol L1), V is the volume of the reaction solution (L),
and m is the mass of sorbent (g).
2.5. Sorption data ﬁtting
2.4. Sorption experiments
Batch experiments were conducted at 25 ± 1 °C in an incubator
shaker. Firstly, PFOS was dissolved in methanol to prepare PFOS
stock solution (100 g L1), and then the PFOS aqueous solution
was obtained by diluting stock solution with high-purity water.
The methanol volume fraction in each PFOS aqueous solution
was controlled to less than 0.001 to avoid any cosolvent effects.
During all sorption experiments, three pieces of membranes
(1 cm 1 cm, total wt. 50–55 mg) were added into polypropylene
conical ﬂasks containing 50 mL of PFOS aqueous solution, and the
reaction mixture was incubated with stirring (150 rpm) for 40 h. In
our experiments, the sorption kinetics were determined from solutions containing 100 lg L1 of PFOS (pH = 6.0, adjusted by 5 M HCl
and NaOH solutions). The sorption isotherm experiments were
conducted with an initial PFOS concentration that ranged from 1
to 100 000 lg L1 (pH = 6.0). At speciﬁc time points, a volume of
0.5 mL sample was taken from the reaction system for analysis
(the detailed method for the PFOS determination is described in
the Supporting information). The pH values of the reaction solutions during the sorption process were monitored, and they
showed essentially no change in pH. The inﬂuence of pH was determined by measuring the sorption efﬁciency with pH values ranging
from 2.0 to 10.0 at 25 ± 1 °C in an incubator shaker (150 rpm).
Experimental uncertainties evaluated in ﬂasks without the membranes were less than 5% of the initial concentrations. The recoveries of the controls from the polypropylene conical ﬂasks ranged
from 95% to 103%. All experiments were run in triplicate, and the
average value was adopted. The sorption amount was calculated
according to the PFOS concentrations before and after sorption
using the following equation:
qt ¼ ðC 0 C t ÞV=m
Among all the sorption kinetic models, the pseudo-ﬁrst-order
(PFOM) and pseudo-second-order (PSOM) models were frequently
used. Some researchers have proposed that the modiﬁed PFOM and
PSOM had better adaptability to sorption kinetic data (Pan and
Xing, 2010). The modiﬁed PFOM and PSOM were thus adopted in
the present study. Because both the MWCNT-ENFMs and pure
ENFMs are porous sorbents, intra-ﬁber diffusion could exist during
the sorption processes. Therefore, the intraparticle diffusion model
proposed by Weber and Morris (WMM) was also used to describe
the data. Three sorption isotherm models, the Freundlich, Langmuir and partition–adsorption equations, were applied to ﬁt the
experimental data. The mean-weighted squared error (MWSE)
and correlation coefﬁcient (r2) were used to evaluate the goodness
of the model ﬁtting (Yang et al., 2006). The sorption kinetics and
isotherm models are listed in Table S1 of the Supporting
3. Results and discussion
3.1. Morphology and properties of the MWCNT-ENFMs
The SEM images of the MWCNT-ENFMs are shown in Fig. 1. The
images illustrate that the electrospun nanoﬁbers possess a common feature of being bead free and randomly arrayed, with some
MWCNTs ﬁlling the membranes. The average ﬁber diameter of
the MWCNT-ENFMs was approximately 100 ± 20 nm, and the
diameter of the MWCNTs was about 30–50 nm. Fig. 1a shows that
the MWCNTs were successfully immobilized into the ENFMs during the electrospinning process as one of the following three forms:
(1) the MWCNT penetrated into the ﬁbers with both ends exposed
outside the ﬁbers (see Fig. 1b); (2) one end of the MWCNT was in-
Fig. 1. SEM micrographs of MWCNTs-ENFMs, the images (b), (c and d) show the respective enlarged image of the marked areas using red ellipses and words in image (a). (For
interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of this article.)
Y. Dai et al. / Chemosphere 93 (2013) 1593–1599
serted into the ﬁbers, while the other end was exposed outside (see
Fig. 1c); and (3) the MWCNTs were twined around or between the
ﬁbers (see Fig. 1c and d) and were interweaved into the network
structure with the ﬁbers (see Fig. 1d). The immobilized forms of
the MWCNTs in the ENFMs were related to the interactions of several physical instability processes, such as the Rayleigh instability,
axisymmetric instability and bending (or whipping) instability, under the strong electrostatic ﬁeld of electrospinning (Li and Xia,
2004; Greiner and Wendorff, 2007). Due to these immobilized
forms, the MWCNTs not only retain their sorption properties, but
they also may not be easily released into the environment.
It is worth noting that the electrospinning voltage for preparing
the MWCNT-ENFMs in our experiments was 11 kV. However, the
voltage was increased to 20 kV to obtain similar ﬁber diameters
of the pure ENFMs (see Fig. S1 of the Supporting information). This
was mainly due to the fact that the MWCNTs are conductive, and
they enhanced the conductivity of the electrospinning solution,
resulting in a decrease of the ﬁber diameter. The addition of the
MWCNTs (0.5 wt.% vs. ENFMs) affected both the electrospinning
process and the properties of the ENFMs. The speciﬁc surface area
of the MWCNT-ENFMs (13.2 m2 g1) was doubled compared to the
pure ENFMs (5.86 m2 g1), and the tensile stress of the membrane
was also enhanced from 0.587 MPa (pure ENFMs) to 1.225 MPa
3.2. Sorption kinetics
Fig. 2 shows the sorption kinetics of PFOS onto the pure ENFMs
and MWCNT-ENFMs. Both of the sorption processes were mostly
completed within the ﬁrst 5 h of reaction and then became more
gradual until equilibrium was reached. Although the sorption
trends of PFOS onto the two sorbents were similar, their kinetic
proﬁles were quite different. The sorption rate of PFOS onto the
pure ENFMs was low, and the adsorbed amount was only
0.29 lmol g1 at 5 h, while the adsorbed amount reached approximately 1.13 lmol g1 onto the MWCNT-ENFMs at 5 h. The obviously faster sorption rate was mainly due to the strong
hydrophobicity of MWCNTs and the larger speciﬁc surface area
of the MWCNT-ENFMs, which provided more sorption sites for
PFOS. However, because of the complex porous structure of the
MWCNT-ENFMs, the sorption equilibrium of PFOS onto the
MWCNT-ENFMs was reached after 40 h, while only 20 h was required to achieve equilibrium for the pure ENFMs.
To further understand the sorption kinetics, the modiﬁed PFOM
and PSOM were applied to describe the sorption kinetics data. As
seen from Fig. 2a and Table S2 of the Supporting information, the
estimated correlation coefﬁcient (r2) demonstrated that the PSOM
ﬁtted the experimental data better than the PFOM. This was also
veriﬁed as a result of the equilibrium sorption amounts of PFOS ﬁtted by PSOM being much closer to the experimental values
(0.47 lmol g1 onto the pure ENFMs and 1.44 lmol g1 onto the
MWCNT-ENFMs) than those ﬁtted by the PFOM. The modiﬁed
parameter k2 was considered to be a more applicable rate constant
to directly describe the adsorption kinetic process based on the
PSOM (Pan and Xing, 2010). The k2 of the PFOS adsorption onto
the MWCNT-ENFMs was higher than the k2 of the PFOS adsorption
onto the pure ENFMs, indicating a faster sorption rate of PFOS onto
the MWCNT-ENFMs. Furthermore, the initial sorption rate (t0) of
PFOS onto the MWCNT-ENFMs (1.81 lmol h1 g1) was also much
faster than that onto the pure ENFMs (0.11 lmol h1 g1), and the
MWCNT-ENFMs contained nearly triple the sorption capacity
(1.45 lmol g1) as did the pure ENFMs (0.47 lmol g1). For a given
adsorbate, the sorption rate is mainly determined by the available
sites of the sorbent, typically displaying a positive correlation. In
the present study, the addition of the MWCNTs into the ENFMs increased the speciﬁc surface area of the membranes and provided
more sorption sites for PFOS. Additionally, the immobilized forms
(as discussed above) of the MWCNTs in the ENFMs provided many
nanotube ends (strong sorption sites) that were located outside the
ﬁbers and enhanced the sorption rate (Yang et al., 2006).
The WMM was adopted to ﬁt the sorption kinetics because the
PSOM cannot provide a deﬁnite process for sorption. The model
illustrates that if intra-sorbent diffusion is the sole rate-controlling
factor in a given system, a good linear relationship should be obtained from the plot of adsorbate uptake (qt) vs. the square root
of time (t1/2), and the line should also pass through the origin
(Yu et al., 2009; Pan et al., 2010). As shown in Fig. 2b and
Table S2 of the Supporting information, the WMM ﬁtted the sorption data of PFOS onto the two sorbents with relatively high correlation coefﬁcients (r2 > 0.97), but the plots did not pass through the
origin. This implied that intra-sorbent diffusion might have been a
rate-controlling step. The sorption rate might also have been inﬂuenced by the morphology and the natural properties of the sorbent,
the concentration of the adsorbate, and its afﬁnity to the adsorbent
(Cheung et al., 2007).
As seen in Fig. 2b, the plots were not linear throughout the
experiments, suggesting that more than one process affected the
sorption. In this study, the sorption process of PFOS onto the sorbents involved three possible steps. The initial linear portion demonstrated a rapid sorption phase that included two steps: (1) PFOS
diffusion in the liquid phase (2) and external mass transfer to the
membrane surface. The later plateau indicated the equilibrium
stage, namely intra-sorbent diffusion onto the sorption sites. In
addition to the remaining PFOS concentration in the solution, this
Fig. 2. Sorption kinetics of PFOS on the pure ENFMs and MWCNTs-ENFMs ﬁtted by (a) the pseudo-ﬁrst-order model (PFOM) and pseudo-second-order model (PSOM); and (b)
Weber and Morris model (WMM).
Y. Dai et al. / Chemosphere 93 (2013) 1593–1599
step was also inﬂuenced by the morphology and the structure of
the sorbent. The slope of the later stage for the MWCNT-ENFMs
was larger than that of the pure ENFMs (see Fig. 2b), which might
have been because the existence of the MWCNTs made the
MWCNT-ENFMs more porous. Therefore, the diffusion and sorption
of PFOS onto the MWCNT-ENFMs required more time while it also
achieved a greater sorption capacity. This three-stage sorption is
similar to the results reported from other studies (Cheung et al.,
2007; Yu et al., 2009).
3.3. Sorption isotherms
The sorption isotherms of PFOS onto the pure ENFMs and
MWCNT-ENFMs are shown in Fig. 3. Isotherm ﬁtting with model
equations is critical to the exploration of sorption mechanisms.
Three widely used models, including the Freundlich model, Langmuir model and partition–adsorption model, were adopted to describe our experimental data. The ﬁtted parameters for all
isotherms are summarized in Table S3 of the Supporting
As seen in Fig. 3 and Table S3, the sorption isotherms of PFOS
onto the pure ENFMs and MWCNT-ENFMs were ﬁtted well by all
tested models. However, the Freundlich model ﬁtted the data
slightly better than the other two models, judging from the higher
r2 (>0.90) and lower MWSE values. According to the results of the
Freundlich ﬁtting, both isotherms were nonlinear, with values of
n1 ranging from 0.75 to 0.79 (n1 = 1 for a linear isotherm). Nonlinearity can result from sorption site heterogeneity and/or sorbate–sorbate interactions such as electrostatic repulsion,
considering the anionic property of PFOS due to its low pKa
(3.27) (Yu et al., 2009) in the pH range studied. Moreover,
Fig. 3 shows that the MWCNT-ENFMs, with their higher speciﬁc
surface areas, resulted in much lower equilibrium concentrations
in the aqueous phase. These results indicated that the sorption
capacity of PFOS onto the MWCNT-ENFMs (16.29 ± 0.26 lmol g1)
was higher than the sorption capacity of PFOS onto the pure ENFMs
(0.92 ± 0.06 lmol g1), an increase of nearly 18 times. These results
also agreed with the order of Q0 and Kl, which are also indicators of
the sorption capacity of the adsorbents. Although the r2 value ﬁtted
by the Langmuir model was high, the adsorption behavior of PFOS
onto the membrane surface might be not only the monolayer coverage. Because of the hydrophobic perﬂuorinated chain of PFOS,
the multilayer sorption might also occur at higher equilibrium concentrations (Chen et al., 2011).
The partition–adsorption model used in this study was a Langmuir-type model that assumed the ‘‘site type’’ for the Langmuir
adsorption and was a combination of the Freundlich model and
the Langmuir model. Its major differences from the Freundlich
model were that it reverts to linearity at very low aqueous concentrations and that it usually results in a maximum capacity for
adsorption (Yang and Xing, 2010). However, the sorption data of
the PFOS onto the MWCNT-ENFMs from our experiments made it
difﬁcult to accurately estimate the sorption capacity using the partition–adsorption model. Because the overall sorption behaviors
were dominated by partitioning within the observed concentration
range, which was reﬂected by that the estimated partitioning contribution accounted for the bulk of the total amount of PFOS that
was adsorbed (Yang and Xing, 2010). Nevertheless, for the sorption
of PFOS onto the pure ENFMs, the total sorption was dominated by
the adsorption component, and the estimated sorption capacity
(0.85 ± 0.11 lmol g1) of PFOS was closer to the value that was ﬁtted by the Langmuir model (0.92 ± 0.06 lmol g1).
3.4. Effect of pH
The inﬂuence of pH on the sorption of PFOS onto the pure
ENFMs and MWCNT-ENFMs is illustrated in Fig. 4a. The PFOS sorption efﬁciencies of both sorbents decreased as pH values increased,
and the pH produced a stronger effect on the PFOS sorption onto
the MWCNT-ENFMs than onto the pure ENFMs. The PFOS sorption
efﬁciency onto the MWCNT-ENFMs decreased by >30% as the pH
was increased from 2.0 to 10.0, while the sorption efﬁciency onto
the pure ENFMs decreased by <20%. Regardless, the sorption efﬁciency onto the MWCNT-ENFMs was still much higher compared
to that onto the pure ENFMs within the pH range investigated in
The solution pH not only affects the sorbate speciation in solution, but it also inﬂuences the properties of the sorbent surface. Because the pKa value of PFOS (3.27) was lower than the pH values
(2.0–10.0) investigated in the present study, PFOS mainly existed
as an anion in solution throughout the entire experiments. Therefore, the sorption of PFOS at the different pH values was affected by
the characteristics of the sorbent surface. Fig. 4b presents the zeta
potentials of the sorbents at different pH solutions. The MWCNTENFMs were negatively charged within the pH range studied, indicating that electrostatic interaction between the MWCNT-ENFMs
and PFOS were repulsive. Furthermore, the zeta potentials of the
MWCNT-ENFMs decreased with increasing pH, which made the
interactions between sorbent and PFOS more repulsive and decreased the sorption efﬁciencies. However, the point of zero charge
(PZC) of the pure ENFMs in this study was measured at pH 3.1.
Thus, the surfaces of the pure ENFMs were positively charged at
pH 2.0 and 3.0 and were negatively charged at solution pH 4.0–
10.0. The higher sorption efﬁciencies of PFOS onto the pure ENFMs
at pH 2.0 and 3.0 were attributed to the electrostatic attraction between the sorbent and the PFOS. As the pH was increased, the
increasing electrostatic repulsion between the pure ENFMs and
PFOS decreased the sorption efﬁciencies.
3.5. Sorption mechanism
Fig. 3. Sorption isotherms of PFOS on the pure ENFMs and MWCNTs-ENFMs ﬁtted
by Freundlich model, Langmuir model and partition–adsorption model.
The experimental results indicated that the sorption capacity of
the MWCNT-ENFMs for PFOS was signiﬁcantly higher than that of
the pure ENFMs, which was mainly attributed to the addition of
the MWCNTs. The large speciﬁc surface areas of the MWCNTs provided more sorption spaces and sites for PFOS while the extreme
hydrophobicity of the MWCNTs also enhanced the sorption of
PFOS. Hydrophobic interactions between the MWCNTs and PFOS
existed during the sorption processes because the C–F chains in
PFOS exhibit hydrophobic properties. Because the PDLLA was a
type of hydrophobic polymer (contact angle: 95.1°), hydrophobic
interactions between the ENFMs and PFOS occurred. Furthermore,
Y. Dai et al. / Chemosphere 93 (2013) 1593–1599
Fig. 4. The effects of solution pH on (a) the sorption efﬁciency of PFOS on the pure ENFMs and MWCNTs-ENFMs, and (b) the zeta potential of the pure ENFMs and MWCNTsENFMs.
the good ﬁt of the PSOM for the kinetics data indicated that chemical interactions were likely involved during the sorption processes
(Yu et al., 2009). As discussed above, an electrostatic repulsion existed between the MWCNT-ENFMs and PFOS, which would prevent
the PFOS anions from approaching the MWCNT-ENFMs if this
repulsive force was stronger than the combined sum of the other
attractive interactions. However, a high sorption efﬁciency of PFOS
(>75%, pH = 6.0) was achieved, indicating that the hydrophobic
interaction between the MWCNT-ENFMs and PFOS was much
stronger than the electrostatic repulsion. Hydrophobic interactions
between the MWCNTs and PFOS played a more signiﬁcant role
during the sorption of PFOS onto the MWCNT-ENFMs.
For the pure ENFMs, although the electrostatic repulsion existed
when the solution pH was greater than 3.1, the sorption efﬁciencies for PFOS were over 20%. The hydrophobic interactions between the pure ENFMs and PFOS were responsible for the
sorption of PFOS onto the pure ENFMs. Moreover, the electrostatic
attraction between the pure ENFMs and PFOS enhanced the sorption efﬁciencies of PFOS when the solution pH was between 2.0
A schematic diagram of the sorption of PFOS onto the pure
ENFMs and MWCNT-ENFMs is shown in Fig. 5. The hydrophobic
interactions and electrostatic interactions are the main
mechanisms for the sorption of PFOS onto the pure ENFMs. The
Fig. 5. The schematic diagram of PFOS sorption on the pure ENFMs and MWCNTs-ENFMs. The predicted sorption mechanisms of PFOS on and MWCNTs-ENFMs included
hydrophobic interaction (between MWCNTs and PFOS, between nanoﬁbers and PFOS) and electrostatic repulsion (between MWCNTs-ENFMs and PFOS) at the pH range
studied (2.0–10.0). The hydrophobic interaction could overcome the electrostatic repulsion and facilitate the sorption of PFOS onto the MWCNTs and electrospun nanoﬁbers.
Besides the hydrophobic interaction between nanoﬁbers and PFOS, the electrostatic repulsion (pH > 3.1) and electrostatic attraction (pH < 3.1) was involved in the sorption
process of PFOS on the pure ENFMs, respectively.
Y. Dai et al. / Chemosphere 93 (2013) 1593–1599
electrostatic attraction and repulsion between the pure ENFMs and
PFOS existed when the solution pH was less than and greater than,
respectively, 3.1. The sorption mechanisms of PFOS onto the
MWCNT-ENFMs included mainly hydrophobic interactions (between the MWCNTs and PFOS, and between the ENFMs and PFOS)
and electrostatic repulsion. The C–F chains of PFOS can be adsorbed
in parallel or random to the MWCNTs and ﬁber axis, or the long C–
F chains may be adsorbed closely onto the MWCNTs and ﬁber surfaces along the curvature.
The MWCNT-ENFMs were successfully prepared and used for
the sorption of PFOS in aqueous solutions. The sorption kinetic results showed faster sorption rates of PFOS onto the MWCNTENFMs than onto the pure ENFMs, and the PSOM describes the
sorption kinetics well. The sorption isotherms showed that the
maximum adsorption capacities of PFOS onto the pure ENFMs
and the MWCNT-ENFMs were 0.92 ± 0.06 lmol g1 and
16.29 ± 0.26 lmol g1, respectively. The enhanced sorption of PFOS
onto the MWCNT-ENFMs was attributed to the larger speciﬁc surface area and stronger adsorption capacity of MWCNTs. The results
suggested that the solution pH produced a signiﬁcant effect on
PFOS sorption, and the sorption efﬁciencies of PFOS decreased with
the increasing solution pH. The hydrophobic interactions between
the sorbents and PFOS were much stronger than the electrostatic
repulsion, which played a predominant role in the sorption of
Because the morphology and structure of the MWCNT-ENFMs
can be adjusted by altering the electrospinning parameters, it is
possible to further optimize the MWCNT-ENFMs to increase their
sorption capacities. Moreover, the immobilization of the MWCNTs
into the ENFMs not only maintained the sorption properties of the
MWCNTs, but also prevented the release of the MWCNTs into the
water. The abovementioned advantages combined with their operational simplicity indicated that MWCNT-ENFMs are promising
sorbents for PFOS removal from aqueous solutions.
This work was supported by the Fund for Creative Research
Groups of the National Natural Science Foundation of China (No.
51121003), the Fundamental Research Funds for the Central Universities (No. 2012LZD03) and the Program of the Co-Construction
with Beijing Municipal Commission of Education of China.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.chemosphere.
Bang, H., Gopiraman, M., Kim, B.S., Kim, S.H., Kim, I.S., 2012. Effects of pH on
electrospun PVA/acid-treated MWNT composite nanoﬁbers. Colloid. Surf. A 409,
Burger, C., Hsiao, B.S., Chu, B., 2006. Nanoﬁbrous materials and their applications.
Ann. Rev. Mater. Res. 36, 333–368.
Chen, X., Xia, X., Wang, X., Qiao, J., Chen, H., 2011. A comparative study on sorption
of perﬂuorooctane sulfonate (PFOS) by chars, ash and carbon nanotubes.
Chemosphere 83, 1313–1319.
Cheng, J., Vecitis, C.D., Park, H., Mader, B.T., Hoffmann, M.R., 2008. Sonochemical
degradation of perﬂuorooctane sulfonate (PFOS) and perﬂuorooctanoate (PFOA)
in landﬁll groundwater: environmental matrix effects. Environ. Sci. Technol. 42,
Cheung, W.H., Szeto, Y.S., McKay, G., 2007. Intraparticle diffusion processes during
acid dye adsorption onto chitosan. Bioresour. Technol. 98, 2897–2904.
Dai, Y.R., Niu, J.F., Yin, L.F., Xu, J.J., Xi, Y.H., 2011. Sorption of polycyclic aromatic
hydrocarbons on electrospun nanoﬁbrous membranes: sorption kinetics and
mechanism. J. Hazard. Mater. 192, 1409–1417.
Deng, S., Zhang, Q., Nie, Y., Wei, H., Wang, B., Huang, J., Yu, G., Xing, B., 2012.
Sorption mechanisms of perﬂuorinated compounds on carbon nanotubes.
Environ. Pollut. 168, 138–144.
Greiner, A., Wendorff, J., 2007. Electrospinning: a fascinating method for the
preparation of ultrathin ﬁbers. Angew. Chem. Int. Ed. 46, 5670–5703.
Hagenaars, A., Knapen, D., Meyer, I.J., van der Ven, K., Hoff, P., De Coen, W., 2008.
Toxicity evaluation of perﬂuorooctane sulfonate (PFOS) in the liver of common
carp (Cyprinus carpio). Aquat. Toxicol. 88, 155–163.
Hu, J., Yu, J., Tanaka, S., Fujii, S., 2011. Perﬂuorooctane sulfonate (PFOS) and
perﬂuorooctanoic acid (PFOA) in water environment of Singapore. Water Air
Soil Pollut. 216, 179–191.
Huang, H., Huang, C., Wang, L., Ye, X., Bai, C., Simonich, M.T., Tanguay, R.L., Dong, Q.,
2010. Toxicity, uptake kinetics and behavior assessment in zebraﬁsh embryos
following exposure to perﬂuorooctanesulphonicacid (PFOS). Aquat. Toxicol. 98,
Jin, Y.H., Liu, W., Sato, I., Nakayama, S.F., Sasaki, K., Saito, N., Tsuda, S., 2009. PFOS
and PFOA in environmental and tap water in China. Chemosphere 77, 605–611.
Kunacheva, C., Fujii, S., Tanaka, S., Seneviratne, S.T.M.L.D., Nguyen Pham Hong, L.,
Nozoe, M., Kimura, K., Shivakoti, B.R., Harada, H., 2012. Worldwide surveys of
perﬂuorooctane sulfonate (PFOS) and perﬂuorooctanoic acid (PFOA) in water
environment in recent years. Water Sci. Technol. 66, 2764–2771.
Li, D., Xia, Y., 2004. Electrospinning of nanoﬁbers: reinventing the wheel? Adv.
Mater. 16, 1151–1170.
Lien, N.P.H., Fujii, S., Tanaka, S., Nozoe, M., Tanaka, H., 2008. Contamination of
perﬂuorooctane sulfonate (PFOS) and perﬂuorooctanoate (PFOA) in surface
water of the Yodo River basin (Japan). Desalination 226, 338–347.
Niu, J.F., Xu, J.J., Dai, Y.R., Xu, J.R., Guo, H.Y., Sun, K., Liu, R.L., 2013. Immobilization of
horseradish peroxidase by electrospun ﬁbrous membranes for adsorption and
degradation of pentachlorophenol in water. J. Hazard. Mater. 246–247, 119–
Pan, B., Xing, B.S., 2010. Adsorption kinetics of 17 a-ethinyl estradiol and bisphenol
A on carbon nanomaterials. I. Several concerns regarding pseudo-ﬁrst order and
pseudo-second order models. J. Soil Sediment. 10, 838–844.
Pan, G., You, C., 2010. Sediment-water distribution of perﬂuorooctane sulfonate
(PFOS) in Yangtze River Estuary. Environ. Pollut. 158, 1363–1367.
Pan, B., Sun, K., Xing, B.S., 2010. Adsorption kinetics of 17 a-ethinyl estradiol and
bisphenol A on carbon nanomaterials. II. Concentration-dependence. J. Soil
Sediment. 10, 845–854.
Senevirathna, S.T.M.L.D., Tanaka, S., Fujii, S., Kunacheva, C., Harada, H., Ariyadasa,
B.H.A.K.T., Shivakoti, B.R., 2010a. Adsorption of perﬂuorooctane sulfonate (nPFOS) onto non ion-exchange polymers and granular activated carbon: Batch
and column test. Desalination 260, 29–33.
Senevirathna, S.T.M.L.D., Tanaka, S., Fujii, S., Kunacheva, C., Harada, H., Shivakoti,
B.R., Okamoto, R., 2010b. A comparative study of adsorption of perﬂuorooctane
sulfonate (PFOS) onto granular activated carbon, ion-exchange polymers and
non-ion-exchange polymers. Chemosphere 80, 647–651.
Shintani, M., Naito, Y., Yamada, S., Nomura, Y., Zhou, S., Nakashimada, Y., Hosomi,
M., 2008. Degradation of perﬂuorooctansulfonate (PFOS) and perﬂuorooctanoic
acid (PFOA) by mechanochemical treatment. Kag. Kog. Ronbunshu 34, 539–544.
Thavasi, V., Singh, G., Ramakrishna, S., 2008. Electrospun nanoﬁbers in energy and
environmental applications. Energy Environ. Sci. 1, 205–221.
Thompson, J., Eaglesham, G., Mueller, J., 2011. Concentrations of PFOS, PFOA and
other perﬂuorinated alkyl acids in Australian drinking water. Chemosphere 83,
Tsuda, T., Inoue, A., Igawa, T., Tanaka, K., 2010. Seasonal changes of PFOS and PFOA
concentrations in Lake Biwa water. Bull. Environ. Contam. Toxicol. 85, 593–597.
Vecitis, C.D., Park, H., Cheng, J., Mader, B.T., Hoffmann, M.R., 2009. Treatment
perﬂuorooctanoate (PFOA). Front. Environ. Sci. Eng. China 3, 129–151.
Wang, F., Shih, K., 2011. Adsorption of perﬂuorooctanesulfonate (PFOS) and
perﬂuorooctanoate (PFOA) on alumina: inﬂuence of solution pH and cations.
Water Res. 45, 2925–2930.
Wang, T., Wang, Y., Liao, C., Cai, Y., Jiang, G., 2009. Perspectives on the inclusion of
perﬂuorooctane sulfonate into the Stockholm convention on persistent organic
pollutants. Environ. Sci. Technol. 43, 5171–5175.
Wang, F., Liu, C., Shih, K., 2012. Adsorption behavior of perﬂuorooctanesulfonate
(PFOS) and perﬂuorooctanoate (PFOA) on boehmite. Chemosphere 89, 1009–
Yamada, S., Naito, Y., Yamamoto, T., Noma, Y., Hosomi, M., 2008. Degradation fate of
perﬂuorooctansulfonate (PFOS) and perﬂuorooctanoic acid (PFOA) by UV
irradiation. Kag. Kog. Ronbunshu 34, 410–414.
Yang, K., Xing, B.S., 2010. Adsorption of organic compounds by carbon
nanomaterials in aqueous phase: Polanyi theory and its application. Chem.
Rev. 110, 5989–6008.
Yang, K., Zhu, L.Z., Xing, B.S., 2006. Adsorption of polycyclic aromatic hydrocarbons
by carbon nanomaterials. Environ. Sci. Technol. 40, 1855–1861.
Yu, Q., Zhang, R.Q., Deng, S.B., Huang, J., Yu, G., 2009. Sorption of perﬂuorooctane
sulfonate and perﬂuorooctanoate on activated carbons and resin: kinetic and
isotherm study. Water Res. 43, 1150–1158.