Journal of Colloid and Interface Science 369 (2012) 193–201

Contents lists available at SciVerse ScienceDirect

Journal of Colloid and Interface Science

www.elsevier.com/locate/jcis

Interactions between surfactants and silver nanoparticles of varying charge

J. Hedberg a, M. Lundin a,b, T. Lowe a,b, E. Blomberg a,c, S. Wold b, I. Odnevall Wallinder a,⇑
a
KTH Royal Institute of Technology, School of Chemical Science and Technology, Surface and Corrosion Science, SE-100 44 Stockholm, Sweden
b
KTH Royal Institute of Technology, School of Chemical Science and Technology, Applied Physical Chemistry, SE-100 44 Stockholm, Sweden
c
Institute for Surface Chemistry, P.O. Box 5607, SE-114 86 Stockholm, Sweden

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

Article history: The interaction between silver nanoparticles (Ag NPs) of different surface charge and surfactants relevant
Received 20 October 2011 to the laundry cycle has been investigated to understand changes in speciation, both in and during trans-
Accepted 2 December 2011 port from the washing machine. Ag NPs were synthesized to exhibit either a positive or a negative surface
Available online 11 December 2011
charge in solution conditions relevant for the laundry cycle (pH 10 and pH 7). These particles were
characterized in terms of size and surface charge and compared to commercially laser ablated Ag NPs.
Keywords: The surfactants included anionic sodium dodecylbenzenesulfonate (LAS), cationic dodecyltrimethylam-
Silver nanoparticles
moniumchloride (DTAC) and nonionic Berol 266 (Berol). Surfactant–Ag NP interactions were studied
Laundry detergents
Surfactants
by means of dynamic light scattering, Raman spectroscopy, zeta potential, and Quartz Crystal Microbal-
Surface Enhanced Raman Spectroscopy ance. Mixed bilayers of CTAB and LAS were formed through a co-operative adsorption process on
Zeta potential positively charged Ag NPs with pre-adsorbed CTAB, resulting in charge reversal from positive to negative
Dynamic light scattering zeta potentials. Adsorption of DTAC on negatively charged synthesized Ag NPs and negatively charged
commercial Ag NPs resulted in bilayer formation and charge reversal. Weak interactions were observed
for nonionic Berol with all Ag NPs via hydrophobic interactions, which resulted in decreased zeta poten-
tials for Berol concentrations above its critical micelle concentration. Differences in particle size were
essentially not affected by surfactant adsorption, as the surfactant layer thicknesses did not exceed more
than a few nanometers. The surfactant interaction with the Ag NP surface was shown to be reversible, an
observation of particular importance for hazard and environmental risk assessments.
Ó 2011 Elsevier Inc. All rights reserved.

1. Introduction shown to significantly influence the rate and chemical state of

Although classified as an environmental hazard, the use of silver
nanoparticles (Ag NPs) in consumer products is rapidly increasing
on the global market. Silver nanoparticles are used in a large
variety of applications such as sport wear, washing machines,
and food packaging materials [1,2], primarily due to the beneficial
antibacterial properties [3,4].
The rapid and extensive use of Ag NPs in different applications
is potentially problematic given that the possible adverse effects
on the environment and humans have not been extensively ex-
plored. Even though Ag NPs are one of the most commonly used
nanomaterials in consumer products [1,2], the mechanisms behind
the anti-bacterial properties of Ag NPs, their environmental fate,
metal release mechanisms, and the chemical state of released sil-
ver are far from defined. The transition between different chemical
surroundings has a major influence on the chemical form and any
environmental interaction [2,4,5]. For example, the presence of
organic and inorganic compounds and the effect of pH have been
⇑ Corresponding author. Fax: +46 8 208284.
E-mail address: ingero@kth.se (I.O. Wallinder).
released silver from silver nanoparticles [6,7]. An alarming fact is
that adsorption of molecules, such as capping agents, in some cases
may give rise to enhanced toxicity [8].
There is hence an urgent need for in-depth investigations on the
characteristics, reactivity, and fate of Ag NPs dispersed to the envi-
ronment via different routes. A relatively recent study showed that
significant amounts of silver, as ions and colloids, were released
from Ag NP impregnated sport socks when washed in deionized
water [9]. However, the chemical environment is significantly
more complex in a laundry cycle with components such as anionic
and cationic surfactants, zeolites, builders, and bleach [10]. From a
broader perspective, the applications of Ag NPs interacting with
surfactants are many, for example in the production of useful
nanostructures in microelectronic and catalysts, where surfactants
are often used as capping agents [11,12]. Questions that arise
include how surfactants interact with the Ag NP surface, what type
of structures are formed, and the effect of surfactant adsorption on
toxicity [8,13]. For example, the formation of bilayers of ionic sur-
factants on Ag NPs can reverse the surface charge of the Ag NPs
resulting in different toxicological behavior compared to the
unmodified Ag NPs. To complicate the understanding of surfactant

0021-9797/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcis.2011.12.004
194 J. Hedberg et al. / Journal of Colloid and Interface Science 369 (2012) 193–201
interaction with Ag NPs, there are numerous types of Ag NPs
synthesized and studied in the literature [2,5,6,8,9,14,15]. There
is a need for a fundamental understanding of the surfactant
adsorption on Ag NPs, like the presence of hydroxyl groups on
the silver surface, especially since these are known to play an
important role in surfactant adsorption on silica surfaces [16].
Furthermore, knowledge of the surface charge of the surfactant
modified Ag NPs and the structure of the adsorbed layer is crucial
to determine the stability of the Ag Nps,
The main focus of this study has been to investigate the adsorp-
tion of selected surfactants relevant to laundry detergents on Ag
NPs using Surface Enhanced Raman Spectroscopy (SERS). SERS
has the ability to distinguish between molecules adsorbed onto
nanostructured surfaces of silver and solvent molecules by utiliz-
ing the enhanced Raman signal from species adsorbed onto the
Ag NPs. As the SERS effect depends on several parameters such
as particle size and surface charge [17,18], complementary mea-
surements were performed using dynamic light scattering (DLS)
to determine the size distribution of the Ag NPs and zeta potential
measurements to determine their surface charge. Parallel sorption
studies were performed using Quartz Crystal Microbalance with
Dissipation (QCM-D) to measure the adsorbed amount of surfac-
tant on silver surfaces. The study includes different types of Ag
NPs (synthesized particles with negative and positive charge,
respectively, as well as commercially available laser ablated parti-
cles), as characteristics of commercially available Ag NPs are hard
to come by [19]. The negatively charged Ag NPs were synthesized
according to the procedures of Creighton et al. using sodium boro-
hydride to reduce silver nitrate into Ag NPs [20], and the positively
charged particles following the procedure described in Sui et al.
[21,22], where CTAB (cetyl trimethylammonium bromide) capped
silver particles were synthesized using borohydride and silver
nitrate. To simplify the system, the interaction between different
Ag NPs (with a particle concentration comparable with the maxi-
mum silver concentration that leaked out from sport socks in
Ref. [9]) and a selection of laundry detergent components (in
mM concentrations, roughly relevant for washing powder [10])
was investigated. The study included sodium dodecylbenzenesul-
fonate (LAS), representing a negatively charged surfactant, an
alcohol ethoxylate (Berol 266) representing a nonionic surfactant,
and finally dodecyltrimethylammoniumchloride (DTAC), repre-
senting a positively charged surfactant.
LAS and Berol 266 are common surfactants in laundry formula-
tions [10], whereas positively charged surfactants are more often
used in softeners. DTAC was selected because it has the same chain
length as LAS and is a commonly used surfactant, although not
necessarily as a fabric softener. All measurements were performed
at a laundry-cycle relevant pH (pH 10). Studies were also con-
ducted at pH 7 for comparison and to enable the investigation of
the transition of Ag NPs from alkaline to more pH neutral condi-
tions, which will be the case upon release of the washing solution.
2. Materials and methods
2.1. Commercial silver NPs
A commercial powder of Ag NPs, produced by laser ablation,
was purchased from EV NANO Technology Co., Ltd., China. The
powder was used as received and a colloidal solution produced
via sonication (Sonifier 250%, 30% duty cycle) for 5 min. The parti-
cle concentration in solution was set to 0.12 mg/10 mL of ultrapure
water (resistivity 18.2 MO cm), with a measured concentration of
4.4 mg/L in solution after sonication and filtration (0.45 lm, What-
man) determined by means of atomic absorption spectroscopy
(AAS) used in all experiments.
2.2. Negatively charged silver NPs
The negatively charged Ag NPs were synthesized following the
procedure of Creighton et al. [20]. Briefly, 17 mL of a 1 mM AgNO3
solution was added drop-wise into an ice-cold 3 mM NaBH4 solu-
tion (50 mL) under vigorous stirring. Prior to use, the solution
was stirred for 1 h followed by storage at dark conditions for
1 week to allow borohydride decomposition. The final silver
concentration, measured by means of AAS, was 25 mg/L and used
in all experiments.
2.3. Positively charged silver NPs
The positively charged Ag NPs were synthesized using the
method described by Sui et al. [21,22]. 50 mL of 2 mM AgNO3 with
0.4 M NH3 and 5  10 4 M CTAB solution was added drop-wise
during approximately 15 min to an ice-cold solution containing
50 mL of 12 mM NaBH4 and 5  10 4 M CTAB. The solution was
vigorously stirred during the addition and kept at stirring condi-
tions overnight. After synthesis, the appearance of the solution
changed from yellow to green/yellow. The mixture was thereafter
heated to approximately 60 °C for 3 h to remove remnant NH3 and
decompose excess NaHB4. The solution was finally filtered through
a 0.45 lm filter (Whatman), and stored at dark conditions prior to
use. The concentration of silver was when following this proce-
dure, 6.7 mg/L (AAS) in all experiments.
2.4. Solutions
Dodecyltrimethylammoniumchloride (DTAC), sodium dodecyl-
benzenesulfonate (LAS), sodium nitrate (NaNO3), and sodium
borohydride (NaHB4) were purchased form Sigma Aldrich (Reagent
grade) and used without further purification. Berol 266 (Akzo
Nobel), an alcohol ethoxylate with chain length C9–11 and 4–7
ethylene oxide groups (molecular weight range 320–480 g/mole
with median 402 g/mole quoted by the supplier), was used with-
out further purification. To aid readability, we will refer to the sur-
factant we used as Berol from here on in the text, although this
name usually refers to the whole class of surfactants supplied by
Akzo Nobel.
In water, the critical micelle concentration (CMC) values for the
surfactants used in this study are approximately 1.1 mM (CTAB),
20 mM (DTAC), 1.2 mM (LAS), and 75 lM (Berol) according to liter-
ature values [17] or the supplier (Akzo Nobel).
The pH of the Ag NP solutions was adjusted to pH 7 or to pH 10
by addition of Na2CO3 and HNO3. Measurements were usually car-
ried out within 2 h from sample preparation.
2.5. Surface Enhanced Raman Spectroscopy (SERS)
The Raman Spectrometer in a reflection configuration is in es-
sence similar to the Total Internal Reflection Raman spectrometer
described elsewhere [23]. Briefly, a highly stable 532 nm laser
(Quantum lasers) is directed to the sample in an external reflection
configuration at an angle of incidence of 78°, in S polarization (per-
pendicular to the plane of incidence). The Raman scattered light is
collected using an ultra-long high numerical aperture objective
with 50 magnification attached to a modified upright Axio micro-
scope from Zeiss. The scattered light is then passed through a sharp
long pass filter that blocks the 532 nm light and finally focused to a
spectrograph (Shamrock–Andor) and detected with a CCD camera
(Newton 940–Andor). Glass cuvettes were used as sample cells,
and all measurements were conducted at 20 °C. Raman spectra
from corresponding bulk solution without added Ag NPs were sub-
tracted unless otherwise noted.
 J. Hedberg et al. / Journal of Colloid and Interface Science 369 (2012) 193–201
2.6. Particle size distribution
Particle size distribution measurements were conducted by DLS
on an instrument employing photon cross-correlation spectros-
copy, PCCS (NanoPhox, Sympatec, Germany). Duplicate samples
were investigated, and all measurements were performed at 25 °C.
2.7. Zeta potential
Zeta potential measurements were carried out using a Malvern
Zetasizer Nano ZS instrument (Malvern Instruments, UK). In gen-
eral, duplicate samples were measured three times each at 25 °C.
However, the reported zeta potential values for Ag NP-surfactant
complexes are results from one single sample measured three
times. A single data set was selected to allow accurate comparison
of data between the same Ag NP solution with and without the
presence of surfactants. The deviation in zeta potential between
independent samples varied by 5–15% but the trend was the same.
The pH titrations were conducted by starting at neutral pH using
the auto-titration feature of the Malvern instrument to acidic pH
with 0.1 M HNO3 or alkaline pH with 0.1 M NaOH. Standard Latex
samples (Malvern Instruments) were tested prior to analysis to
ensure the accuracy of the measurements. The reported zeta po-
tential values were included after validating the quality of the
phase distribution.
2.8. Atomic absorption spectroscopy (AAS)
Total concentrations of silver in solutions were determined by
means of flame atomization atomic absorption spectroscopy (Per-
kin Elmer AAnalyst 800). Calibration was conducted using 1, 3,
10 mg/L silver standards (prepared from 1000 ppm standard, Per-
kin Elmer). Prior to the analysis, all samples were acidified with
concentrated HNO3 to a pH < 1. Triplicate measurements were per-
formed on each sample. Blank and standard samples were mea-
sured throughout the analysis. Samples with concentrations
exceeding 10 mg/L were diluted to fall within the calibration range
(1–10 mg/L). To confirm total atomization of the Ag NPs in the
flame, acid digested samples (24 h, 60 °C, concentrated HNO3)
were also measured. No difference in concentration was observed.
2.9. Quartz Crystal Microbalance (QCM)
Quartz Crystal Microbalance with Dissipation monitoring
(QCM-D) was conducted using a QCM-D E4 from Q-sense AB (Swe-
den). The adsorbed mass of surfactant molecules adsorbed onto
evaporated silver surfaces (QSX 304, Q-sense AB) was estimated
with the Sauerbrey equation [24]. Measurements were performed
at 25 °C with a continuous surfactant solution flow rate of 200 lL/
min during the entire measurement. The QCM Ag surfaces were
cleaned for 10 min in isopropylic alcohol and for 15 min in abso-
lute ethanol prior to use.
3. Results and discussion
Generated findings are divided into four parts, focusing on the
characteristics of the different Ag NPs in Section 3.1, and their
individual interactions with three laundry relevant surfactant
molecules (LAS, DTAC, and Berol) in Sections 3.2–3.4.
3.1. Silver colloid characteristics
In order to understand the behavior of the Ag NPs interacting
with different surfactants, the particles were characterized
as-synthesized (positively and negatively charged particles) or
195
as-received (commercial particles) at pH 7 and 10 in water
solutions pH adjusted to pH 7 or pH 10.
According to particle size distribution measurements (by
number), the particle size varied between 1 and 10 nm in all solu-
tions of Ag NPs, independent of surface charge, presence of added
surfactants and pH. A representative result can be seen in Fig. 1
where the relative number of particles at different sizes is shown.
The measured zeta potential values for the Ag NPs in water of
different pH are compiled in Table 1. In line with literature findings
[17], a zeta potential of approximately 58 mV was determined for
the negatively charged synthesized Ag NPs at pH 10 while the pos-
itively charged synthesized Ag NPs exhibited a zeta potential of
approximately +37 mV. Lowering the pH to 7 shifted the zeta
potentials to 46 mV and +34 mV, respectively. The commercial
Ag NPs displayed negative potentials, 44 mV and 23 mV at pH
10 and 7, respectively. These values were slightly less negative than
the negatively charged synthesized Ag NPs ( 58 mV (pH 10) and
46 mV (pH 7)), an effect that may be attributed to differences in
surface roughness depending on the difference in preparation.
The isoelectric point (IEP) was approximately three (pH 3) for
both the commercial and negatively charged synthesized Ag NPs
(Fig. 2), in agreement with previously reported findings [25]. The
origin of the negative charge will be discussed further below.
Raman spectroscopy was employed to provide insight in the
surface composition on the three different types of Ag NPs. The
results are shown in Fig. 3. Note that the negatively charged syn-
thesized Ag NPs were approximately five times more concentrated
(25 mg/L) than the commercial and positively charged synthesized
NPs in the presented Raman spectra.
The same concentration of negatively charged Ag NPs as for the
other types (about 5 mg/L) yielded no observable peaks in the Ra-
man spectra. Thus, to enable improved Raman spectra and a more
detailed understanding of adsorbed species, the 25 mg/L particle
concentration of negatively charged synthesized Ag NPs was used
throughout this work in all measurements including surface poten-
tial and size estimations. Several things can be concluded from the
results in Fig. 3. As expected, the positively charged Ag NPs are cov-
ered by CTAB, seen from peaks at 758,1020, 1380, 1449, 2725,
2850, 2880, 2930, 2970 cm 1 corresponding to CH3 rocking mode
from N+(CH3)3 group, CC stretch, C–CH3 symmetric bending, CH2
and CH3 bending, CH2 symmetric stretch, CH3 symmetric stretch,
asymmetric CH2 stretch, and asymmetric CH3 stretch of adsorbed
CTAB, respectively [26–28]. HNO3, used to adjust the pH, resulted
in additional peaks (marked with an asterisk) in the Raman spectra
for the positively charged synthesized silver NPs. Na2CO3, used to
Fig. 1. Particle size distribution (by number) positively and negatively charged
synthesized and commercial Ag NPs in ultrapure water (pH 10).
196 J. Hedberg et al. / Journal of Colloid and Interface Science 369 (2012) 193–201
Table 1
Zeta potential measurements of commercial and synthesized Ag NPs in pure water at
pH 10 and pH 7. The reported error is the standard deviation of three replicate
measurements on at least two independent samples.
Zeta potential (mV) Ultra-pure water
pH 7 10
Commercial NPs (laser ablated) 23 ± 10 44 ± 3
Negative NPs (BH4 synthesis) 46 ± 3 58 ± 2
Positive NPs (CTAB synthesis) +34 ± 3 +37 ± 6
Fig. 2. The zeta potential of commercial Ag NPs (triangles) and negatively charged
synthesized Ag NPs (squares) as a function of solution pH. The titration was carried
out with 0.1 M NaOH and 0.1 M HNO3. The reported zeta potential values are
obtained from a single measurement point at each pH for both synthesized and
commercial Ag NPs.
increase the pH to 10, also gave rise to weak interfering Raman
bands found to originate from bulk species, as confirmed in supple-
mentary experiments (data not shown). As previously discussed,
the native surface of the Ag NPs exhibited a negative charge at both
pH 7 and pH 10. Since a monolayer of CTAB would essentially give
rise to an uncharged surface, these findings imply the presence of
more than one monolayer of CTAB present on the particles. The
Raman spectrum for the positively charged Ag NPs was very differ-
ent from a pure CTAB solution, as the SERS effect enhanced bands
for the part of the molecule closest to the particle surface, in this
case the trimethylammonium head group, represented, for
example, by the band at 758 cm 1 [18,27]. Thus, the Raman results
provide strong evidence for the adsorption of CTAB via electrostatic
Fig. 3. Raman spectra of negatively charged and commercial Ag NPs in pure water solutions (left) and positively charged Ag NPs (right). Raman spectra from bulk water were
subtracted for the corresponding spectra with Ag NPs added in the case of the positively charged Ag NPs. Asterisks indicate interference from bands originating from the
buffers used and spectra are offset for clarity.
interaction with the head group oriented toward the negative
charged Ag NP surface. The net positive charge on these particles
arises from a combination of electrostatic attraction between the
cationic head group of the surfactant and the negatively charged
Ag NPs, coupled with hydrophobic interactions between surfactant
tails to promote formation of some type of surfactant bilayer on
the surface. However, the small size of the particles measured by
DLS indicates that the electrostatic repulsion between the cationic
head groups limits the formation of higher order aggregates, as
evidenced by the small size of the Ag NPs in the DLS results, c.f.
Fig. 1. It was therefore concluded that bilayers of CTAB were
formed with the positively charged trimethylammonium group
interacting with the silver surface, results that support previous
observations and the model postulated model by Sui [21,22]. In
that model, they showed that the surfactant molecules in the inner
or outer layer of the bilayer do not pack in a full-extended state but
rather bend or twist to each other resulting in a much smaller bi-
layer thickness than would be expected from the length of a CTAB
molecule (about 1–2 nm). In a previous study by Penfold et al., it
was shown that the adsorption of CTAB on hydrophilic silica re-
sulted in bilayer formation with a disordered layer structure,
where a fraction of the surface was covered by bilayer patches
[29]. In order to estimate the exact structure (disordered bilayer/
patchy bilayer) of the CTAB layer on the Ag NPs, a more surface
sensitive technique is needed such as the small angle neutron scat-
tering technique (SANS). Finally, it should be noted that the overall
weak Raman signals observed throughout this work, partly can be
explained by the small particle size (<10 nm), as the optimum for
SERS is regarded to be of the order 10–100 nm [12].
For the negatively charged synthesized Ag NPs, the only detect-
able peak in the Raman spectra in Fig. 2, which did not originate
from water, was located at 242 cm 1. The commercial silver NPs
showed no Raman bands; however, the measurements on precipi-
tated Ag NPs in aqueous solutions revealed a band at 242 cm 1.
Others have attributed this band to an Ag–O vibration of atomic,
reversibly adsorbed oxygen on the surface of silver particles
[30–32], and we support this assignment. In this work, no shift
was observed for this band (which would indicate surface Ag–OH
bonds) between the different pH values investigated. This may be
due to that the investigated pH values were significantly higher
than the IEP of the Ag NPs (Fig. 2). In addition, no isotope-induced
frequency shift was observed for measurements conducted in D2O,
and thus, it was concluded that the band does not correspond to
Ag–OH species. A hypothesis is that adsorbed OH species are
 J. Hedberg et al. / Journal of Colloid and Interface Science 369 (2012) 193–201
completely deprotonated at these pH values, in line with the low
IEP measured by zeta potential for the commercial Ag NPs
(Fig. 2). No peaks from borate, which could be present as a bi-prod-
uct of the synthesis of the negatively charged silver NPs [31,32],
were observed.
3.2. Surface interactions with the negatively charged surfactant LAS
Surface interactions between Ag NPs and the negatively charged
surfactant LAS were investigated using two different surfactant
concentrations (0.1 and 5 mM). These concentrations were
selected based on being below and above the CMC for LAS
(1.2 mM). The presence of LAS did not result in any observable ef-
fects on the size distribution measurements at any concentration
compared to pure water solutions (data not shown). However,
the DLS technique may not be able to resolve small changes due
to the surfactant interaction with the particles since the particles
are too small, the adsorbed surfactant layer is too thin or not struc-
tured enough to affect the hydrodynamic radius.
It is evident from the zeta potential results in Table 2 that the
addition of LAS moves the zeta potentials of the different Ag NPs
in the negative direction, with the most notable change for the pos-
itively charged particles, where a charge reversal was observed in a
surfactant concentration of 5 mM LAS. A surfactant concentration
below CMC (0.1 mM LAS) was not sufficient to reverse the sign of
the surface charge. According to Raman measurements (Fig. 1),
the positively charged Ag NPs yielded Raman bands corresponding
to LAS at 1135 and 1610 cm 1. These bands were assigned to
aromatic CH bending and CC aromatic stretch, respectively [33].
Since the main peaks attributed to CTAB, c.f. Fig. 1 and Table S1
(Supporting information), are located at 760, 1025 and
1449 cm 1, as well as CH-stretching vibrations in the 2800–
3000 cm 1 region, individual bands could not be identified in these
regions when LAS was added due to overlapping. Similar to find-
ings in pure water, the Raman spectra revealed bands assigned to
NO3 originating from HNO3 and the Na2CO3 buffers, confirmed as
originating from bulk species in complementary experiments. Ob-
served Raman results for the positive Ag NPs containing 5 mM LAS
are in line with the zeta potential measurements, showing spectra
with more intense peaks due to LAS in the 5 mM solution com-
pared with 0.1 mM (Fig. 1). However, the occurrence of weak LAS
bands in the Raman spectra for 0.1 mM concentration at pH 7
did not correlate with a markedly changed zeta potential. In fact,
the zeta potential is slightly more positive upon the addition of
0.1 mM LAS (at pH 7) as shown in Table 2. As the Raman signal
is not quantitative, there is a possibility that the number of con-
tributing LAS molecules to the Raman signal is low due to the pres-
ence of hot spots. Hot spots are locations where the electrical field
is very high due to a pronounced surface roughness with sharper
edges in the material [18,34] thus not affecting the zeta potential,
which is a macroscopic property. The attraction between the ad-
sorbed CTAB on the positively charged Ag NPs and LAS is expected
due to electrostatic interactions and hydrophobic interactions
Table 2
Zeta potential measurements of commercial and synthesized Ag NPs in pure water with and without LAS of different concentration at pH 10 and pH 7. The reported error is the
standard deviation of three replicate measurements on the same sample.
Zeta potential (mV) Ultrapure water
Concentration
pH 7 10
Ag NPs – commercial 24 ± 2 45 ± 1
Ag NPs – negative 46 ± 1 58 ± 1
Ag NPs – positive +33 ± 3 +29 ± 1
197
between their hydrophobic tails [35,36]. The LAS molecules prob-
ably form complexes with the pre-adsorbed CTAB on the positively
charged Ag NPs. A likely scenario is that LAS molecules are incor-
porated in the ‘‘unordered’’ CTAB bilayer, forming a mixed bilayer,
rather than adsorbing as a bilayer on top of the CTAB bilayer. This
explains the observed Raman signal from LAS molecules, the
charge reversal due to addition of LAS but absence of an increased
particle size due to adsorption of LAS molecules. Further, when
viewing the Raman data, the CTAB bands weaken when adding
LAS (5 mM), implying that some CTAB molecules may be removed
upon addition of LAS. Desorption of CTAB molecules will also con-
tribute to the observed high negative charge (approx. 60 mV) of
the positively charged Ag NPs (capped with CTAB) in the presence
of 5 mM LAS (Table 2).
The negatively charged synthesized and commercial Ag NPs did
not exhibit any vibrational bands in the Raman spectra after addi-
tion of LAS (data not shown). However, the zeta potential findings
presented shifts in the negative direction upon addition of LAS,
compared with the zeta potentials of the pure Ag NPs (Table 2).
For the commercial Ag NPs in solution at pH 10, the zeta potential
was substantially higher in the presence of 5 mM LAS ( 88 mV)
compared to water ( 45 mV). This implies that the LAS molecules
are at least in close proximity to the Ag NP surface. LAS surfactants
in a 5 mM solution (in absence of Ag Nps) gave rise to a measured
zeta potential value of approximately 10 mV. This appears to be a
small contribution to the measured potential of Ag NP in the pres-
ence of 5 mM LAS; however, the possibility that LAS micelles
contribute to the measured zeta potential values in Table 2 cannot
be completely ruled out. Adsorption of ionic surfactants on like-
charged substrates is less well understood but can occur due to
hydrophobic interactions, hydrogen bonding, and attractive dis-
persion forces as described in several reviews, e.g. [37]. As the SERS
effect reaches approximately 10 nm from the surface, with the first
monolayers being the most enhanced [38], the lack of Raman
bands shows that no close or strong interaction takes place be-
tween the Ag NPs and the LAS molecules.
To conclude, LAS influence the surface reactivity of the Ag NPs,
most notably observed by the increased negative surface potential
of Ag NPs in the presence of a LAS solution. The most drastic
change was observed for positively charged Ag NPs, causing a
charge reversal and the presence of LAS molecules detected by Ra-
man (indicative of a close LAS-surface interaction). This is in accor-
dance with the work of Fan and Guo [12] where the mixtures of
CTAB/LAS added to Ag NPs resulted in an inhomogeneous nucle-
ation of silver, with the formation of different nanostructures. It
was suggested that surfactant adsorption on Ag NPs caused the
inhomogeneous nucleation. Thus, based on our results, we can
confirm that CTAB and LAS do form a mixed bilayer since the zeta
potential changed from net positively charged for Ag NPs pre-ad-
sorbed with CTAB to net negatively charged when LAS was added
at surfactant concentrations exceeding CMC (5 mM). Further,
Raman results confirmed the presence of both surfactants on the
Ag NP surface. Although the experimental conditions were not
the same, the formation of not fully covered bilayers of CTAB/LAS
Sodium dodecylbenzenesulfonate (LAS)
0.1 mM 5 mM
7 10 7 10
54 ± 1 61 ± 1 73 ± 1 88 ± 4
63 ± 1 70 ± 2 72 ± 3 72 ± 2
+46 ± 1 +32 ± 2 65 ± 3 56 ± 1
198 J. Hedberg et al. / Journal of Colloid and Interface Science 369 (2012) 193–201
may explain the formation of different silver nanostructures in the
work of Fan and Guo [12]. In addition, cooperative adsorption of
LAS above CMC on Ag NPs with CTAB pre-adsorbed is observed,
similar to previous findings of nonionic and cationic surfactants
on hydrophilic silica [29]. Negatively charged surfactants are
widely used in laundry detergents to solubilize dirt and also avoid
adsorption of the surfactant to the fabric, which is also negatively
charged [10]. Hence, the potential release of Ag NPs from textiles
washed in the laundry cycle interacting with LAS would be ex-
pected to have a greater chance in escaping the laundry cycle
due to electrostatic repulsion with the negative charge of the fabric
of clothes.
3.3. Surface interactions with the positively charged surfactant DTAC
Positively charged surfactants are used in softeners for laundry
applications. Even though DTAC is not a component of laundry
applications formulations, its selection was justified from its simi-
lar chain length as the negatively charged laundry surfactant LAS.
The DTAC concentrations investigated, 0.1 and 5 mM, were the
same as studied for the LAS interactions with the Ag NPs. Both
concentrations were below the CMC (20 mM) of DTAC. CTAB was
already adsorbed on the positively charged silver NPs, and there-
fore, the interaction with DTAC was not examined with Raman.
DTAC and CTAB are very similar molecules, which also explain
why the addition of DTAC to the positively charged synthesized
Ag NP particles (with CTAB) did not result in any change in zeta po-
tential (Table 3). This is expected since the only difference between
the two surfactants (except the different counter-ion Br versus
Cl ) is that CTAB has four extra carbons in the hydrocarbon chain
in comparison with DTAC. It is therefore unlikely that the less
hydrophobic surfactant DTAC would replace the already adsorbed
CTAB since it would be energetically unfavorable.
Zeta potential measurements for the negatively charged synthe-
sized and commercial Ag NPs revealed a charge reversal, compared
to water, at a concentration of 5 mM DTAC, Table 3, and less neg-
ative zeta potentials at a concentration of 0.1 mM. Positive zeta
potentials in 5 mM DTAC indicate the formation of some kind of bi-
layer of DTAC on the particle surface. Similar to findings with LAS,
no significant change in particle size distribution was observed in
contact with DTAC compared with pure water conditions. These
findings are expected since the zeta potentials are large enough
(>20 mV) to prevent particle aggregation.
Weak bands at 760, 1310, 1448, 2855, 2930, 2966 cm 1, all orig-
inating from DTAC [27,28] were observed in the Raman spectra for
the negatively charged Ag NPs (Fig. 5). Similar to the findings for
LAS, the Raman results are consistent with the zeta potential mea-
surements and show that adsorption of the detergent occurs at the
Ag NP surface through electrostatic interactions. Similar to the find-
ings for the pure positively charged Ag NP particles, Raman bands at
760 and 1448 cm 1 were prominent. Due to the short range SERS
effect, which enhances vibrational bands for compounds close to
the surface [34,39], the trimethylammonium group interacts with
the particle surface. It should be noted that the band at
Table 3
Zeta potential measurements of commercial and synthesized Ag NPs in pure water with and without DTAC of different concentration at pH 10 and pH 7. The reported error is the
standard deviation of three replicate measurements on the same sample.
Zeta potential (mV) Ultrapure water
Concentration
pH 7 10
Ag NPs – commercial 24 ± 2 45 ± 1
Ag NPs – negative 46 ± 1 58 ± 1
Ag NPs – positive +33 ± 3 +29 ± 1
1310 cm 1, originating from CH2 twisting vibrations [28] was
absent for the 0.1 mM concentration. Since the zeta potential
measurements at the 5 mM concentration imply the adsorption of
DTAC as bilayers, the occurrence of the band at 1310 cm 1 may
indicate that hydrocarbon chains are, to a larger extent, present in
the proximity of the silver surface as the band might originate from
CH2 groups near the end of the chain [28], possibly intercalated into
the first layer of adsorbed DTAC. For the commercial Ag NPs, inves-
tigated with five times lower particle concentration compared with
the negative charged Ag NPs, c.f. discussion above, also weak but
detectable Raman signals from DTAC were observed indicative for
DTAC–surface interactions. The band at 1635 cm 1, Fig. 4, origi-
nates from water, a residual after subtraction of corresponding bulk
water spectrum. The Raman spectra of the negatively charged Ag
NPs displayed, in addition, a band at approximately 242 cm 1,
possibly assigned to the Ag–O band observed for synthesized and
as-received commercial negatively charged Ag NPs in water
(Fig. 3). As the DTAC counter ion was chloride, the band may also
be attributed to AgCl, which has been reported previously [40].
To conclude, the initial driving force for adsorption was the
electrostatic attraction between the positively charged DTAC and
the negatively charged commercial and synthesized Ag NPs. This
was followed by hydrophobic interactions between surfactant tails
in a co-operative process resulting in a charge reversal and bilayer
formation for particles exposed to 5 mM DTAC concentrations.
3.4. Surface interactions with the nonionic surfactant alcohol
ethoxylate Berol
Similar to the investigation with LAS and DTAC, two different
concentrations of Berol (CMC 75 lM) were investigated, one above
CMC (5 mM) and one below (5 lM).
When comparing the particle size distribution and measured
zeta potentials upon interaction between Berol and the pure Ag
NPs, a strong neutralizing effect was evident on the zeta potentials
at 5 mM concentrations, an effect considerably weaker at 5 lM
(Table 4). Thus an interaction with the Ag NPs and charge screen-
ing caused by Berol was evident. The 5 mM Berol concentration is
very high, almost a factor of 10 times the CMC value. Adsorption of
Berol as large aggregates, with the large bulky head groups
(ethoxyl groups) may move the slipping plane (plane of charge)
out from the particle surface and thereby lower the measured zeta
potential, shown in Table 4. Note that Berol in itself did not give
rise to a measurable zeta potential value (data not shown), an
expected result since it is a nonionic surfactant.
Raman spectra for the negatively charged commercial Ag NPs
and the positively charged synthesized Ag NPs mixed with Berol
exhibited no bands, except from bulk water and solution species
of Berol. These spectra are hence not included in Fig. 6. However,
possible overlap between bands related to Berol and the adsorbed
CTAB in, for example, the CH-stretching region hindered the inter-
pretation for the positively charged Ag NPs. The negatively charged
Ag NPs on the other hand exhibited weak Raman bands from Berol.
For example, as shown in the inset in Fig. 6, weak bands in the
Dodecyltrimethylammoniumchloride (DTAC)
0.1 mM 5 mM
7 10 7 10
+5 ± 1 10 ± 2 +59 ± 1 +65 ± 1
17 ± 1 22 ± 1 +38 ± 1 +37 ± 1
+33 ± 1 +26 ± 1 +28 ± 1 +31 ± 2
 J. Hedberg et al. / Journal of Colloid and Interface Science 369 (2012) 193–201 199

Fig. 4. Raman spectra of positively charged Ag NPs in interaction with 0.1 mM or 5 mM LAS at pH 7 (left) and pH 10 (right). Pure water, 0.1 mM LAS, or 5 mM LAS Raman
spectra were subtracted for the corresponding spectra with Ag NPs added. Asterisks indicate interference from bands originating from the buffers used. Spectra are offset for
clarity.

Fig. 5. Raman spectra of commercial (right) and negatively (left) charged silver NPs interacting with 0.1 mM or 5 mM DTAC. Water spectra, 0.1 mM DTAC, or 5 mM DTAC bulk
spectra were subtracted from the corresponding spectra with Ag NPs added. Spectra are offset for clarity.

Table 4
Zeta potential measurements of commercial and synthesized Ag NPs in pure water with and without Berol of different concentration at pH 10 and pH 7. The reported error is the
standard deviation of three replicate measurements on the same sample.

Zeta potential (mV) Ultrapure water Alcohol ethoxylate (Berol)

Concentration 5 lM 5 mM
pH 7 10 7 10 7 10
Ag NPs – commercial 24 ± 2 45 ± 1 24 ± 1 44 ± 1 6±1 11 ± 2
Ag NPs – negative 46 ± 1 58 ± 1 47 ± 1 56 ± 1 9±1 7 ± 0.2
Ag NPs – positive +33 ± 3 +29 ± 1 +26 ± 2 +29 ± 3 +8 ± 0.4 +4 ± 0.4

CH-region were evident upon addition of 5 mM Berol. Several rea-
sons could explain these low Raman intensities. A tentative conclu-
sion is a weak interaction between Berol and the surface of the Ag
NPs. Such an interaction would in general result in a higher inten-
sity (such as the Ag–O band observed at 242 cm 1) due to chemical
SERS effect [34,39]. However, the change in zeta potential also
supports the Berol–surface interaction. As elucidated in Fig. 6, evi-
dently the Ag–O band at 242 cm 1 persists as Berol is added to the
negatively charged Ag NPs. This weak surface interaction was fur-
ther probed by means of QCM-D, and the results are shown in
Fig. 7. At pH 10, Berol surfactants (5 mM) were adsorbed onto
the silver QCM-D substrate giving a final adsorbed mass of approx-
imately 2.5 mg/m2. The maximum surfactant layer thickness is
then 25 Å assuming that the layer density is the same as the
200 J. Hedberg et al. / Journal of Colloid and Interface Science 369 (2012) 193–201
Fig. 6. Raman spectra of negatively charged Ag NPs interacting with 5 mM Berol in
water of pH 7 and pH 10, respectively.
Fig. 7. QCM-D mass changes for a silver QCM crystal surface upon addition of 5 mM
Berol at pH 10 followed by rinsing with water.
density of water (1000 kg/m3). The length of a linear alkyl chain of
9–11 carbon atoms (information provided for Berol 266 by the sup-
plier) is according to Tanford, in the range 13–15 Å [16]. Taking
into consideration that the length of the 4–7 ethylene oxide groups
of Berol is not included in this calculation, it is obvious that the
surfactant layer thickness on an evaporated silver surface is far less
than a bilayer. In addition, it is clear that the interaction between
silver and Berol is weak given that rinsing with water for 5 min
resulted in a complete removal of Berol.
The presented results show that a non-electrostatic type of
interaction, such as hydrophobic interactions, occurs between the
nonionic surfactant and the differently charged Ag NPs. Moreover,
the neutralizing effect on the zeta potential, which increased for
concentrations above the CMC, generally agrees with the notion
that adsorption of the nonionic surfactant is drastically increased
for concentrations close to the CMC [16]. The interaction between
the CTAB-covered Ag NPs and Berol is in line with what has been
observed on hydrophilic silica, where mixtures of CTAB and non-
ionic C12E6 yielded mixed, fragmented, bilayers [29]. To conclude,
it is possible for Berol to interact with Ag NPs in a washing cycle,
for example for particles covered by hydrophobic layers such as
biomolecules and dirt, and also with charged particles without
pre-adsorbed molecules. As previously concluded, this emphasizes
the attraction between different types of Ag NPs and laundry
surfactants, and the fact that potential Ag NPs that exit the laundry
cycle is likely to have laundry surfactants adsorbed onto their sur-
faces. The transition from pH 10 to pH 7 did not result in any major
changes in the Ag NPs characteristics.
4. Concluding remark
The interaction between surfactants relevant to the laundry cy-
cle and three differently charged silver nanoparticles (Ag NPs) was
studied to increase the understanding of Ag NP speciation, essen-
tial for environmental fate assessments. Nanoparticles are, at
times, assumed to be released from consumer products without
transformation during their subsequent transportation and accu-
mulation in the environment. This is probably not the case and
recent reviews highlight the importance, and lack of information,
regarding how the adsorbed species on the NP surface, rate of
ion release, and transport route affect environmental availability
and accumulation [41,42]. This study shows that Ag NPs readily
and reversibly interact with laundry surfactants depending on
the surfactant concentration and surface chemistry of the Ag
NP’s. The implication of these results is that Ag NP’s change speci-
ation during washing and upon dilution when exiting the laundry
cycle, before ending up in the waste water treatment plant. In
every transient stage, the transport and retardation processes will
depend on the speciation of the Ag NPs.
Both the synthesized and commercial Ag NPs used in this study
were stable in solution because of their high negative zeta poten-
tial values above pH 3, due to the formation of deprotonated
Ag–O groups as observed by Raman spectroscopy. These negatively
charged particles interacted strongly by electrostatic interactions
with positively charged surfactants (CTAB and DTAC). Hydrophobic
interactions between surfactant tails resulted in bilayer formation
through cooperative adsorption, which caused net charge reversal.
A similar bilayer formation of unordered structure was also
proposed for LAS at a surfactant concentration above CMC onto
modified Ag NPs with pre-adsorbed CTAB. The cooperative adsorp-
tion of surfactants on silver is similar to previous findings of
adsorption of nonionic and cationic surfactants at hydrophilic silica
[29]. This suggests that the negatively charged Ag–O covered silver
surface (IEP at pH 3) and the negatively charged Si–O covered silica
surface (IEP at pH 2) [43] behave similarly; perhaps because of the
negatively charge and oxygen rich structure of both surfaces. The
results further showed a weak interaction between the nonionic
surfactant Berol and both negatively and positively charged Ag
NPs through hydrophobic interactions. This resulted in more
neutral zeta potentials, which would be expected to reduce their
stability and environmental mobility.
This study shows that adsorption of surfactants on the silver sur-
face is an important process during washing of textiles containing Ag
NP’s. One environmental implication of these results is that surfac-
tant covered Ag NPs released from textiles in the laundry cycle could
potentially become more environmentally hazardous, as adsorption
of surfactants on Ag NPs has been reported previously to be more
toxic toward Escherichia coli bacteria compared to unmodified Ag
NPs [8]. However, it has also been shown that the release of silver
ions, thought to be the toxic species, is suppressed in the presence
of some surfactants [6,44]. Hence, there is a substantial knowledge
gap in the role of surfactants on the toxicity of Ag NPs and an imme-
diate need for data combining specific exposure scenarios and
temporal speciation measurements to fully assess potential envi-
ronmental risk of using Ag NPs in consumer products.
Acknowledgments
The authors acknowledge Associate Professor Eric Tyrode, Div.
of Surface and Corrosion Science, KTH, for the use of the Raman
spectrometer.
 J. Hedberg et al. / Journal of Colloid and Interface Science 369 (2012) 193–201
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.jcis.2011.12.004.
References
[1] S.A. Blaser, M. Scheringer, M. MacLeod, K. Hungerbühler, Science of the Total
Environment 390 (2008) 396–409.
[2] Y. Ju-Nam, J.R. Lead, Science of the Total Environment 400 (2008) 396–414.
[3] P.V. AshaRani, G. Low Kah Mun, M.P. Hande, S. Valiyaveettil, ACS Nano 3 (2008)
279–290.
[4] Q. Li, S. Mahendra, D.Y. Lyon, L. Brunet, M.V. Liga, D. Li, P.J.J. Alvarez, Water
Research 42 (2008) 4591–4602.
[5] J.S. Kim, E. Kuk, K.N. Yu, J.-H. Kim, S.J. Park, H.J. Lee, S.H. Kim, Y.K. Park, Y.H.
Park, C.-Y. Hwang, Y.-K. Kim, Y.-S. Lee, D.H. Jeong, M.-H. Cho, Nanomedicine:
Nanotechnology, Biology and Medicine 3 (2007) 95–101.
[6] J. Liu, R.H. Hurt, Environmental Science and Technology 44 (2010) 2169–2175.
[7] J. Liu, D.A. Sonshine, S. Shervani, R.H. Hurt, ACS Nano 4 (2010) 6903–6913.
[8] E. Bae, H.J. Park, J. Park, J. Yoon, Y. Kim, K. Choi, J. Yi, Bulletin of the Korean
Chemical Society 32 (2011) 613–619.
[9] T.M. Benn, P. Westerhoff, Environmental Science and Technology 42 (2008)
4133–4139.
[10] E. Smulders, Laundry Detergents, John Wiley & Sons, Weinheim, 2007.
[11] Y.-H. Chen, C.-S. Yeh, Colloids and Surfaces A: Physicochemical and
Engineering Aspects 197 (2002) 133–139.
[12] L. Fan, R. Guo, Crystal Growth and Design 8 (2008) 2150–2156.
[13] L. Kvitek, A. PanacÌŒek, J. Soukupova, M. Kolar, R. VecÌŒerova, R. Prucek, M.
Holecova, R. Zboril, The Journal of Physical Chemistry C 112 (2008) 5825.
[14] R.P. Bagwe, K.C. Khilar, Langmuir 16 (1999) 905–910.
[15] S. Kittler, C. Greulich, J. Diendorf, M. Köller, M. Epple, Chemistry of Materials
22 (2010) 4548.
[16] F. Tiberg, J. Brinck, L. Grant, Current Opinion in Colloid and Interface Science 4
(1999) 411–419.
[17] R.n.A. Alvarez-Puebla, E. Arceo, P.J.G. Goulet, J.n.J. Garrido, R.F. Aroca, The
Journal of Physical Chemistry B 109 (2005) 3787–3792.
[18] K. Kneipp, H. Kneipp, I. Itzkan, R.R. Dasari, M.S. Feld, Journal of Physics-
Condensed Matter 14 (2002) R597–R624.
[19] T.M. Tolaymat, A.M. El Badawy, A. Genaidy, K.G. Scheckel, T.P. Luxton, M.
Suidan, Science of the Total Environment 408 (2010) 999–1006.
201
[20] J.A. Creighton, C.G. Blatchford, M.G. Albrecht, Faraday Transactions 2:
Molecular and Chemical Physics 75 (1979) 790–798.
[21] Z.M. Sui, X. Chen, L.Y. Wang, Y.C. Chai, C.J. Yang, J.K. Zhao, Chemistry Letters 34
(2005) 100–101.
[22] Z.M. Sui, X. Chen, L.Y. Wang, L.M. Xu, W.C. Zhuang, Y.C. Chai, C.J. Yang, Physica
E-Low-Dimensional Systems and Nanostructures 33 (2006) 308–314.
[23] E. Tyrode, M.W. Rutland, C.D. Bain, Journal of the American Chemical Society
130 (2008) 17434–17445.
[24] G. Sauerbrey, Zeitschrift für Physik 155 (1959) 206–222.
[25] S. Elzey, V.H. Grassian, Journal of Nanoparticle Research 12 (2009) 1945–1958.
[26] K. Kalyanasundaram, J.K. Thomas, The Journal of Physical Chemistry 80 (1976)
1462–1473.
[27] E. Koglin, A. Tarazona, S. Kreisig, M.J. Schwuger, Colloids and Surfaces A:
Physicochemical and Engineering Aspects 123-124 (1997) 523–542.
[28] R. Foucault, R.L. Birke, J.R. Lombardi, Langmuir 19 (2003) 8818–8827.
[29] J. Penfold, E.J. Staples, I. Tucker, R.K. Thomas, Langmuir 16 (2000) 8879–8883.
[30] H. Matsuta, K. Hirokawa, Applied Spectroscopy 43 (1989) 239–245.
[31] T.H. Wood, M.V. Klein, D.A. Zwemer, Surface Science 107 (1981) 625–635.
[32] T.M. Herne, R.L. Garrell, Analytical Chemistry 63 (1991) 2290–2294.
[33] Z.P. Xu, P.S. Braterman, Journal of Materials Chemistry 13 (2003) 268–273.
[34] A. Otto, I. Mrozek, H. Grabhorn, W. Akemann, Journal of Physics-Condensed
Matter 4 (1992) 1143–1212.
[35] L. Zhai, X. Lu, W. Chen, C. Hu, L. Zheng, Colloids and Surfaces A:
Physicochemical and Engineering Aspects 236 (2004) 1–5.
[36] E.W. Kaler, A.K. Murthy, B.E. Rodriguez, J.A.N. Zasadzinski, Science 245 (1989)
1371–1374.
[37] L.L. Schramm, E.N. Stasium, D.G. Marangoni, Rep. Prog. Chem. 99 (2003) 3–48.
[38] G.J. Kovacs, R.O. Loutfy, P.S. Vincett, C. Jennings, R. Arcoa, Langmuir 2 (1986)
689–694.
[39] I. Mrozek, A. Otto, Journal of Electron Spectroscopy and Related Phenomenon
54-55 (1990) 895–911.
[40] A. Otto, A. Bruckbauer, Y.X. Chen, Journal of Molecular Structure 661-662
(2003) 501–514.
[41] E. Navarro, A. Baun, R. Behra, N.B. Hartmann, J. Filser, A.-J. Miao, A. Quigg, P.H.
Santschi, L. Sigg, Ecotoxicology 17 (2008) 372–386.
[42] J. Fabrega, S.N. Luoma, C.R. Tyler, T.S. Galloway, J.R. Lead, Environment
International 37 (2011) 517–531.
[43] G.H. Bolt, The Journal of Physical Chemistry 61 (1957) 1166–1169.
[44] L. Geranio, N. Heuberger, B. Nowack, Environmental Science and Technology
43 (2009) 8113–8118.