Chemical Physics Letters 675 (2017) 124–130

Contents lists available at ScienceDirect

Chemical Physics Letters

journal homepage: www.elsevier.com/locate/cplett

Research paper

Structure and interaction in pathway of charged nanoparticles

aggregation in saline water as probed by scattering techniques
Sohrab Abbas a, Indresh Yadav a, Sugam Kumar a, Vinod K. Aswal a,⇑, Joachim Kohlbrecher b
a
Solid State Physics Division, Bhabha Atomic Research Centre, Mumbai 400 085, India
b
Laboratory for Neutron Scattering, Paul Scherrer Institut, CH-5232 PSI Villigen, Switzerland

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

Article history: The evolution of interaction and structure during the aggregation of anionic silica nanoparticles in saline
Received 20 January 2017 water has been examined. The net attraction between charged nanoparticles with existing of salt ions is
In final form 8 March 2017 found to be significantly higher than that accounted by van der Waals force of the DLVO theory. The
Available online 10 March 2017
increase in the net attraction with salt leads to systematic transformation of nanoparticles into dimers,
tetramers and higher-mers which finally results into fractal aggregate. Such multi-step aggregation is
consistent with diffusion limited cluster aggregation (DLCA) fractal morphology but the fractal dimension
(
2.6) obtained is much larger than usually comprehended by DLCA.
Ó 2017 Elsevier B.V. All rights reserved.

1. Introduction between long-range electrostatic repulsion and long-range attrac-

tion arising from particle ion clouding around the protein mole-
The quest to understand the stability of colloidal suspension, its cules [12]. On the other hand, the clouding phenomenon is
aggregation and the underlying steps leading to aggregation gener- observed in various charged and mixed micelles in presence of dif-
ate great scientific interest because of its multitude applications in ferent hydrophobic counterions [11]. The nanoparticles form an
the fields ranging from molecular biology to food industry [1,2]. In ideal system to understand these attractive and repulsive interac-
general, the stabilization of colloids is controlled by the interplay of tions leading to particle clustering. Moreover, these particle clus-
repulsive interactions (e.g. electrostatic or steric repulsion) against ters being ubiquitous phenomenon among all nanoparticles,
attractive forces (e.g. van der Waals or depletion) [3,4]. For charged influence mediation of the physiochemical properties like reactiv-
colloids, its stability is very well explained by Derjaguin-Landau- ity, environmental fate, transport, and biological interactions (e.g.,
Verwey-Overbeek (DLVO) theory which combines short range bioavailability and uptake) of nanoparticles [7,16]. Therefore, it has
attractive van der Waals force with long range repulsion caused always been of interest to decipher the mechanism of the aggrega-
by the overlap of electric double layers between interacting parti- tion and to provide the model responsible for interactions leading
cles [4,5]. Aggregation of colloidal particles occurs when physical to the aggregation.
processes bring particle surfaces in contact with each other and There have been a large number of studies on the aggregation of
thermodynamic stability requires the particle-particle attachment charged nanoparticles through various means such as addition of
[6,7]. This can be achieved by modifications in interparticle inter- electrolyte [9,13,14] and/or oppositely charged macromolecule
actions through addition of selective additives leading to particle [10] or through depletion mechanism [17]. Many of these studies
aggregation [8–11]. These variations in interactions and hence show a systematic evolution of attraction leading to phase separa-
the colloidal structure may either be in accordance with DLVO the- tion of particle in the form of large aggregates [8,14,15]. In partic-
ory or may require involvement of non-DLVO forces. ular, electrolyte induced cluster phase in a variety of nanoparticles
Recently, the colloidal aggregation driven by non-DLVO interac- has been extensively studied for the various disordered structure
tions has been investigated in many systems like proteins, micelles formed, process kinetics and effect of different system parameters
and nanoparticles [10–15]. In all the cases, the aggregation has on their growth and morphology. Various experimental techniques
been explained in terms of short/long-range attractive and repul- including dynamic and static light scattering and different simula-
sive interactions present in the systems. For example, the emer- tion methodologies were used to investigate the aggregates [18–
gence of finite sized protein clusters is attributed to competition 20]. However, most of these studies lack in providing information
on the evolution of interaction at each step of these aggregation
from initiation of attraction to final phase separation and in quan-
⇑ Corresponding author.
tifying the interaction parameters. The understanding of these
E-mail address: vkaswal@barc.gov.in (V.K. Aswal).

http://dx.doi.org/10.1016/j.cplett.2017.03.024
0009-2614/Ó 2017 Elsevier B.V. All rights reserved.
 S. Abbas et al. / Chemical Physics Letters 675 (2017) 124–130 125

steps provides useful inputs in predicting the phase behavior of the S(Q) depends on the interparticle interactions. The interaction
nanoparticles as well as in obtaining equilibrium cluster phase between the particles may be a combination of attractive and
with unique features. repulsive forces. As per the DLVO theory, the total potential
In this work, we have examined the systematic transition of sil- between two charged spherical particles is the sum of van der
ica nanoparticles from individual to micron sized aggregates on Waals attraction (UA) and electrostatic double layer repulsion
addition of salt. The silica nanoparticles are used as model system (UR), which can be expressed as [4,5,32]
due to their high stability even in the conditions where DLVO the-
ory predicts coagulation [14,21]. The systems are examined using a U ¼ UA þ UR ð2Þ
combination of small-angle neutron scattering (SANS) and
dynamic light scattering (DLS). SANS is used as it provides informa- 

A 1 1 1
tion on both structure as well as interaction through form factor UA ¼  þ þ 2 ln 1  ð3Þ
12 x2  1 x2 x2
and structure factor of scattering intensity, respectively [22]. DLS
is used as a complementary technique since it measures the diffu-
sion of the particles which is also related to the structure and inter- q2 e2 exp½jðx  1Þ
UR ¼ ð4Þ
action present in the system [23]. The evolution of interaction in 4pee0 rð1 þ jr=2Þ2 x
the SANS data have been modeled using a two-Yukawa potential
accounting for both attractive and repulsive interactions present where x (>1) is the ratio of center-to-center distance (r) of particles
in the system [15,24,25]. The choice of such potential can establish to the diameter of particle (r). A is the Hamaker constant, q is the
the range and strength of the individual parts of the total potential charge on particle, e0 is the permittivity of the free space, e is the
without any predefined assumption [15]. dielectric constant of medium and j is the inverse Debye length.
For non-DLVO interactions, the total interaction potential can
be described by the following two-Yukawa potential accounting
2. Experimental details
for attractive (characterized by parameters K1 and Z1) and repul-
sive (characterized by parameters K2 and Z2) interactions present
Electrostatically stabilized colloidal suspensions of silica
in the system [12,24]
nanoparticles (Ludox LS30) was purchased from Sigma-Aldrich,
USA. Samples were prepared by dissolving 1 wt% amount of silica expfZ 1 ðx  1Þg expfZ 2 ðx  1Þg
nanoparticles with varying amount of NaCl (0–2 M) in H2O solvent U ¼ K 1 þ K2 ð5Þ
x x
for better contrast. Small-angle neutron scattering experiments
were performed at SANS-I facility, Swiss Spallation Neutron Source where K (in units of kBT, kB is Boltzmann constant and T is absolute
SINQ, Paul Scherrer Institut, Switzerland [26]. The wavelength (k) temperature) is proportional to magnitude of the potential while 1/
of monochromatic neutron beam was 0.6 nm with a nominal Z is proportional to the range of the potential. The two-Yukawa
spread (Dk/k) of 10%. Data were collected at three sample-to- potential is different from DLVO as it can take account of different
detector distances 2, 8 and 20 m to cover a wide wave vector trans- attractions than van der Waals interaction. The repulsive part
fer [Q = (4psin h)/k; 2h is scattering angle] range of 0.02–3 nm1. remains same as governed by the screened Coulomb potential for
The scattered neutrons were detected by a large (96  96 cm2) charged particles. The effective charge (q) on the particle and Debye
3
He based, 2d-position sensitive neutron detector. Corrections length (j-1) are related to K2 and Z2 by
were made for background and empty cell contributions and data
were normalized to absolute cross-sectional unit using standard q2 e 2
procedure. DLS measurements were performed on nanoparticle ¼ K 2 kB T ð6Þ
4pee0 rð1 þ jr=2Þ2
size analyzer SZ-100 (Horiba, Japan). The wavelength of the inci-
dent laser light was 532 nm and the data were collected at scatter-
r
ing angle 173°. It may be mentioned that all the samples were j1 ¼ ð7Þ
equilibrated for one day after preparation. These samples did not Z2
show any significant changes observed during the period of mea- The interparticle structure factor S(Q) for two-Yukawa potential
surements (about 1–3 h for SANS). is calculated numerically by solving the Ornstein-Zernike (OZ)
equation with a Mean Spherical Approximation closure relation.
The details about the calculation of this S(Q) may be found else-
3. Data analysis
where [24].
For particle aggregation modeled by a mass fractal, S(Q) is given
3.1. Small-angle neutron scattering
by [33]

In SANS experiments, one measures the coherent differential 1 Df CðDf  1Þ

scattering cross-section per unit volume as a function of scattering SðQ Þ ¼ 1 þ h i sinfðDf  1Þ
ðQRb ÞDf 1 þ 1 ðDf 1Þ=2
vector (Q) which can be given by [22,27] ðQnÞ2

dR  tan1 ðQ nÞg ð8Þ

ðQ Þ ¼ nV 2 ðqp  qs Þ2 PðQ ÞSðQ Þ þ B ð1Þ
dX
where n signifies the maximum length up to which fractal
where n is the number density of particles and V is particle volume. microstructure exists, Rb is the size of building block and Df denotes
qp and qs are scattering length densities of particle and solvent, the mass fractal dimension and U is mathematical Gamma function.
respectively. P(Q) signifies intraparticle structure factor (square of Throughout the data analysis, corrections were made for instru-
form factor) and S(Q) denotes interparticle structure factor. B is a mental smearing. The calculated scattering profiles for the suitable
constant term representing incoherent background. P(Q) is decided models of interest were smeared by the appropriate resolution
by the shape and size of the particle and the expressions of P(Q) for function and compared with the measured data. The parameters
different shapes as well as aggregates (dimmers, tetramers) can be in the analysis were optimized by means of nonlinear least-
found elsewhere [27–31]. square fitting programs [27].
126 S. Abbas et al. / Chemical Physics Letters 675 (2017) 124–130

3.2. Dynamic light scattering (Q) as calculated by two-Yukawa potential under Mean Spherical
Approximation [24]. The fitted interaction potential is displayed
In DLS, the intensity autocorrelation function [g2(s)] of the scat- in inset of Fig. 1. The potential has four unknown parameters K1,
tered light is measured at a fixed scattering angle as a function of K2, Z1 and Z2. The parameters K1 and Z1 represents the strength
delay time (s). For an ergodic system, it is related to the field auto- (proportional to K1) and range (varies as 1/Z1), respectively of the
correlation function [g1(s)] as [23,34] attractive potential. For pure silica nanoparticle system, these
parameters are fixed to that corresponding to van der Waals attrac-
g 2 ðsÞ ¼ 1 þ bjg 1 ðsÞj2 ð9Þ tion between particles as calculated using Hamaker constant
where b (0 < b < 1) is the spatial coherence factor and depends on (0.5  1020 J) [35,36]. On the other hand, K2 and Z2 are the param-
the instrument optics. In case of a monodisperse system of particles, eters of the repulsive part of the potential providing the strength
g1(s) decays exponentially and for a polydisperse system it can be (related to effective charge) and range (related to ionic strength),
written as respectively [37]. The effective charge has been thus calculated
Z from K2 and Z2 using Eq. (7). The magnitude (K2) and the range
1
(1/Z2) of repulsive part are found to be much larger than that of
g 1 ðsÞ ¼ FðCÞ expðCsÞdC ð10Þ
0 attractive part in the total interaction potential, providing stability
to the nanoparticle system. The mean size of silica nanoparticles is
where U (=DQ2) is the decay constant for a given size, which
found to be 15.4 nm with a polydispersity (standard deviation) of
depends on the diffusion coefficient (D) of the particles and the

0.17 for log-normal size distribution (Table 1).
magnitude of wave vector transfer (Q). F(U) is the weight factor in The silica nanoparticles being charged attain long term stability
the decay constant distribution representing the particles distribu- as per DLVO theory where electrostatic repulsion overcomes the
tion with different diffusion coefficients relative to the mean value. van der Waals attraction. The presence of salt ions screens the
For a narrow monomodal distribution, the cumulant analysis repulsion between charged nanoparticles giving rise to a net
yields the mean value of diffusion coefficient (Dm) and polydisper- attraction in the DLVO potential which finally leads to nanoparticle
sity index (PI) according to following equation [34] aggregation at sufficient salt concentration [4,5]. However, the
 
l s2 transition from monomer nanoparticles under attractive interac-
g 1 ðsÞ ¼ exp Dm Q 2 s þ 2 ð11Þ tion to nanoparticle aggregates is not a single step process. Some
2
of the studies report about trapping the particles in the secondary
where l2 is variance. The PI is calculated by the ratio of variance minima known as particle crowding leading to aggregation
(l2) to the square of mean of the decay rate (Cm = DmQ2). The whereas some of the other studies describe it as a systematic pro-
Stokes-Einstein relation then correlates the hydrodynamic radius cess of formation of different sized particle aggregates [32,38]. In
of particles to the diffusion coefficient. The hydrodynamic size order to understand the evolution of interaction at each step of
could be very different from actual size depending on hydration the underlying pathway of aggregation, we have carried out SANS
and/or interaction between the particles. measurements on 1 wt% silica nanoparticles with varying amount
of NaCl in H2O as shown in Fig. 2. Based on the features, the SANS
4. Results and discussion data have been divided into three parts: (i) the first data set [Fig. 2
(a)] depict gradual increase in intensity at low Q for smaller
The SANS data of 1 wt% (volume fraction
0.0045) pure silica amount of salt addition with almost no change in intermediate
nanoparticle solution in H2O is displayed in Fig. 1. H2O is used as and high Q range, (ii) the second data set [Fig. 2(b)] exhibits the
solvent in spite of higher scattering background due to the better large scattering at low Q values with significant changes at inter-
contrast for silica nanoparticles in H2O compare to D2O. The ratio mediate Q values and (iii) in the third data set [Fig. 2(c)] the large
of contrast factor [(qp  qs)2] for silica nanoparticles in H2O and linear increase in scattered intensity (log-log scale) without any
D2O is about 2.9 [10]. The nanoparticle system shows a correlation low Q cut-off, is observed. The rise in the low Q intensity in the first
peak at about (Q = 0.1 nm1) indicating the presence of repulsive data set [Fig. 2(a)] clearly indicates the evolution of the net attrac-
interaction between nanoparticles. Such peaks usually appears at tion whereas in the third set [Fig. 2(c)], the linearity suggests the
Q = 2p/d, where d signifies the interparticle separation [10]. The formation of large size fractal aggregates [10,33]. However the
data is fitted employing a P(Q) for polydispersed spheres and S steps in between these two extremes from present observations
cannot be predicted.
To ascertain the nature of system evolution with varying salt
concentration, DLS measurements were also carried out. The corre-
sponding intensity auto correlation functions are shown in Fig. 3
(a). As can be seen in Fig. 3(a), the autocorrelation function is sys-
tematically modified (slower decay of correlations in time) with
the increasing salt concentration as approaching from an individ-
ual particle to aggregated structures. The variations imply decrease
in the effective decay rate of the autocorrelation function and
hence the decrease in the effective diffusion coefficient with salt
concentration [23]. This may be because of increase in attractive
interaction between nanoparticles and/or due to formation of large
size nanoparticle aggregates [34]. In order to confirm the same,
DLS measurements were further performed on the system after
10 times dilution [Fig. 3(b)]. The dilution is supposed to suppress
the apparent decrease in effective decay rate as caused by attrac-
tive interactions (if any). The auto correlation function after dilu-
tion remains same to that of nanoparticle system (1 wt% LS30
Fig. 1. SANS data of 1 wt% silica nanoparticles in H2O. Inset shows the fitted + 0.1 M NaCl) up to 0.5 M of the salt concentration indicating that
interaction potentials. modifications in the correlation function [Fig. 3(a)] up to this
 S. Abbas et al. / Chemical Physics Letters 675 (2017) 124–130 127

Table 1
Structure and interaction parameters of 1 wt% silica nanoparticles in H2O. K (in units of kBT, kB is Boltzmann constant and T is absolute temperature) and 1/Z are proportional to
magnitude and range of the potential, respectively.

Size (nm) Polydispersity r Attractive potential Repulsive potential Charge q (e.u.)

K1 (kBT) Z1 K2 (kBT) Z2
15.4 ± 0.8 0.17 ± 0.03 5.0 ± 0.5 30.0 ± 2.5 8.0 ± 1.0 1.0 ± 0.2 40

Fig. 3. DLS data of 1 wt% silica nanoparticles mixed with varying NaCl concentra-
tions (a) without dilution (b) after 10 times dilution. Inset of (b) shows the
comparison of effective hydrodynamic size before and after dilution.

concentration arises solely because of evolution of net attraction.

However, above this concentration (0.5 M NaCl), effectively slow
relaxation rates were observed in both the cases (with and without
dilution) indicating the formation of larger size aggregates. The
variation of calculated effective hydrodynamic sizes is shown in
the inset of Fig. 3(b). It may be mentioned that the viscosity and
refractive index increase on increasing the salt concentration
[39,40]. The corresponding values of viscosity and refractive index
have been taken into account while calculating the effective hydro-
dynamic size. The effective hydrodynamic size in the salt concen-
tration range from 0.6 to 1.2 M NaCl is slightly less than that
observed at same salt concentration without dilution indicating
the presence of the attraction as well as aggregation in the system.
Finally, corresponding to the salt concentration in the third data
set [inset of Fig. 3(b)], the effective hydrodynamic size becomes
independent of salt concentration as well as dilution, suggesting
the formation of large aggregates of micron size.
Based on understandings of DLS data, the first SANS data set
Fig. 2. SANS data of 1 wt% silica nanoparticles mixed with varying NaCl concen-
[Fig. 2(a)] is fitted by a structure factor as calculated from two-
tration in the range (a) 0.0–0.5 M, (b) 0.6–1.2 M and (c) 1.4–2.0 M.
128 S. Abbas et al. / Chemical Physics Letters 675 (2017) 124–130

Yukawa potential. The S(Q) as obtained by dividing the scattered Table 2

intensity by form factor P(Q), are plotted in Fig. 4(a). The P(Q) is Fitted parameters corresponding to SANS data of 1 wt% silica LS30 nanoparticles
(size = 15.4 nm, polydispersity = 0.17) with varying salt (NaCl) concentration. The
obtained from the data of the 1 wt% LS30 in presence of 0.1 M NaCl, interaction between nanoparticles has been fitted using a two-Yukawa potential
where the scattering is governed only by form factor and S(Q)
1. where K (in units of kBT, kB is Boltzmann constant and T is absolute temperature) is
This is confirmed by measuring the SANS intensity of diluted sam- proportional to magnitude of the potential while 1/Z is proportional to the range of
ples and scaling the data with that of original sample (1 wt% LS30 the potential.
+ 0.1 M NaCl). The S(Q = 0) and hence scattering intensity I(Q) at Salt conc. (M) Attractive potential Repulsive potential
Q ? 0 is governed by isothermal compressibility of the system.
K1 Z1 K2 Z2
The diverging nature of S(Q) in the low Q region and, in particular,
(a) Interaction parameters in the low salt concentration range from 0.0 to 0.5 M
the increase in the value of S(Q
0) clearly suggests enhancement
corresponding to Fig. 2(a), where nanoparticles experiences evolution of
in the net attraction with the increasing salt concentration [15]. attractive interaction
The fitted parameters are listed in Table 2(a). The corresponding 0 5.0 ± 0.5 30.0 ± 2.5 8.0 ± 1.0 1.0 ± 0.2
fitted total interaction potentials and their respective repulsive 0.1 5.0 ± 0.5 30.0 ± 2.5 3.0 ± 0.3 12.2 ± 1.0
0.2 1.4 ± 0.2 5.0 ± 0.5 1.0 ± 0.1 16.1 ± 1.5
and attractive parts are shown in Fig. 4(b)–(d), respectively.
0.3 4.9 ± 0.5 4.6 ± 0.4 0.13 ± 0.03 22.8 ± 2.0
According to the DLVO theory the aggregation is driven by the 0.4 8.0 ± 1.0 3.6 ± 0.4 0.08 ± 0.02 32.3 ± 2.5
reduction in the repulsive part of the total interaction potential 0.5 11.0 ± 1.0 3.0 ± 0.3 0.06 ± 0.01 35.0 ± 3.0
causing an effective attraction in the system due to van der Waals
attraction. This salt induced reduction in electrostatic repulsion Salt conc. (M) Attractive potential Structures formed
[Fig. 4(c)] has been taken into account during data analysis. At K1 Z1
higher salt concentrations, where the repulsion becomes negligibly (b) Interaction parameters in the intermediate salt concentration range from 0.6
small [Fig. 4(c)], the DLVO potential is expected to be mostly gov- to 1.2 M corresponding to Fig. 2(b), where there is a systematic evolution of
erned by van der Waals interaction. For particle aggregation to dimers, tetramers and higher-mers
takes place, the attraction at an average distance between the par- 0.6 21.4 ± 1.8 3.6 ± 0.3 Dimers
0.7 29.4 ± 2.5 4.0 ± 0.4 Dimers
ticles should be more than the average thermal kinetic energy 0.8 42.4 ± 4.0 4.0 ± 0.4 Tetramers
(1.5 kBT). This condition is not fulfilled by the van der Waals attrac- 1.0 43.0 ± 4.0 3.7 ± 0.3 Tetramers
tion for the system under study. Also, the van der Waals interaction 1.2 50.0 ± 4.8 3.6 ± 0.3 Tetramers
alone is not enough to account the SANS data, in particular the low Salt conc. (M) Building block radius Rb (nm) Fractal dimension Df
Q rise in the scattering intensity. Therefore, the attraction beyond (c) Structure parameters in the higher salt concentration range from 1.4 to 2.0 M,
DLVO is required to account the data. This is also clear from the where most of the nanoparticles are aggregated
1.4 7.7 ± 0.4 2.6 ± 0.1
inset of Fig. 4(b) where the calculated DLVO potential is compared
1.6 7.7 ± 0.4 2.6 ± 0.1
with the fitted two-Yukawa potential for 1 wt% LS30 + 0.3 M NaCl 1.8 7.7 ± 0.4 2.6 ± 0.1
system. For DLVO potential, the van der Waals attraction is calcu- 2.0 7.7 ± 0.4 2.6 ± 0.1
lated using Hamaker constant (0.5  1020 J) for silica and
screened Coulomb potential is calculated from Debye length and of calculated DLVO potential. Moreover, the van der Waals interac-
charge on nanoparticles [32,35,36]. Both the magnitude and range tion does not have any salt dependence and therefore for DLVO
of the fitted potential are found to be significantly larger than those interaction, there should not be any changes in the interaction

Fig. 4. Variation of (a) structure factor, (b) total interaction potential, (c) repulsive part of the total potential and (d) attractive part of the total potential for 1 wt% LS30 silica
nanoparticles with varying salt concentration in the range 0.0–0.5 M corresponding to first data set [Fig. 2(a)]. The data of the structure factor for different salt concentrations
are shifted vertically for clarity. Inset of (b) shows the differences between DLVO potential and the two-Yukawa potential for 1 wt% LS30 + 0.3 M NaCl.
 S. Abbas et al. / Chemical Physics Letters 675 (2017) 124–130 129

potential and hence in the scattering data with increase in salt con-
centrations. On the contrary, we observe the significant increase in
the low Q scattering intensity at higher salt concentrations [Fig. 2
(a)] which is accounted by increasing the attractive part of the
two-Yukawa potential. Such dissension of the phase behavior of
silica nanoparticles with DLVO theory has also been reported in
other studies [14,21,41–44]. For example, contrary to the DLVO
predictions, the silica nanoparticles are claimed to be stable at
low pH, where silica has small charge [14]. It has been understood
that the validity of the DLVO theory is restricted to the systems,
where the repulsive potential barrier is located at distances
exceeding a few nanometers. At higher salt concentrations, this
condition is violated and various other interactions like solvation
forces, ion mediated attraction and/or repulsion owing from hairy
layers of poly(silicilic acid) chains on the particle surface become
important [14,42,43]. However, it may be mentioned that such
anomalies have been mainly reported for nanosized silica particles,
while sub-micrometer silica particles seem to behave in accor-
dance to the DLVO.
The SANS data for salt concentrations greater than 0.5 M cannot
be solely understood by considering the attractive interaction
among monomeric silica nanoparticles. Moreover, the DLS data
also suggests the formation of larger size structures above this salt
concentration. Therefore, the data in the second set with salt con-
centration range from 0.6 to 1.2 M [Fig. 2(b)], can be analyzed by
considering two situations to account for the presence of larger
aggregates. In the first case, one may consider the formation of
fractal aggregates, coexisting with individual nanoparticles where
the population of un-aggregated nanoparticles decreases with
increasing salt concentration due to their transformation from
individual to a part of aggregates. This kind of modeling is as per
diffusion limited (DLA) kind of fractal aggregation where the aggre-
gation process is controlled by solely diffusion motion of particles
under very weak repulsive forces [45]. Though DLA model is also
widely accepted for explaining the aggregation process in a variety
Fig. 5. Variation of (a) structure factor and (b) total interaction potential for 1 wt%
of systems [46,47] but in the present case, our DLS results do not
LS30 silica nanoparticles with varying salt concentration in the range 0.6–1.2 M
support this kind of aggregation growth mechanism. It is clear corresponding to second data set [Fig. 2(b)]. The data of the structure factor for
from DLS that the effective size of the aggregates for salt concen- different salt concentrations are shifted vertically for clarity.
trations below 1.0 M is about 100 nm which is significantly less
than the sizes of the final aggregates (micron size) as assumed in
this case. In DLS, the intensity weighted size distribution is highly distribution for the clusters. But in order to simplify the model
dominated by large particles present in the system. Therefore, if and to limit the fitting parameters, only single population for a
there are any such large clusters present in the system from the given salt concentration without any size distribution has been
beginning, these should be reflected in the DLS data and the effec- considered. As can be seen in Fig. 5(b), the magnitude of the attrac-
tive size should be much larger than that has been experimentally tion increases with increasing salt concentration which is respon-
obtained. Therefore, nanoparticle aggregation is considered as sible for the conversion of the nanoparticles into dimers to
multi-step process where monomers converting to dimers and tetramers, etc.
then to tetramers etc. and finally to higher-mers with increasing For salt concentrations more than 1.2 M, the SANS data cannot
salt concentration. This kind of modeling is consistent with that be accounted by attractive S(Q) and the values of magnitude (K1)
governed by diffusion limited cluster-cluster aggregation (DLCA) and range (1/Z1) becomes unacceptably large. Therefore, the SANS
process and is suggested by others also [20,48]. data in the set 3 [Fig. 2(c)] are fitted using a S(Q) of mass fractal as
In order to analyze the data in this case, first S(Q) have been cal- suggested by the linearity in the low Q data in log-log scale. In the
culated by dividing the observed SANS intensities I(Q) with the SANS data from fractal system, usually there exists two Q cut offs
respective intraparticle structure factors P(Q) of dimers, tetramers at low and high Q values corresponding to overall size of the fractal
and are shown in Fig. 5(a). The interparticle structure factors S(Q) and building block size, respectively [33]. Since the low Q cut off is
thus obtained, are fitted by attractive interaction considering neg- not observed in the present Q range of the measurements, the
ligible repulsion between them. The K1 and Z1 are the only fitting overall size of the aggregates (n) is fixed to a value that obtained
parameters whereas r is calculated by considering the scatteres from DLS which is more than 2p/Qmin throughout the data analysis.
as equivalent hard spheres. The fitted parameters are given in The high Q cut off in this case is decided by the nanoparticle radius
Table 2(b) and the corresponding interaction potentials are shown (building block radius). The fitted parameters are listed in Table 2
in Fig. 5(b). The data for salt concentrations (0.6 and 0.7 M) could (c). It is interesting to note that the fractal dimension is found to
be well fitted by dimers under the influence of attractive interac- be about 2.6 irrespective of the salt concentration, though the
tion whereas for salt concentrations 0.8–1.2 M, the analysis shows structure formation is directed by a DLCA kind of aggregation pro-
that increased attraction converts these dimers into tetramers. It cess. The fractal dimension for DLCA driven morphology has been
may further be mentioned that there could be possibility of mixed mostly reported in the range 1.6–1.8. Also for reaction limited
population of monomers, dimers, trimers, etc. and also size cluster-cluster aggregation (RLCA), the fractal dimension is found
130 S. Abbas et al. / Chemical Physics Letters 675 (2017) 124–130

to be varying in the range 2.1–2.4 [20,48], whereas consideration [9] R. Pamies, J.G.H. Cifre, V.F. Espín, M.C. González, F.G.D. Baños, J.G. de la Torre, J.
Nanopart. Res. 16 (2014) 2376.
of restructuring of the DLCA aggregates can result into a fractal
[10] S. Kumar, V.K. Aswal, J. Kohlbrecher, Langmuir 28 (2012) 9288.
dimension of about 2.2 [49]. It has been recently reported by sim- [11] Y. Liu, E. Fratini, P. Baglioni, W.-R. Chen, S.-H. Chen, Phys. Rev. Lett. 95 (2005)
ulations that the fractal dimension in DLCA increases with particle 118102.
concentration and can have values up to 2.8 at higher particle con- [12] T.F. Mohry, A. Maciołek, S. Dietrich, J. Chem. Phys. 136 (2012) 224903.
[13] A.J. Chinchalikar, V.K. Aswal, J. Kohlbrecher, A.G. Wagh, Colloid Polym. Sci. 288
centrations [50]. The fractal dimension of 2.6 in our systems (1 wt% (2010) 1601.
silica nanoparticles corresponding to
1015 particles/cm3) sup- [14] M. Kobayashi, F. Juillerat, P. Galletto, P. Bowen, M. Borkovec, Langmuir 21
ports to these results. Our studies thus provide the understanding (2005) 5761.
[15] S. Kumar, D. Ray, V.K. Aswal, J. Kohlbrecher, Phys. Rev. E 90 (2014) 042316.
of the evolution of interaction in the pathway followed by the [16] A. Albanese, W.C.W. Chan, ACS Nano 5 (2011) 5478.
charged nanoparticles during salt induced aggregation and the [17] D. Banerjee, J. Yang, K.S. Schweizer, Soft Matter 11 (2015) 9086.
resultant morphology. The results also show that the primary par- [18] S.-W. Bian, I.A. Mudunkotuwa, T. Rupasinghe, V.H. Grassian, Langmuir 27
(2011) 6059.
ticle density is a major parameter influencing the fractal dimension [19] P.T. Serrano, C.J. Chin, S. Yiacoumi, C. Tsouris, Curr. Opin. Colloid Interface Sci.
along with the mechanism of the aggregation. 10 (2005) 123.
[20] M. Carpineti, M. Giglio, Phys. Rev. Lett. 70 (1993) 3828.
[21] H.E. Bergna, W.O. Roberts, Colloidal Silica: Fundamentals and Applications,
5. Conclusions Taylor & Francis Group, New York, 2006.
[22] D.I. Svergun, M.H.J. Koch, Rep. Prog. Phys. 66 (2003) 1735.
[23] R. Pecora, Dynamic Light Scattering, Plenum, New York, 1985.
The systematic evolution of interaction and structure in the [24] Y. Liu, W.-R. Chen, S.-H. Chen, J. Chem. Phys. 122 (2005) 044507.
nanoparticle aggregation as induced by salt has been investigated [25] S. Kumar, V.K. Aswal, J. Kohlbrecher, Langmuir 32 (2016) 1450.
[26] J. Kohlbrecher, W. Wagner, J. Appl. Crystallogr. 33 (2000) 804.
using scattering techniques (SANS and DLS). The presence of salt
[27] J.S. Pedersen, Adv. Colloid Interface Sci. 70 (1997) 171.
gives rise a net attraction between nanoparticles which can not [28] H. Kaya, J. Appl. Cryst. 37 (2004) 223.
be simply accounted by van der Waals interaction, implying the [29] S. Cusack, J. Mol. Biol. 145 (1981) 541.
[30] I. Bressler, J. Kohlbrecher, A.F. Thünemann, SASfit: J. Appl. Cryst. 48 (2015)
presence of non-DLVO forces. The increase in the net attraction
1587.
gives rise to systematic evolution of monomers into dimers, tetra- [31] A. Senesi, B. Lee, J. Appl. Cryst. 48 (2015) 565.
mers and higher-mers finally resulting into nanoparticle aggrega- [32] A.J. Chinchalikar, V.K. Aswal, J. Kohlbrecher, A.G. Wagh, Chem. Phys. Lett. 542
tion consistent with diffusion limited cluster aggregation (DLCA) (2012) 74.
[33] J. Teixeira, J. Appl. Crystallogr. 21 (1988) 781–785.
type of fractal morphology. The nanoparticle aggregates have been [34] P.A. Hassan, S. Rana, G. Verma, Langmuir 31 (2015) 3.
characterized by mass fractal structure having a fractal dimension [35] S. Kumar, M.-J. Lee, V.K. Aswal, S.-M. Choi, Phys. Rev. E 87 (2013) 042315.

2.6, which is much larger than that is usually explained by DLCA. [36] L. Bergström, Adv. Colloid Interface Sci. 70 (1997) 125.
[37] A.J. Chinchalikar, V.K. Aswal, J. Kohlbrecher, A.G. Wagh, Phys. Rev. E 87 (2013)
The higher fractal dimension is accounted by higher number den- 062708.
sity of primary particles present in the system in accordance to the [38] Y.-H. Peng, C.-P. Tso, Y.-C. Tsai, C.-M. Zhuang, Y.-H. Shih, Sci. Total Environ.
recent simulation results. 530–531 (2015) 183.
[39] Z.H. Lang, H.S. Jun, J. Chem. Eng. Data 41 (1996) 516.
[40] K.M. Aly, E. Esmail, Opt. Mater. 2 (1993) 195.
References [41] S.H.L. Klapp, D. Qu, R.V. Klitzing, J. Phys. Chem. B 111 (2007) 1296.
[42] M. Boström, D.R.M. Williams, B.W. Ninham, Phys. Rev. Lett. 87 (2001) 168103.
[43] C.O. Metin, L.W. Lake, C.R. Miranda, Q.P. Nguyen, J. Nanopart. Res. 13 (2011)
[1] R. Mezzenga, P. Schurtenberger, A. Burbidge, M. Michel, Nat. Mater. 4 (2005) 839.
729. [44] S. Jenkins, S.R. Kirk, M. Persson, J. Carlen, Z. Abbas, J. Chem. Phys. 130 (2009)
[2] B.Y. Feng, B.H. Toyama, H. Wille, D.W. Colby, S.R. Collins, B.C.H. May, S.B. 134702.
Prusiner, J. Weissman, B.K. Shoichet, Nat. Chem. Bio. 4 (2008) 197. [45] T.A. Jr, L.M. Witten, Sander, Phys. Rev. Lett. 47 (1981) 1400.
[3] L. Belloni, J. Phys.: Condens. Matter 12 (2000) R549. [46] L.M. Sander, Contemp. Phys. 41 (2000) 203.
[4] B.V. Deryaguin, L. Landau, Acta Phys. Chem. 14 (1941) 633. [47] J.N. Wilking, S.M. Graves, C.B. Chang, K. Meleson, M.Y. Lin, T.G. Mason, Phys.
[5] E.J.W. Verwey, J.T.G. Overbeek, Theory of the Stability of Lyophobic Colloids, Rev. Lett. 96 (2006) 015501.
Elsevier, Amsterdam, 1948. [48] M.Y. Lin, H.M. Lindsay, D.A. Weitz, R.C. Ball, R. Klein, P. Meakin, Nature 339
[6] P.D. Godfrin, N.E.V. Pérez, P.R.C. Priego, N.J. Wagner, Y. Liu, Soft Matter 10 (1989) 360–362.
(2014) 5061. [49] P. Meakin, R. Jullien, J. Chem. Phys. 89 (1988) 246.
[7] E.M. Hotze, T. Phenrat, G.V. Lowry, J. Environ. Qual. 39 (2010) 1909. [50] M.D. Camejo, D.R. Espeso, L.L. Bonilla, Phys. Rev. E 90 (2014) 012306.
[8] X. Han, J. Goebl, Z. Lu, Y. Yin, Langmuir 27 (2011) 5282.