See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/371660808

Aggregation of nanoparticles and morphology of aggregates in porous media

with computations

Article in Journal of Colloid and Interface Science · June 2023

DOI: 10.1016/j.jcis.2023.06.045

CITATIONS READS

5 84

3 authors:

Vi Nguyen Ngoc H. Pham

University of Oklahoma SC Johnson
5 PUBLICATIONS 41 CITATIONS 32 PUBLICATIONS 164 CITATIONS

SEE PROFILE SEE PROFILE

Dimitrios Papavassiliou
University of Oklahoma
259 PUBLICATIONS 5,151 CITATIONS

SEE PROFILE

All content following this page was uploaded by Ngoc H. Pham on 05 July 2023.

The user has requested enhancement of the downloaded file.

 Journal of Colloid and Interface Science 650 (2023) 381–395

Contents lists available at ScienceDirect

Journal of Colloid And Interface Science

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

Regular Article

Aggregation of nanoparticles and morphology of aggregates in porous

media with computations
Vi T. Nguyen , Ngoc H. Pham , Dimitrios V. Papavassiliou *
School of Chemical Biological and Materials Engineering, The University of Oklahoma, Norman, OK 73019, USA

G R A P H I C A L A B S T R A C T

A R T I C L E I N F O A B S T R A C T

Keywords: Hypothesis: The main hypothesis is that the aggregation process for nanoparticles (NPs) propagating in porous
Porous media media is affected by the structure of the flow field as well as by the properties of the primary NPs. If this were
Nanoparticle interactions true, then the aggregation could be predicted and controlled. However, to obtain reliable results from compu­
Dispersion
tations, one needs to account for the interactions between the NPs as well as the details of the fluid velocity, thus
Lagrangian methods
Aggregation
making advances over prior efforts that either ignored the aggregation of NPs, or used probabilistic methods to
model aggregation.
Experiments: Computational experiments were conducted using the lattice Boltzmann method in conjunction with
Lagrangian particle tracking (LPT). The LPT accounted for the physicochemical interaction forces among NPs.
Computationally obtained aggregation kinetics and fractal dimensions of Cerium oxide (CeO2) particles, sus­
pended in potassium chloride (KCl) solutions with different concentration, were verified against experimental
results. The model was then employed to investigate the effects of ionic strength, fluid velocity, and particle size
on the aggregation kinetics and the aggregate morphology, as NPs propagated in the pore space between
randomly packed spheres.
Findings: The aim of this study was to develop a computational model to simulate the aggregation of NPs and
obtain the morphology of aggregates in confined geometries, based on the physics of NP interactions and the flow
field. The most important factor that impacted both the aggregation process and the aggregate structure was
found to be the concentration of the electrolyte. The pore velocity influenced the aggregation kinetics and the NP

* Corresponding author.
E-mail address: [email protected] (D.V. Papavassiliou).

https://doi.org/10.1016/j.jcis.2023.06.045
Received 27 April 2023; Received in revised form 30 May 2023; Accepted 8 June 2023
Available online 16 June 2023
0021-9797/© 2023 Elsevier Inc. All rights reserved.
V.T. Nguyen et al. Journal of Colloid And Interface Science 650 (2023) 381–395

fractal dimension, especially in diffusion-limited aggregation. The primary particle size affected the diffusion-
limited aggregation kinetics and the fractal dimension of reaction-limited aggregates noticeably.

1. Introduction particle and the solution properties are known.

In brief, the contributions of the paper are (a) developing a validated
The use of nanoparticles (NPs) has been increasing in various fields multiscale methodology for nanoparticle aggregation in confined flow
including the petroleum industry, agriculture, medicine, biotechnology, based on the physics of particle interactions, and (b) probing the effects
and electronics [1–7] because of their unique physical and chemical of the ionic strength, fluid velocity, and particle size on the aggregation
properties. In addition, NPs have been observed to influence the flow of kinetics of NPs and the morphology (i.e., fractal dimension and
fluid in porous media, which is crucial in enhanced oil recovery [8,9], asphericity) of aggregates formed in porous media.
groundwater remediation [10], geothermal energy [11], and so on. For
example, in the field of petroleum exploration and production, NPs 2. Theory and previous computational studies
could be injected into the reservoir to reduce the surface tension of
oil–water or change the wettability of oil-rock interfaces, thus enhancing The Smoluchowski model has been used to predict the aggregation
oil recovery [8,9]. In addition, the injection of nanoparticles into the kinetics of nanoparticles [12,30,34,37,38]. This model considers the
sub-surface could facilitate the degradation of contaminants and reme­ aggregation between two particles as a second-order rate process, where
diate environmental damage [10]. In geothermal energy production, the change in concentration of an aggregate consisting of k primary
particle movement and blockage could hinder the flow of water rein­ particles is expressed as follows [12,30,34,37–40]:
jected into geothermal reservoirs [11].
dnk 1 ∑ ∑c
Aggregation of NPs in porous media is inevitable. It is also significant = α β(i, j)ni nj − αnk β(i, k)ni (1)
because it changes the size, shape, and the properties of NPs [16–18]. dt 2 i+j=k i=1

Clogging of pores may even occur; however, interactions among NPs

have often been neglected in computations while many studies consider where ni , nj , and nk are the concentrations of aggregate particles con­
particle movement to be caused only by advection and diffusion taining i, j, and k single particles, respectively; t is time; and c is the
[19–22]. Several experiments have reported on the formation of ag­ maximum number of particles in a cluster. The physicochemical prop­
gregates from NPs, such as CeO2 [14], TiO2 [17,23,24], ZnO erties involved in attachment between two particles are represented by
[5,16,25–27], and nanoscale zero-valent irons (NZVI) [28–32], while the collision efficiency α, while the collision mechanism is reflected by
several researchers have proposed simulation models for particle the collision frequency function β. This model assumes that the
agglomeration in porous media [13,30,33,34]. Most of these models are agglomeration of particles is caused by three mechanisms: perikinetic,
based on stochastic approaches, which can predict the gross aggregation orthokinetic, and differential sedimentation. In the perikinetic mecha­
kinetics but fail to provide a detailed look into the morphology of ag­ nism, particle–particle collisions are driven by Brownian motion
gregates [13,33]. Quite often models assume that the NP morphology whereas the orthokinetic mechanism is induced by flow in a fluid that
does not change during the aggregation process [30,34], despite fluid creates shear stress. The last mechanism, differential sedimentation, is a
velocity or particle size being known to affect the morphology of ag­ result of size differences among particles. In other words, the small
gregates [35,36]. particles can move faster, while the large ones tend to settle quicker and
The modeling of aggregation based on probabilities, rather than the collide with the fast-moving particles on their paths. The total collision
physics of particle–particle interactions, is a result of the difficulty in frequency function is the sum of collision frequency functions arising
addressing the multiscale problem that arises when combining the from all three mechanisms given below [39,40]:
movement of nanoparticles with aggregation [33]. The movement of 2kB T 1/3
NPs is governed by diffusion and convection; and it can be simulated at βBR (i, j) = (v + v1/3 −
j )(vi
1/3
+ v−j 1/3
) (2)
3μ i
relatively large timesteps, on the order of 10-4 or 10-5s. However, ag­
gregation is controlled by particle–particle interactions, which are G 1/3 3
important when the distance between two particles is as small as a few βSH (i, j) = (vi + v1/3
j ) (3)
π
nanometers; and it requires much smaller simulation timesteps (10-8 or
10-9s) [33]. More information about previous computational models is g (π)− 1/3
1/3 3
(4)
1/3 1/3 1/3
βDS (i, j) = (ρp − ρf )(vi + vj ) |vi − vj |
presented in the next section. 12μ 6
Our aim is to build a new Lagrangian particle tracking model to study
the aggregation of NPs and the morphology of aggregates in a porous where βBR , βSH , and βDS represent the collision frequencies due to
medium, by taking physicochemical interactions among particles into Brownian motion, shear stress, and size differences, respectively; kB is
consideration (to avoid using probabilities). The fluid flow through the Boltzmann constant; T is temperature; vi or vj is the volume of par­
randomly packed spheres is obtained by the lattice Boltzmann method ticle i or j; μ is the dynamic viscosity of the fluid; G is the average velocity
(LBM). Then, the NPs are uniformly released in the simulation domain; gradient; g is the gravitational constant; ρp and ρf are the densities of the
and their movement as well as agglomeration is examined by a force particles and the fluid, respectively. This model considers aggregates as
balance approach and Newton’s second law of motion. The particles are compact spheres; however, in reality, aggregates have fractal-like
subject to forces including gravity, buoyancy, van der Waals attraction, shapes [3]. Thus, this model ignores the effects of aggregate
electrostatic repulsion, drag, and random forces. This model keeps track morphology on the kinetic predictions.
of the positions of all particles at each timestep. When the distance be­ Lee et al. modified the Smoluchowski model by adding the fractal
tween two particles is less than the primary minimum (where the dimension of aggregates into the kinetic model [39]. The fractal
attraction force between two particles is much larger than the repulsion dimension, Df , is an index characterizing the density of an aggregate
force), it is assumed that they aggregate. The model is validated by structure; and it varies from 1 to 3. Known values of fractal dimension
comparing computational results for the aggregation of CeO2 NPs to for specific aggregate structures can be found in Sections 3.3 and 3.4, but
data from experiments. Then the transport of CeO2 NPs suspended in KCl as a rule of thumb, the higher the fractal dimension, the denser the
aqueous solutions for flow through randomly packed spheres is simu­ structure. The collision frequency functions caused by Brownian motion
lated. This system is used as an example of NP suspensions where the and fluid shear stress are expressed as [39]

382

View publication stats