Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Journal of Colloid and Interface Science 650 (2023) 13–18

Contents lists available at ScienceDirect

Journal of Colloid And Interface Science


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

Regular Article

Nucleation by a charged particle in fluids containing salt


Roni Kroll, Yoav Tsori ∗
Department of Chemical Engineering, Ben-Gurion University of the Negev, 8510501, Israel

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

Keywords: We present a model to describe ion-induced nucleation in fluids. Nucleation is induced by a charged molecular
Nucleation aggregate, a large ion, a charged colloid, or an aerosol particle. This model generalizes the Thomson model to
Colloids polar environments. Solving the Poisson-Boltzmann equation, we find the potential profiles around the charged
Electric field
core and calculate the energy. Our results are analytical in the Debye-Hückel limit and numerical otherwise.
From the Gibbs free energy curve vs. nucleus size, we find the metastable and stable states and the energy
barrier between them, for varying saturation values, core’s charge, and amount of salt. The nucleation barrier
decreases with increasing core charge or Debye length. We calculate the phase lines in the phase diagram of
supersaturation and core charge. We find regions of one phase, electro-prewetting, spontaneous nucleation, ion-
induced nucleation, and classical-like nucleation.

1. Introduction dipole-ion interactions. [18,19] A nucleation model that includes the


presence of ions in the fluids is surprisingly absent from the literature.
The influence of electrostatic forces on nucleation was first studied Here we present a simple ion-induced nucleation model in polar flu-
in the 1890s when Wilson observed nucleation enhancement due to the ids. The model can be analytically solved in the Debye-Hückel limit
presence of ions in a supersaturated vapor. [1] Since then, experiments and numerically in the nonlinear case. The Gibbs energy profiles as a
in ion-induced nucleation, mostly conducted in cloud chambers, proved function of nucleus size are then calculated for different saturation/un-
that the presence of ions lowers the supersaturation needed to initiate dersaturation values and salt content. Phase diagrams are found for
phase transition and significantly increases the nucleation rates com- different core ion charges, saturation, and Debye length.
pared to homogeneous nucleation. [2–4] Recent works demonstrated
the ability to manipulate drop condensation by controlling the elec- 2. Model and results
trode’s structure and electrowetting properties. [5,6] Despite the rapid
improvements in the control of liquid condensation, the nucleation The model assumes an isolated charged spherical core of radius 𝑅1
mechanism is often poorly understood. The nucleation rates are com- and fixed surface charge density per unit area 𝜎. The core can be a large
monly based on the classical nucleation theory, where the rates depend agglomerate of several molecules, some of which are ionic, or charged
exponentially on the free energy cost of creating the critical nucleus. aerosol particles. We assume that the core is surrounded by a spheri-
[7,8] For ion-induced nucleation, the Thomson model from 1933 allows cal shell of dense, polar, fluid embedded in a less polar fluid close to
the calculation of the nucleation energy of a dense liquid layer from a their coexistence. The system is spherically-symmetric and all quanti-
vapor around a core ion. [9] Both models use the bulk properties for the ties depend on 𝑟, the distance from the center of the core. The interface
liquid and vapor phases and assume a sharp interface between them. between the fluids is located at 𝑟 = 𝑅, and is taken to be infinitely sharp.
Thomson’s model fits nucleation experiments well in water but not The polar and nonpolar phases are uniform and characterized by per-
in other fluids. [10–12] Numerous theories were developed in order mittivities 𝜀𝑝 and 𝜀np , respectively.
to improve the description of ion-induced nucleation. Recent works Within the mean-field theory, the ionic densities of the cations and
used density-functional theory with a squared-gradient term to account anions obey the Boltzmann distribution 𝑛± = 𝑛0 exp(∓𝑒𝜓∕𝑘𝐵 𝑇 ), where
for two coexisting bulk phases separated by an interface with finite 𝑛0 is the salt concentration in the nonpolar phase infinitely far from the
thickness. [13–15] Other models included binary nucleation [16,17] or core particle, 𝜓 is the electrostatic potential, 𝑒 the elementary charge,

* Corresponding author.
E-mail address: tsori@bgu.ac.il (Y. Tsori).

https://doi.org/10.1016/j.jcis.2023.06.149
Received 16 February 2023; Received in revised form 15 June 2023; Accepted 22 June 2023
Available online 26 June 2023
0021-9797/© 2023 Elsevier Inc. All rights reserved.
R. Kroll and Y. Tsori Journal of Colloid And Interface Science 650 (2023) 13–18

𝑘𝐵 is the Boltzmann constant, and 𝑇 is the absolute temperature. All is the dimensionless ratio between the electrostatic energy stored in a
ions are assumed to be monovalent. sphere of radius 𝑅1 and the thermal energy. An equivalent represen-
The electrostatic potential 𝜓(𝑟) obeys the Poisson-Boltzmann equa- tation of 𝑝2 is possible using a dimensionless surface charge density
tion 𝜎̃ = 𝜎𝑅21 ∕𝑒 as
( )
2 2𝑒𝑛0 𝑒𝜓𝑖 1 𝑅1 2
𝜓𝑖′′ + 𝜓𝑖′ = sinh . (1) 𝑝2 = 𝜎̃ , (8)
𝑟 𝜀𝑖 𝑘𝐵 𝑇 2 𝑙𝐵
The index 𝑖 is 𝑖=p or np, for polar and nonpolar phases, respectively. where 𝑙𝐵 = 𝑒2 ∕𝜀0 𝑘𝐵 𝑇 is the “vacuum” Bjerrum length.
𝜓𝑝 and 𝜓np are the potentials of the respective phases. Figure (1) shows the energy curves from Eq. (5) as a function of the
The four boundary conditions are: (i) 𝜓𝑝′ (𝑟 = 𝑅1 ) = −𝜎∕𝜀𝑝 at the sur- nucleus size 𝑅̃ for varying values of the core’s charge 𝑝2 at constant
face of the charged core, (ii) 𝜓𝑝 (𝑟 = 𝑅) = 𝜓np (𝑟 = 𝑅) and (iii) 𝜀𝑝 𝜓𝑝′ (𝑟 = vacuum Debye length 𝜆̃0 (a), and varying 𝜆̃0 at constant 𝑝2 (b). At large
𝑅) = 𝜀np 𝜓np ion content, 𝑛0 is large, 𝜆̃0 is small, and the Debye lengths 𝜆𝑖 (i=𝑝, 𝑛𝑝)
′ (𝑟 = 𝑅) are continuous across the interface, and (iv) 𝜓 (𝑟 →
np
∞) = 0 in the bulk of the nonpolar fluid. The ionic density, being are small, hence 𝑈 decreases rapidly with 𝑅. ̃
Boltzmann-weighed, tends to a constant at 𝑟 → ∞, but is considerably It is instructive to look at several limiting cases.
higher within the polar phase. Once the potentials in the two domains
are known, one can integrate the electrostatic and ions energy densities Large nucleus size.
as When the nucleus is very large, i.e., when 𝑅 ≫ 𝑅1 and 𝑅 ≫ 𝜆𝑖 , we
{ [ ] look at the limit 𝑅̃ → ∞. In this limit, the energy profiles tend to a
𝑈=

𝑘𝐵 𝑇 𝑛+ (ln(𝑣0 𝑛+ ) − 1) + 𝑛− (ln(𝑣0 𝑛− ) − 1) constant given by ℎ(𝑅̃ → ∞) = 1∕(𝜀̃𝑝 (1 + 1∕𝜆̃𝑝 )). The curves in Fig. 1 tend
𝑣 ̃
to this constant at large values of 𝑅.
}
1
+ 𝜀(∇𝜓)2 − 𝜇+ 𝑛+ − 𝜇− 𝑛− 𝑑𝐫, (2)
2 High salt concentration.
where = 𝜇+ 𝜇−
= 𝑘𝐵 𝑇 ln(𝑛0 𝑣0 ) are the chemical potentials of the cation In this limit, the core size is much larger than both screening lengths,
and anion. 𝑈 can then be added to the classical bulk and interfacial 𝜆̃𝑖 ≪ 1 (𝜆𝑖 ≪ 𝑅1 in physical lengths), leading to the following expression
terms to yield the model’s predictions. for the energy per unit area,

𝑈 𝜎 2 𝜆𝑝
2.1. Debye-Hückel limit ≈ . (9)
2
4𝜋𝑅1 2𝜀𝑝

In the Debye-Hückel limit, 𝑒𝜓 ≪ 𝑘𝐵 𝑇 , and the Poisson-Boltzmann In this limit, the field is effectively localized at the surface of the
equation can be linearized, yielding charged core and does not reach the interface separating the two phases.
Consider now the core surrounded by a nonpolar phase only. If there
2 1 was no condensed layer, then 𝑎𝑝 = 𝑏𝑝 = 0 and 𝑈 = 4𝜋𝑅21 𝜎 2 𝜆2np (2𝜆np +
𝜓𝑖′′ + 𝜓𝑖′ − 𝜓𝑖 = 0. (3)
𝑟 𝜆2𝑖 𝑅1 )∕(2𝜀np 𝜆np 𝑅1 ). In the limit 𝜆̃np ≪ 1 one finds
The Debye lengths 𝜆𝑖 are defined by 𝜆2𝑖 = 𝜀̃𝑖 𝜆20 , where the relative di- 𝜎 2 𝜆np
𝑈
electric constants are defined by 𝜀̃𝑖 = 𝜀𝑖 ∕𝜀0 . 𝜆0 is the “vacuum Debye ≈ . (10)
length” defined as 𝜆20 = 𝑘𝐵 𝑇 𝜀0 ∕2𝑛0 𝑒2 , where 𝜀0 is the vacuum permittiv- 4𝜋𝑅21 2𝜀np
1∕2
ity. The Debye lengths 𝜆𝑖 are inversely proportional to 𝑛0 , the squared If the condensed polar phase is liquid and the nonpolar phase is a
root of the bulk ion density in the nonpolar phase. gas, these energies renormalize the solid-liquid and solid-gas interfa-
The solutions to Eq. (3) are cial tensions. The difference between the interfacial tensions, 𝛾sl − 𝛾sg ,
becomes an effective interfacial tension difference:
𝑒𝑟∕𝜆𝑝 𝑒−𝑟∕𝜆𝑝 ( )
𝜓𝑝 = 𝑎𝑝 + 𝑏𝑝 , 𝑅1 ≤ 𝑟 ≤ 𝑅, 𝜆𝑝 𝜆np
𝑟 𝑟 1
(𝛾sl − 𝛾sg )ef f = 𝛾sl − 𝛾sg + 𝜎 2 −
𝑒−𝑟∕𝜆np 2 𝜀𝑝 𝜀np
𝜓np = 𝑏np , 𝑅 ≤ 𝑟, (4) ( )
𝑟
1 2 1 1
where we used the boundary condition (iv) at 𝑟 → ∞ to eliminate the = 𝛾sl − 𝛾sg + 𝜎 √ −√ , (11)
2 𝜀̃𝑝 𝜀̃np
diverging exponential in 𝜓np . The remaining boundary conditions lead
to a system of three linear equations in three unknowns, 𝑎𝑝 , 𝑏𝑝 and 𝑏np , where we used Eqs. (9) and (10) and 𝜆2𝑖 = 𝜀̃𝑖 𝜆20 .
that can be readily solved by matrix inversion for a given value of 𝜎.
The energy 𝑈 in Eq. (2) is given analytically once these coefficients are Low salt concentration.
known. The dimensionless energy can be written as In the low salt regime, 𝜆̃𝑖 ≫ 𝑅̃ > 1, the field is not screened, and we
find
𝑈 ̃ ( )
= 𝑝2 ℎ(𝑅), (5) 𝑈 𝑝 𝑝2
4𝜋𝑘𝐵 𝑇 = 2 + − +
1 1
, (12)
4𝜋𝑘𝐵 𝑇 𝜀̃𝑝 𝜀̃𝑝 𝜀̃np 𝑅̃
where all distances are measured by the radius of the charged core:
𝑅̃ = 𝑅∕𝑅1 , 𝑟̃ = 𝑟∕𝑅1 and 𝜆̃𝑖 = 𝜆𝑖 ∕𝑅1 . The variables are redefined as which is the result of the Thomson model.
̃
𝑎̃𝑝 = 𝑎𝑝 𝜀0 ∕𝜎𝑅21 , 𝑏̃ 𝑝 = 𝑏𝑝 𝜀0 ∕𝜎𝑅21 , and 𝑏̃ np = 𝑏np 𝜀0 ∕𝜎𝑅21 . The function ℎ(𝑅)
is given by 2.1.1. Gibbs energy profiles
The summation of the ions and electrostatic energies in Eq. (5) can
[ ]|𝑟̃=𝑅̃
̃ = 𝜀̃𝑝 𝑎̃2 𝑒2̃𝑟∕𝜆̃𝑝 (1∕𝜆̃𝑝 − 1∕̃𝑟) − 𝑏̃ 2 𝑒−2̃𝑟∕𝜆̃𝑝 (1∕𝜆̃𝑝 + 1∕̃𝑟) − 2𝑎̃𝑝 𝑏̃ 𝑝 ∕̃𝑟 | now be added to the standard surface tension and chemical potential
ℎ(𝑅) 𝑝 𝑝 |
|𝑟̃=1 terms in the free energy. The total dimensionless Gibbs free energy
̃ ̃ Δ𝐺̃ = Δ𝐺∕4𝜋𝑘𝐵 𝑇 is written as
+ 𝜀̃np 𝑏̃ 2np 𝑒−2𝑅∕𝜆np (1∕𝜆̃np + 1∕𝑅)
̃ (6)
( )
and Δ𝐺̃ = 𝛾̃ 𝑅̃ 2 − 𝑅̃ 3 − 1 𝑝1 + 𝑝2 ℎ(𝑅)
̃ + const. (13)

𝜎 2 𝑅31 The first term, proportional to 𝑅̃ 2 , accounts for the interfacial energy
𝑝2 = (7) due to a surface tension 𝛾 between the fluids. This tension is taken as
2𝑘𝐵 𝑇 𝜀0

14
R. Kroll and Y. Tsori Journal of Colloid And Interface Science 650 (2023) 13–18

Fig. 1. (a) Energy as a function of the dimensionless nucleus size 𝑅̃ for different values of the core charge, 𝑝2 , and a constant value of the vacuum Debye length,
𝜆̃0 = 2. (b) Similarly but for a constant 𝑝2 = 5 and varying values of 𝜆̃0 ∼ 𝑛0 . 𝑛0 is the ion density in the nonpolar bulk phase. In both figures the scaled permittivities
−1∕2

are 𝜀̃𝑝 = 20, 𝜀̃np = 1. (For interpretation of the colors in the figure(s), the reader is referred to the web version of this article.)

Orange curve: electroprewetting. Here 𝑝1 < 0 but less negative, and


Δ𝐺̃ has a minimum at a small value of 𝑅.
̃ The stable state is a polar
̃
nucleus with finite radius 𝑅.

Yellow curve: ion-induced nucleation. When 𝑝1 is positive and suffi-


ciently small, Δ𝐺̃ has a metastable minimum with a finite-size radius,
but the global minimum is at 𝑅̃ → ∞. This is ion-induced nucleation
with dissociated ions. The barrier for ion-induced nucleation, Δ𝐺̃ ∗ ,
equals the difference between the maximum and the local minimum.

Purple curve: spontaneous nucleation. When 𝑝1 > 0 and sufficiently


large, Δ𝐺̃ is always decreasing with 𝑅.
̃ A spontaneous phase transition
occurs with no nucleation barrier.

Black curve in the inset: classical-like. When the core charge 𝑝2 is


either zero or very small, the effect of the charge is not strong enough to
create a metastable nucleus, and the curve is similar to that in classical
nucleation theory.
Fig. 2. Gibbs energy Δ𝐺̃ from Eq. (13) as a function of nucleus size 𝑅̃ for dif- Fig. 3 shows Δ𝐺(̃ 𝑅)
̃ curves for a supersaturated nonpolar fluid with
ferent saturations 𝑝1 . We used core charge 𝑝2 = 5 and ion density corresponding varying values of 𝑝2 and 𝜆̃0 . In both parts, the energy barrier decreases
to 𝜆̃0 = 2. The nucleation barrier, Δ𝐺̃ ∗ , is marked for the yellow curve. The in- with either an increase in the core charge or a decrease in the Debye
set shows Δ𝐺̃ for classical nucleation, with 𝑝2 = 0 and 𝑝1 = 0.2. In all figures we length. The nucleation barrier disappears at 𝑝2 ≈ 3 in (a) and at 𝜆̃0 ≈ 1
used the scaled permittivities and surface tension 𝜀̃𝑝 = 20, 𝜀̃np = 1, and 𝛾̃ = 1.
in (b).

2.1.2. Phase diagrams


independent of the field although this is not generally true [20–22]. Based on the insight gained from the above curves, one can construct
The second term, ∼ 𝑅̃ 3 − 1, is the bulk energy due to the undersat- phase diagrams that depend on the parameters 𝑝1 , 𝑝2 , 𝛾̃ , and 𝜆̃0 . Fig. 4
uration chemical potential difference Δ𝜇 between the fluids. Recall (a) is a phase diagram in the plane of saturation 𝑝1 and core charge
that the interface is found at 𝑅̃ > 1 since the radius of the core is at 𝑝2 , at fixed surface tension and several values of ion content given by
𝑅̃ = 1. The dimensionless quantities are defined as 𝛾̃ = 𝛾𝑅21 ∕𝑘𝐵 𝑇 and 𝜆̃0 = 6 (solid lines), and 𝜆̃0 = 1, 0.1 (dashed lines). The bulk binodal curve
𝑝1 = Δ𝜇𝑅31 ∕3𝑘𝐵 𝑇 . In this notation, when Δ𝜇 < 0, the nonpolar phase is corresponds to 𝑝1 = 0. Without the presence of a charged core (𝑝2 = 0),
undersaturated and 𝑝1 < 0; when 𝑝1 > 0 it is supersaturated. When the at negative values of 𝑝1 , the stable bulk phase is a homogeneous fluid,
core is uncharged, 𝑝2 = 0, and nucleation is classical. while if 𝑝1 > 0 the fluid is supersaturated and nucleation is described
It is instructive to examine the total free energy, Eq. (13), at a given by classical nucleation theory. When 𝑝2 is small and 𝑝1 < 0, the fluid is
value of core charge 𝑝2 and surface tension 𝛾̃ , and varying values of in a one-phase nonpolar state (blue). Electro-prewetting with a finite-
saturation 𝑝1 , see Fig. 2. Each curve represents a qualitatively different size nucleus occurs when 𝑝1 < 0 and 𝑝2 is sufficiently large (orange);
behavior. This type of field-induced phase transition in undersaturated conditions
was also found in a previous work. [20] The line separating between
the two states is defined as 𝜕Δ𝐺∕𝜕 ̃ 𝑅| ̃ 𝑅=1
̃ = 0. This line continues to the
Blue curve: one phase. When 𝑝1 is sufficiently negative, Δ𝐺̃ is mono- supersaturated area (𝑝1 > 0) and divides the metastable region into two
̃ The minimum energy is at 𝑅̃ = 1. The stable
tonically increasing with 𝑅. areas. In the first (purple), at small values of 𝑝2 , the curve Δ𝐺( ̃ 𝑅)
̃ is
fluid state is a single homogeneous fluid. similar to the Δ𝐺̃ ∼ (...)𝑅̃ 2 − (...)𝑅̃ 3 of classical nucleation, and has no

15
R. Kroll and Y. Tsori Journal of Colloid And Interface Science 650 (2023) 13–18

Fig. 3. (a) Gibbs energy Δ𝐺̃ from Eq. (13) as a function of the nucleus size 𝑅̃ for varying values of core charge 𝑝2 and constant ion content (𝜆̃0 = 2). The inset shows
the nucleation barrier Δ𝐺̃ ∗ as functions of 𝑝2 . (b) The same, with varying 𝜆̃0 ∼ 𝑛0
−1∕2
and constant 𝑝2 = 5. In both parts 𝑝1 = 0.3.

Fig. 4. (a) Phase diagram in the 𝑝1 –𝑝2 (saturation-core charge) plane calculated from Eq. (13). The diagram is for 𝜆̃0 = 6 (bold lines) and 𝜆̃0 = 0.1, 1 (dashed lines,
grey, and peach, respectively). (b) Phase diagram in the 𝑝1 –𝜆̃0 plane. The lines correspond to 𝑝2 = 2.4 (bold lines), 𝑝2 = 1.5 (grey) and 𝑝2 = 0 (dashed red). The
“electrostatic spinodal” curve is the solid black line separating spontaneous and ion-induced nucleations in both (a) and (b).

minimum at a finite value of 𝑅. ̃ We call this classical-like nucleation. Fig. 4 (b) is a phase diagram in the 𝜆̃ 0 –𝑝1 plane for constant sur-
The second area (yellow), at large values of 𝑝2 , corresponds to ion- face tension and core charge 𝑝2 = 2.4, for which only four phases ex-
induced nucleation, where a metastable thin polar layer exists at a finite ist. The grey lines correspond to a smaller value of 𝑝2 = 1.5, and the
radius 𝑅̃ > 1. This behavior is similar to the Thomson model. electro-prewetting zone disappears. The solid grey line separates be-
The classical stability limit for neutral cores, 𝑝2 = 0, is defined by tween classical-like and ion-induced nucleation areas. The dashed grey
the bulk spinodal curve, which separates the regions of nucleation and line is the electrostatic spinodal for 𝑝2 = 1.5, while the red dashed line
growth and spinodal decomposition. For the charged core (𝑝2 > 0), we is the “regular” spinodal of 𝑝2 = 0.
define an “electrostatic spinodal” line by vanishing of the energy barrier
2.2. Nonlinear Poisson-Boltzmann
Δ𝐺∗ = 0, or, equivalently, by Δ𝐺̃ ′′ (𝑅)
̃ = Δ𝐺̃ ′ (𝑅)
̃ = 0 at some value of 𝑅. ̃
These two equations in the variable 𝑅, ̃ 𝑝1 , and 𝑝2 trace a line 𝑝1 vs. 𝑝2 in
We now go beyond the Debye-Hückel limit and solve the full non-
Fig. 4 (a) and (b). The electrostatic spinodal line separates the regions
linear Poisson-Boltzmann equation, Eq. (1), at arbitrary values of core
of ion-induced and spontaneous nucleation. This line does not exist in charge 𝑝2 . The purely electrostatic part of the energy, defined as the
classical nucleation. Although the green area describes a spontaneous volume integral 𝑈es = ∫ (1∕2)𝜀𝐸 2 𝑑𝐫, behaves non trivially: At small val-
phase transition, it represents nucleation rather than decomposition as ̃ is monotonic. At large values of 𝑝2 , however, 𝑈es (𝑅)
ues of 𝑝2 , 𝑈es (𝑅) ̃ is
the barrier vanishes at discrete points in the fluid. The electrostatic spin- non-monotonic and has a minimum at small 𝑅̃ values, see in Fig. 5 (a).
odal displaces to the left as the core charge 𝑝2 increases, enlarging the The minimum at a finite polar layer thickness is a result of a delicate
area of spontaneous nucleation. The dashed lines show the shift in the balance: In both phases, the electric field is approximately exponen-
diagram for 𝜆̃0 = 1, 0.1 (dashed lines), reflecting that the electrostatic ̃ the electric field has a
tially decaying with 𝑟̃. At the interface (̃𝑟 = 𝑅),
spinodal changes markedly as a function of the salt concentration. positive jump upon moving from the polar to the nonpolar phase due

16
R. Kroll and Y. Tsori Journal of Colloid And Interface Science 650 (2023) 13–18

̃ at
Fig. 5. Results from the nonlinear Poisson-Boltzmann equation Eq. (1) and Eq. (2). (a) Electrostatic energy profiles and (b) full energy profiles vs. nucleus size 𝑅,
different values of the core charge, 𝑝2 . We used 𝜆̃0 = 6 and 𝑙𝐵 ∕𝑅1 = 50.

both fluids. Secondly, we consider nucleation in undersaturated fluids,


a phenomenon we call “electroprewetting”.
We calculate the system’s energy analytically in the Debye-Hückel
approximation and numerically otherwise. This energy has two interest-
ing limits: In the low salt regime, we retrieve the results of the Thomson
model. At high salt content, we obtain expressions analogous to classi-
cal electrowetting of a drop at a charged surface. We find a surprising
minimum in the purely electrostatic energy 𝑈es (𝑅), ̃ which we rational-
ize above.
The total Gibbs free energy function Δ𝐺( ̃ 𝑅)
̃ as a function of the in-
̃
terface location 𝑅 from Eq. (13), reveals several areas in the phase
diagram. At moderately supersaturated fluids (small and positive 𝑝1 ),
an increase of the core charge leads to a transition from classical-like
nucleation to ion-induced nucleation (see Fig. 6). At larger supersat-
urations 𝑝1 , one enters the area of spontaneous nucleation. The line
separating spontaneous and ion-induced nucleations, called “electro-
static spinodal”, corresponds to disappearance of the nucleation barrier,
and obeys Δ𝐺̃ ′ (𝑅) ̃ = Δ𝐺̃ ′′ (𝑅)
̃ = 0. The amount of dissociated ions, ap-
pearing in 𝜆̃0 , significantly affects this curve. For undersaturated fluids
(𝑝1 < 0), an increase of the core’s charge leads to an electroprewetting
Fig. 6. Phase diagram in the 𝑝1 − 𝑝2 plane, calculated for the nonlinear Poisson-
Boltzmann case. Here 𝜆̃0 = 6, 𝛾̃ = 100, and 𝑙𝐵 ∕𝑅1 = 50. transition, whereby the core is covered by a layer of polar fluid with
finite thickness.
The curves Δ𝐺( ̃ 𝑅)
̃ allow one to calculate the nucleation barriers and
to the decrease of permittivity and the continuity of the displacement
nucleation rates, e.g., the yellow curve in Fig. 2, which generalizes ion-
field. When the interface location is increased from 𝑅̃ to 𝑅̃ + Δ𝑅, ̃ the
induced nucleation to the presence of salt. It would be interesting to
smaller electric field squared in the gap 𝑅̃ ≤ 𝑟̃ ≤ 𝑅̃ + Δ𝑅̃ exactly balances
study how collisions between metastable nuclei lead to coalescence and
the increase in permittivity such that the electrostatic energy (integral
droplet growth. Another point that requires further study is the merg-
of ∼ 𝜀𝐸 2 ) stays constant. When the ions entropy terms in are included,
ing of several droplets in the electroprewetting regime – do large drops
𝑈 is monotonically decreasing as before, see Fig. 5 (b).
Fig. 6 shows the phase diagram at arbitrary values of 𝑝1 , and 𝑝2 continue to grow by “adsorbing” smaller droplets, or do they sponta-
found using the full numerical solutions, and 𝛾̃ = 100. The phase dia- neously break due to surface tension?
gram is qualitatively similar to the one in Fig. 4, but note that parame- The current model can be used to describe the stability of suspen-
ters values are two orders of magnitude larger. sions of colloids or nanometer-sized particles in binary mixtures. It may
also apply to maritime air and other atmospheric conditions where
3. Conclusions dissociated ions exist at sufficient numbers [23–25]. In this case, the
solvation energy of the ions in the fluids can also play a vital role in
We use a simple mesoscopic model to describe coexistence of fluids the nucleation energy and should be added to the energy in Eq. (2).
around a charged core in the presence of ions. The model considers a Such preferential solvation would lead to a more substantial accumu-
solid core (molecular aggregate or a solid particle) surrounded by a con- lation of ions in the liquid and a discontinuity of their density across
densed polar layer of dimensionless thickness 𝑅. ̃ It describes the fluids the liquid-vapor interface. In all those systems, the thermodynamic be-
by their bulk properties and assumes a sharp interface. These assump- havior is significantly different from the one predicted by the Thomson
tions are reasonable for fluids sufficiently far from the critical point. The model, and this might have important consequences for the kinetics of
first novelty of the present work is the inclusion of dissociated ions in nucleation.

17
R. Kroll and Y. Tsori Journal of Colloid And Interface Science 650 (2023) 13–18

CRediT authorship contribution statement [8] D.W. Oxtoby, Nucleation of first-order phase transitions, Acc. Chem. Res. 31 (2)
(1998) 91–97, https://doi.org/10.1021/ar9702278.
[9] J.J. Thomson, G.P. Thomson, Conduction of Electricity through Gases, 1933.
Both authors, Roni Kroll and Yoav Tsori, are fully responsible for the
[10] P.M. Holland, A.W.J. Castleman, Thomson equation revisited in light of ion-
data presented in the paper. clustering experiments, J. Phys. Chem. 86 (21) (1982) 4181–4188, https://doi.org/
10.1021/j100218a019.
Declaration of competing interest [11] A.G. Nasibulin, J.F. de la Mora, E.I. Kauppinent, Ion-induced nucleation of dibutyl
phthalate vapors on spherical and nonspherical singly and multiply charged
polyethylene glycol ions, J. Phys. Chem. A 112 (6) (2008) 1133–1138, https://
The authors declare that they have no known competing financial doi.org/10.1021/jp0755995.
interests or personal relationships that could have appeared to influence [12] C. Tauber, X. Chen, P.E. Wagner, P.M. Winkler, C.J. Hogan, A. Maisser, Hetero-
the work reported in this paper. geneous nucleation onto monoatomic ions: support for the Kelvin-Thomson the-
ory, ChemPhysChem 19 (22) (2018) 3144–3149, https://doi.org/10.1002/cphc.
201800698.
Data availability
[13] I. Kusaka, Z. Wang, J. Seinfeld, Ion-induced nucleation - a density-functional ap-
proach, J. Chem. Phys. 102 (2) (1995) 913–924, https://doi.org/10.1063/1.469158.
No data was used for the research described in the article. [14] A. Obeidat, G. Wilemski, Gradient theory of nucleation in polar fluids, in: 16th
International Conference on Nucleation and Atmospheric Aerosols, Kyoto Univ.,
Acknowledgements Grad. Sch. Energy Sci., Kyoto, Japan, Jul. 26–30, 2004, Atmos. Res. 82 (3–4) (2006)
481–488, https://doi.org/10.1016/j.atmosres.2006.02.005.
[15] H. Kitamura, A. Onuki, Ion-induced nucleation in polar one-component fluids,
We are grateful for support by the Israel Science Foundation Grant J. Chem. Phys. 123 (12) (2005), https://doi.org/10.1063/1.2039078.
No. 274/19. [16] L.Y. Chan, V. Mohnen, The formation of ultrafine ion h2o h2so4 aerosol particles
through ion-induced nucleation process in the stratosphere, J. Aerosol Sci. 11 (1)
(1980) 35–45, https://doi.org/10.1016/0021-8502(80)90142-1.
References
[17] J. Curtius, E.R. Lovejoy, K.D. Froyd, Atmospheric ion-induced aerosol nucleation,
Space Sci. Rev. 125 (1–4) (2006) 159–167, https://doi.org/10.1007/s11214-006-
[1] C.T.R. Wilson, J.J. Thomson, XI. Condensation of water vapour in the presence of 9054-5.
dust-free air and other gases, Philos. Trans. R. Soc. Lond., Ser. A, Contain. Pap. Math. [18] F. Yu, Modified Kelvin-Thomson equation considering ion-dipole interaction: com-
Phys. Character 189 (1897) 265–307, https://doi.org/10.1098/rsta.1897.0011. parison with observed ion-clustering enthalpies and entropies, J. Chem. Phys.
[2] M. Adachi, K. Okuyama, J.H. Seinfeld, Experimental studies of ion-induced nu- 122 (8) (2005) 084503, https://doi.org/10.1063/1.1845395.
cleation, J. Aerosol Sci. 23 (4) (1992) 327–337, https://doi.org/10.1016/0021- [19] A. Nadykto, F. Yu, Dipole moment of condensing monomers: a new parameter
8502(92)90002-D. controlling the ion-induced nucleation, Phys. Rev. Lett. 93 (1) (2004), https://
[3] J. Kirkby, J. Duplissy, K. Sengupta, C. Frege, H. Gordon, C. Williamson, M. Hein- doi.org/10.1103/PhysRevLett.93.016101.
ritzi, M. Simon, C. Yan, J. Almeida, J. Tröstl, T. Nieminen, I. Ortega, R. Wagner, [20] R. Kroll, Y. Tsori, Liquid nucleation around charged particles in the vapor phase,
A. Adamov, A. Amorim, A.-K. Bernhammer, F. Bianchi, M. Breitenlechner, J. Cur- J. Chem. Phys. 155 (17) (2021) 174101, https://doi.org/10.1063/5.0067249.
tius, Ion-induced nucleation of pure biogenic particles, Nature 533 (2016) 521–526, [21] S. Samin, Y. Tsori, Vapor-liquid equilibrium in electric field gradients, J. Phys.
https://doi.org/10.1038/nature17953. Chem. B 115 (1) (2011) 75–83, https://doi.org/10.1021/jp107529n.
[4] H. Rabeony, P. Mirabel, Experimental study of vapor nucleation on ions, J. Phys. [22] Y. Tsori, Electroprewetting near a flat charged surface, Phys. Rev. B 104 (2021)
Chem. 91 (7) (1987) 1815–1818, https://doi.org/10.1021/j100291a027. 054801.
[5] H. Hoek, R. Dey, F. Mugele, Electrowetting-controlled dropwise condensation [23] L. Horváth, E. Mészáros, E. Antal, A. Simon, On the sulfate, chloride and sodium
with patterned electrodes: physical principles, modeling, and application perspec- concentration in maritime air around the Asian continent, Tellus 33 (4) (1981)
tives, Adv. Mater. Interfaces 8 (2) (2021) 2001317, https://doi.org/10.1002/admi. 382–386, https://doi.org/10.1111/j.2153-3490.1981.tb01760.x.
202001317. [24] U. Platt, G. Hönninger, The role of halogen species in the troposphere, Chemosphere
[6] M. Damak, K.K. Varanasi, Electrostatically driven fog collection using space 52 (2) (2003) 325–338, https://doi.org/10.1016/S0045-6535(03)00216-9.
charge injection, Sci. Adv. 4 (6) (2018) eaao5323, https://doi.org/10.1126/sciadv. [25] P. Loche, D.J. Bonthuis, R.R. Netz, Molecular dynamics simulations of the evapora-
aao5323. tion of hydrated ions from aqueous solution, Commun. Chem. 5 (1) (2022), https://
[7] D.W. Oxtoby, Homogeneous nucleation: theory and experiment, J. Phys. Condens. doi.org/10.1038/s42004-022-00669-5.
Matter 10 (4) (1998) 897, https://doi.org/10.1088/0953-8984/10/4/019.

18

You might also like