Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Journal of Colloid and Interface Science 531 (2018) 523–532

Contents lists available at ScienceDirect

Journal of Colloid and Interface Science


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

Regular Article

Adsorption of surfactant from non-polar liquid into porous material


characterized using conductivity measurement
Andrei Dukhin, Sean Parlia ⇑
Dispersion Technology Inc., 364 Adams Street, Bedford Hills, NY 10507, 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

Article history: Hypothesis: Surfactants cause ionization in non-polar liquids, enabling such liquids to become electrically
Received 12 June 2018 conductive with that conductivity being a linear function of the surfactant concentration. Consequently,
Revised 18 July 2018 measurement of the conductivity can be used as a tool for monitoring surfactant concentration.
Accepted 19 July 2018
Experiments: We describe here a simple method for studying surfactant adsorption from oil into a porous
Available online 20 July 2018
material. The conductivity of solutions containing toluene and porous particles was measured as a
function of time after the addition of surfactant, at various concentrations. We applied this method for
Keywords:
characterizing surfactant (SPAN 20) adsorption by porous particles (silica gel Davisil) suspended in the
Porous material
Pore size
non-polar liquid (toluene). We the suggested a simple theoretical model for the initial stage of this
Porosity adsorption process and tested its prediction experimentally.
Ion pairs Findings: The experimental data confirms all predicted theoretical trends both qualitatively and quanti-
Non-aqueous conductivity tatively. This method can be used for understanding surfactant behavior in rock formations during oil
Liquid mixtures recovery, optimizing surfactant concentration, and analyzing chemical composition.
Ions solvation Ó 2018 Elsevier Inc. All rights reserved.
Non-polar liquid
Surfactant

1. Introduction published in this Journal on the subject of surfactant’s role in oil


recovery, for example papers [1–4]. Optimization of this process
Surfactants are widely used these days for enhancing oil recov- requires methods that would not only enable us to select the most
ery. They can modify the wettability of rock formations, changing suitable surfactant for a particular rock formation based on its
them from oil-wet to water-wet, making possible the displacement composition, but also at the ideal surfactant concentration. Simply
of oil by oil/water gravity drainage. There are multiple papers stated, such methods should be capable of characterizing the
adsorption of surfactant by a porous material that is saturated with
either water, oil, or an oil/water mixture. It would be highly desir-
⇑ Corresponding author. able to have such tools available in the laboratory environment for
E-mail address: sparlia@dispersion.com (S. Parlia). conducting preliminary tests on a small scale.

https://doi.org/10.1016/j.jcis.2018.07.075
0021-9797/Ó 2018 Elsevier Inc. All rights reserved.
524 A. Dukhin, S. Parlia / Journal of Colloid and Interface Science 531 (2018) 523–532

Here we suggest such a method that, as far as we know, has not and verified it with experiments conducted with particles of differ-
been used for such an application. This new method is based on ent sizes, porosities, and pore sizes.
measuring the conductivity of a surfactant solution in non-polar
so does this mean we can't use
liquid. non-ionic surfactants? 1.1. Theoretical model of the initial stage that is controlled by diffusion
It has been known since late 20th century that ionic surfactants
produce stable ions in non-polar liquids, as described in the review In this section we consider the simplest theoretical model of the
by Morrison [5]. The discovery of ionization in non-polar liquids by adsorption process discussed above. This model is based on several
non-ionic surfactants was made by Dukhin and Goetz in 2005 [6]. assumptions, and the experimental setup is designed such that
There have been many papers published on this subject since then these assumptions are satisfied. The experiment was conducted
[7–16], which allows us to make the general statement that prac- on porous particles, with sizes ranging from tens to hundreds
tically all surfactants enhance ionization in non-polar liquids, as microns, that are immersed in non-polar liquid. Intense mixing
long as they are soluble in said liquid. prevents sedimentation of the particles and distributes them
Surfactants, as well as other amphiphilic substances, serve as homogeneously throughout the sample. Such setup allows us to
solvating agents providing steric stabilization for ions in non- introduce following assumptions into the theory:
polar liquids and preventing their re-aggregation into ion-pairs.
Details of this phenomenon can be found in recently published Assumption 1. Initial stage of surfactant penetration into the
reviews [17,18] that follow landmark works by Bjerrum, Onsager, porous particles from the non-polar liquid is controlled by
Fuoss, and Kraus from the early 20th century [19–23]. The corre- diffusion.
sponding models of ion formation in nonpolar liquids are further
described in the Appendix A.
Assumption 2. Concentration of the surfactant [c(t)] is constant
All of these studies revealed a linear dependence between con-
everywhere in the liquid bulk up to the particle surface and decays
ductivity and surfactant concentration at low surfactant concentra-
in time (t) due to irreversible surfactant adsorption inside of the
tion. Some examples of these conductivity-surfactant concentration
particles.
dependences are shown in Figs. 1 and 2.
Linear dependence allows for using conductivity as a simple
measure of the surfactant concentration in a particular non-polar Assumption 3. Measured conductivity of the sample [K(t)] is pro-
liquid, including oil. In order to do this we would only need to gen- portional to the concentration of the surfactant [c(t)] with a coeffi-
erate a calibration curve by measuring the conductivity at a set of cient of proportionality [k] being independent on time:
known surfactant concentrations. Then we can use such calibration
K ðt Þ ¼ kcðtÞ ð1Þ
curve, which would be similar to the curves in Figs. 1 and 2, to deter-
mine the surfactant concentration from the measured conductivity This assumption is based on the several experimental studies pub-
for any solutions of this surfactant in the same non-polar liquid. lished in the papers [6,7,15,16]. All these studies reported a linear
We apply this method here for studying adsorption of the non- relationship between the measured conductivity and surfactant
ionic surfactant SPAN 20 by porous chromatographic particles, concentration in the lower concentration range.
which serve as a convenient model for porous rock. We suspended
the particles in toluene, which models oil. We then developed a Assumption 4. Particles are monodisperse and spherical with
simple theoretical model for the early stages of such adsorption radius R and volume fraction in the sample u.

Fig. 1. Conductivity of the various non-polar liquids with SPAN 80. Abbreviation d.p stands for dielectric permittivity of the liquid, copied from the paper [6].
A. Dukhin, S. Parlia / Journal of Colloid and Interface Science 531 (2018) 523–532 525

Fig. 2. Conductivity of the kerosene solutions with non-ionic surfactants having various HLB numbers, copied from the paper [6].

Assumption 5. Pores have radius [a] with total porosity P. We also in general. But, we can introduce a simple assumption that should
assume the simplest geometry of pores making their length equal be valid for short times during the initial stage of diffusion.
to the diameter of the particle (2R), as well as being straight and
open on both ends. Assumption 7. Diffusional flux at the early stages of this process
occurs mostly at the entrance of the pores, since surfactant that
enters the pores will be completely adsorbed in the vicinity of the
Assumption 6. The non-polar liquid saturates and wets the porous
entrance. We can assume that the concentration of the surfactant
space inside of the porous particles. In order to assure this assump-
drops from c(t) at the very entrance of the pore down to 0 over
tion we add surfactant to the liquid prior to the adding particles.
certain distance D. We assume that this distance D remains
The surfactant promotes wetting of the pore walls and ensures sat-
constant and is time independent during early stages of adsorp-
uration of the pores with the non-polar liquid.
tion. This leads to the following approximate equation for the
The theoretical model that we are looking for should yield an
diffusional flux:
equation describing the evolution of measured conductivity over
time. These kinetics would reflect depletion of the surfactant in
cðtÞ
the bulk liquid due to its adsorption by the porous particles. An J ðt Þ ¼ D S ð4Þ
equation for the total surfactant amount preserved in the bulk [c
D
(t)] is the starting point for such mathematical model. It can be Substituting this equation into Eq. (2) and applying the time
written as the following: derivative leads to the following differential equation for c(t):
Z dcðtÞ DScðtÞ
t
V ¼ ð5Þ
V ½c0  cðtÞ ¼ J ðt Þdt ð2Þ dt D
0
Solving this simple differential equation yields following
where V is the total sample volume, c0 is the initial concentration of expression for c(t)
the surfactant, and J(t) is the diffusional flux of the surfactant from cðtÞ DS
the liquid bulk into the porous particles. ln ¼ t ð6Þ
c0 VD
This diffusional flux can be linked to the surfactant concentra-
tion with the following equation: We can link the diffusion cross-sectional area per unit volume
(S/V) to the geometric properties of particles using Assumptions
dcðt; rÞ 4 and 5.
J ðtÞ ¼ D S ð3Þ The total diffusion cross-section for a single particle (Sp) is equal
dr r¼R
to the product of the pore cross-sectional area (pa2) and the num-
where c(t,r) is concentration of surfactant at time t and distance r ber of the pores Np:
from the particle center inside of the pores, D is the diffusion coef-
Sp ¼ Np pa2 ð7Þ
ficient of the surfactant molecules, and S is an area of the cross-
section that is open for surfactant molecules to diffuse inside the The number of pores can be expressed by using the definition of
particles. the porous material’s porosity P as the ratio of the pore volume
Concentration of the surfactant diminishes from the bulk into (Np * pa2 * 2R) to the total volume of the particle (4/3pR3). This
the particles, which requires a minus sign () in front of the special leads to the following expression for the number of pores in a sin-
derivative. We do not know the value of the concentration gradient gle particle:
526 A. Dukhin, S. Parlia / Journal of Colloid and Interface Science 531 (2018) 523–532

2 R2 the same 60 Å pore size but different particle sizes, as shown in


Np ¼ P ð8Þ Table 1. The other two types of the particles have a pore size of
3 a2
150 Å and different particle sizes, as also shown in Table 1.
Substituting this expression for Np into Eq. (7) leads to the fol-
Table 1 specifies the particle size ranges for the each material
lowing equation for the diffusional cross-section of a single particle:
type in mesh size as is given by Sigma-Aldridge, and in microns
2 as converted by us.
Sp ¼ pPR2 ð9Þ
3 There is no information given for the median particle size. We
The total number of particles Nv in the sample volume V can be were able to measure median particle size using a Dispersion Tech-
expressed as a function of the porous material volume fraction u: nology DT-100 Acoustic spectrometer that functions in accordance
with ISO standard ISO-20998 [24]. It is described briefly below. The
3V u measured particle size values are also shown in Table 1.
Nv ¼ ð10Þ
4pR3 The manufacturer has not reported the porosity values for these
Finally, the total diffusion cross-section equals the diffusion particles. We measured the porosity using a Dispersion Technology
cross-section for a single particle multiplied by the number of DT-900 Porosity Meter, which is described below. Values of poros-
particles: ity are reported in Table 1.
We used the non-ionic surfactant SPAN 20 (Sorbitan Monolau-
uPV rate) purchased from Sigma-Aldrich in these experiments. This sur-
S¼ ð11Þ
2R factant has the following properties: density is equal to 1.032 g/cc,
Substituting this equation into Eq. (6) leads to the following and the HLB number is 8.6. The rather high HLB number ensures a
simple expression describing the initial stage of surfactant concen- substantial conductivity increase will occur when this surfactant is
tration evolution: added to the non-polar liquid.
The non-polar liquid used in these experiments was Toluene.
cðtÞ DuP
ln ¼ t ð12Þ Pure (99.9%) toluene was purchased from Sigma Aldrich.
c0 2RD
Finally, we can apply Assumption 3 in order to link surfactant 2.1. Sample preparation and measurement procedure
concentration to the conductivity, which is measurable parameter.
This leads to the following expression, which is supposed to reflect The first step in the sample preparation was to mix the non-
the initial stage of conductivity evolution due to surfactant adsorp- ionic surfactant SPAN 20 with toluene. We added 1 g of SPAN 20
tion in the porous particles: into 49 g of toluene, yielding a concentration of 2% by weight. This
KðtÞ DuP liquid mixture was prepared in a cup with a diameter sufficiently
ln ¼ t ð13Þ small enough to allow the conductivity probe to be inserted into
K0 2RD
the cup, while minimizing evaporation of the liquid. The small
There are several theoretical predictions that can be derived diameter also ensures that the liquid reaches a sufficiently high
regarding the evolution of conductivity normalized with the con- level such that both electrodes of the conductivity probe
ductivity of the pure surfactant solution: (described in the next section) are covered.
KðtÞ This cup is placed on magnetic mixing plate with a small mag-
K¼ ð14Þ netic cross in the bottom of the cup. The rate of mixing generated
K0
by the rotating magnetic cross is very important. It must be high
Prediction 1. The logarithm of the normalized conductivity is a enough to keep the porous particles suspended in the liquid.
linear function of time at the early stages of the described The second important function of mixing is supporting Assump-
adsorption process; tions 1 and 2 of the suggested theoretical model, which state that
Prediction 2. The rate of this linear time dependence is recipro- the surfactant concentration is constant everywhere in the liquid
cally proportional to the particles size; bulk, and that the adsorption process is controlled by diffusion of
Prediction 3. The rate of this linear time dependence is propor- the surfactant into the particles.
tional to the porosity of the particles; We have verified these assumptions by conducting measure-
Prediction 4. The rate of this linear time dependence is propor- ments at several rates of magnetic cross rotation and showing that
tional to the volume fraction of the particles. the conductivity in the presence of porous particles is independent
Prediction 5. The rate of this time dependence is independent of the rotation rate in the range that we used.
on pore size. This is the most surprising and counterintuitive Continuous measurements of conductivity begin with the
prediction. toluene-surfactant mixture without any porous particles present.
We can verify all these theoretical predictions via experiment, This allows us to verify that the initial concentration of surfactant
which is the main purpose of this paper. (and hence the conductivity) is the same for all samples.
We then add 1 g of porous particles into the cup with toluene-
2. Materials and sample preparation surfactant mixture without interrupting the ongoing conductivity
measurements.
We used 5 different types of silica gel Davisil porous particles Conductivity of the sample begins to decay in time due to
purchased from Sigma-Aldrich. Three types of these particles had surfactant adsorption by the particles. The rate of this decay

Table 1
Properties of silica gel porous particles Davisil. Source of information: * - manufacturer, ** - acoustic particle sizing [24], *** - high frequency conductivity [23].

Name Pore size, Angstrom* Particle size, mesh* Particle size, micron* Particle size d50, micron** Porosity, %***
Davisil 633 60 200–425 30–75 45.8 45.95
Davisil 635 60 60–100 150–250 247.0 43.83
Davisil 636 60 35–60 250–500 345.0 47.02
Davisil 643 150 200–425 30–75 72.6 64.08
Davisil 646 150 35–60 250–500 397.0 60.35
A. Dukhin, S. Parlia / Journal of Colloid and Interface Science 531 (2018) 523–532 527

slows with time. We kept the measurements going until the con- 2.2.2. Porosity measurement
ductivity reaches saturation and remains practically unchanged Porosity measurements were conducted with a Dispersion
with time. Technology DT-900 Porosity Meter. This instrument applies a high
frequency 3 MHz electric field for measuring porosity of porous
2.2. Instruments materials. This instrument is described in detail in the paper [25].

We used 3 different instruments for measuring conductivity, 2.2.3. Particle size measurement
porosity, and particle size. Particle size distribution of the porous particle dispersions was
measured with a Dispersion Technology DT-100 Acoustic spec-
2.2.1. Conductivity measurement trometer (for details, see the recently published book [26]). This
A DT-700 non-aqueous conductivity probe made by Dispersion method is done by measuring the attenuation of ultrasound prop-
Technology, Inc. was used for measuring conductivity in these agating through the sample as a function of frequency, from 1 to
experiments The DT-700 is able to measure from 10E11 to 100 MHz. Then, a theoretical fit is used for finding a particle size
10E4 S/m, with a single-measurement precision of ±(1% + distribution that produces a theoretical attenuation frequency
10E11 S/m) over the complete range. By taking multiple mea- spectrum that matches the experimental one with the least error.
surements for a single data point, this precision is improved. There is a very useful simplification of the particle size search for
The DT-700 probe consists of two concentric coaxial cylindrical sizes exceeding 10 mm where the particles scatter ultrasound. This
electrodes: an inner one and an outer one, along with a guard elec- simple algorithm indicates that the median particle size is approx-
trode which serves to prevent leakage. In order to make a measure- imately 2/3 of the wavelength at the frequency corresponding to
ment, the probe is dipped into the sample such that the holes in the maximum scattering attenuation, expressed in units of dB/
the outer electrode are covered. This outer electrode can also be cm/MHz. All samples studied in this paper have particle sizes much
removed for cleaning. larger than 10 mm, which allowed us to use this simple method for
The newest model of the DT-700 measures current using a log- estimating the median particle size. The results are shown in
amplifier which requires no ‘‘range selection.” A voltage waveform Table 1.
is applied to the outer electrode, while the output signal from the
log amplifier is simultaneously captured by second channel of the 3. Experimental results and discussion
Analog-to-Digital converter. The desired waveform is then calcu-
lated from the captured log-amp values through an inverse log cal- Fig. 3 presents the evolution of measured conductivity over
culation. This technique still allows for measurements to be made time for one type of porous particle – silica gel Davisil 646. The
over many decades of current. conductivity decays over time because the porous particles adsorb
First order Fourier coefficients are used to calculate the ampli- surfactant. The concentration of surfactant in the bulk toluene
tude and phase of both the voltage and current waveforms. The declines, which leads in turn to the drop in conductivity.
Complex Conductance of the cell contents is then computed from The rate of conductivity decline slows down over time because
the current voltage ratio. Lastly, the specific conductivity of the the diffusion flux decreases with diminishing bulk surfactant con-
sample material is determined by computing the real part of the centration. Eventually, equilibrium between the bulk and the por-
complex conductance and then multiplying by a cell constant, ous particles’ interior is established and the conductivity reaches
which is determined via instrument calibration. saturation.

Fig. 3. Conductivity evolution of toluene-Span 20 mixtures with 2% wt porous silica gel Davisil 646. Span 20 was added twice (1 g each) to the mixture after conductivity
reaches saturation.
528 A. Dukhin, S. Parlia / Journal of Colloid and Interface Science 531 (2018) 523–532

Fig. 4. Reproducibility test. Evolution of toluene-Span 20 mixtures with 2% wt porous silica gel Davisil 646, 2 repeated measurements.

Fig. 3 demonstrates that addition of the surfactant at this satu- We have decided to use the only the first leg of this equilibra-
ration stage would lead to a sharp increase in the conductivity, tion kinetics for testing the primitive theoretical model for this
with a second step of the conductivity slow declining to follow. process suggested above. Fig. 4 demonstrates that we can repro-
It is interesting that the second saturation level exceeds the first duce this first leg quite precisely.
one. This reflects the process of the pores gradually filling up with Fig. 5 presents conductivity evolution data for all 5 types of the
surfactant. porous particles that we studied. In order to make the differences
The third addition of surfactant, as shown in Fig. 3, causes only a between these curves more pronounced, and to compare this
very small increase in conductivity, because the pores have appar- experimental data with the theoretical predictions, we converted
ently already been almost filled. this figure to another one, Fig. 6, which presents the logarithm of

Fig. 5. Conductivity evolution of toluene-Span 20 mixtures with 2% wt of various porous silica gel Davisil particles.
A. Dukhin, S. Parlia / Journal of Colloid and Interface Science 531 (2018) 523–532 529

Fig. 6. Evolution of ln(K(t)/K0) over time for dispersion of various porous silica gel Davisil particles at 2% wt in toluene-2% SPAN 20 mixture. Solid lines and equations
correspond to theoretical best linear fit according to Eq. (13).

the so-called ‘‘normalized conductivity”, ln(K(t)/K0). The theory (7/3). The experimental ratio is much bigger than the theory
yields the following expression for this parameter, first described predicts.
in Eq. (13): We suspect that this discrepancy is the result of particle size
differences between the samples. If the particle size of sample
KðtÞ DuP 646 is somewhat smaller than for sample 636, this could reconcile
ln ¼ t ¼ At
K0 2RD the ratio of theoretical and experimental rates.
The coefficient A on the right hand side of this equation can be This observation points towards importance of detailed particle
considered as the ‘‘rate of conductivity decline”. We will use this size information for interpreting these experiments.
term below for the fitting parameter of the linear regression to
the experimental data. Prediction 4. The theory predicts that the rate must be propor-
There have been several theoretical predictions made for this tional to volume fraction of the particles.
parameter in Section 2. We can verify all of them using the curves
shown in Fig. 6, and confirm that these experimental curves con- We have conducted a special test for verifying this prediction
firm all theoretical prediction qualitatively. by measuring the decrease in conductivity over time for samples
with different particle concentrations: 0.5 g, 1 g, and 2 g in 50 ml
Prediction 1. All experimental curves in Fig. 6 exhibit a linear of the toluene-surfactant mixture. We have done this for the
dependence of the logarithm of the normalized conductivity sample 636, which has the largest pore size and largest particle
on time, as predicted with Eq. (13) at the initial stages of the size.
adsorption process. This linearity is especially pronounced for The results are shown in Fig. 7. Each volume fraction was mea-
the 3 samples with larger sizes: Davisil 635, 636 and 646. sured twice in order to ensure reproducibility. It can be seen that
Prediction 2. The rate of this linear time dependence A indeed these measurements are in fact very reproducible.
increases with decreasing particle size. The fastest conductivity The theory predicts that the rates for these samples with differ-
decline is for the samples with sizes ranging from 30 to 75 mm. ent volume fractions should vary as a ratio of volume fractions,
The sample with sizes ranging from 150 to 250 mm is the next meaning 0.5:1:2. The rate of the most diluted sample (with 0.5 g
fastest, and the samples with the sizes of 250–500 mm demon- of particles) must therefore be 4 times slower than the rater for
strate the slowest decline. the most concentrated sample (with 2 g of particles).
Prediction 3. The rate of this time dependence A increases with The experimental results demonstrate a much different ratio
porosity. Samples 646 and 636 have the same particle size surprisingly. The ratio of the averaged linear regression coefficients
range of 250–500 mm. However, sample 646 has a higher poros- shown in Fig. 6 are 0.095:0.31:1.57.
ity of 60% compared to the sample 636 which has a porosity of This means that the ratio of the slowest rate to the fastest rate is
45%. The rate of conductivity decline for sample 646 is much 16.5.
higher than for sample 636. This means that the experiment points towards a square depen-
dence of the conductivity decline rate, instead of the theoretical
However, the difference in the rates between samples 646 and linear dependence shown in Eq. (13).
636 cannot be explained just by porosity differences. The porosity We have no explanation for this difference as of now. Particle
ratio between these two samples is about 4/3, whereas ratio of the size cannot be the reason because this anomaly is observed for
linear regression coefficient for these two curves is 0.3515/0.1549 samples with the same size particles.
530 A. Dukhin, S. Parlia / Journal of Colloid and Interface Science 531 (2018) 523–532

Fig. 7. Volume fraction dependence test. Evolution of ln(K(t)/K0) over time for dispersion of Davisil 646. Particle content varied as 0.5 g, 1.0 g and 2.0 g in 50 ml of the toluene-
surfactant solution. Each sample was measured 2 times. Solid lines and equations correspond to theoretical best linear fit according to Eq. (13).

Prediction 5. The rate of this time dependence is independent samples (2.5 times). The most likely origin of this difference
on the pore size. This is the most surprising and counterintu- between theory and experiment is uncertainty of the particle size.
itive prediction. However, the experiment confirms this conclu- Unfortunately, uncertainty in particle size increases for samples
sion. It turns out that we can reconcile the rates of adsorption with larger sizes. Consequently, we may expect an increase in the
for different samples quantitatively using just particle size deviation between theory and experiment for the samples with
and porosity data. This is shown in the next sub-section. larger particle sizes.
The ratio of the rates for the samples with largest sizes (636 and
3.1. Comparison of the adsorption rates quantitatively 646) are 0.89 for the theory and 0.44 for the experiment, which is
still much less than pore size differences.
According to Eq. (13), the rate of adsorption at the early stages The ratio of the rates for the samples with the intermediate size
of this process is defined in the theory as: (635) and the smallest size (633) are 0.185 for the theory and 0.104
for the experiment, which is still much less than pore size
DuP
A¼ ð14Þ differences.
2RD
We believe in general that the simple theory suggested above
We cannot calculate this parameter because we do not know provides a surprisingly close fit to the experimental data, even
the values of the diffusion coefficient D and diffusion distance D: on a quantitative scale.
However, we can compare the rates for different samples since
the ratio of these rates would depend on the particle size and
porosity only. For instance, the ratio of rates for the two samples 4. Conclusions
with the smaller sizes, samples 633 and 643, is equal to:
This study stresses the ability to study the adsorption of surfac-
DuP 633
A633 2R633 D P633 R643 tant into the pores of porous materials via conductivity measure-
theory : ¼ ¼ ð15Þ
A643 D2RuP643D P643 R633 ments of the bulk nonpolar liquid said porous material is
643
saturated with. We proposed a simple theoretical model to
Substituting in the values of porosity and particle size for these describe such systems and their change in conductivity over time
two samples (from Table 1) leads to the following theoretical value as this adsorption occurs. Our model implied that the logarithm
for the adsorption rate ratio for these two samples: of the normalized conductivity is a linear function of time (at early
the adsorption stage).
A633
theory : ¼ 1:14 ð16Þ From this model we made predictions about how other param-
A643
eters – particle size, porosity, volume fraction of particles, etc. –
On the other hand, we can calculate the ratio of rates from the impacted this linear time dependence. Most interestingly we pre-
linear regression coefficients shown in Fig. 6: dicted that this linear time dependence was independent of pore
size.
A633 5:8211
experiment : ¼ ¼ 0:88 ð17Þ By studying non-polar systems in which non-ionic surfactant
A643 5:1236
SPAN 20 was added to mixtures of toluene and non-conducting
There is a difference between the theoretical value of the ratio porous particles, we were able to confirm each of the predictions
of the rates (1.14) and the experimental value (0.88), but it is much we made in this paper. This includes confirming that conductivity
smaller than the difference in pore sizes between these two trends were independent of pore size for a given particle.
A. Dukhin, S. Parlia / Journal of Colloid and Interface Science 531 (2018) 523–532 531

We have shown first that overall bulk conductivity will The fact that these ions cannot be solvated by the non-polar liquid
decrease as a function of surfactant adsorption, and that this trend molecules due to lack of the molecular dipole moments (in the
can be repeated by adding additional surfactant to the nonpolar non-polar liquid molecules) was simply ignored.
continuous phase (see Fig. 3). We have also demonstrated that On the other hand, Fuoss and Krauss realized that ion-pairs
not only is there a linear dependence of conductivity on surfactant carry dipole moments that could lead to the formation of more
content at low concentrations, but also that there exists a linear complicated structures due to ion-dipole and dipole-dipole
time dependence on the adsorption of surfactant into porous mate- interactions. This approach to non-aqueous electrochemistry,
rials during the initial adsorption stage. It is from this initial linear based on the dissociation model, is still being applied in relevant
adsorption rate that the model was derived for examining the driv- peer-reviewed literature, as can be seen in recent (2009) publica-
ing factors in this adsorption process. tions [27].
We measured suspensions of porous particles that varied in This model ignores the existence of other objects with dipole
concentration, particle size, and pore size. We were able to estab- moments in such non-polar solutions: neutral molecules of the
lish trends that were consistent across both theoretical and exper- added substances. It turns out that practically all substances that
imental data, and are confident that this technique can be used to causes ionization of non-polar liquids are amphiphilic. They must
further understand and predict the adsorption of surfactant by por- have a sufficiently long carbon chain (hydrophobic) on one side
ous materials. However, while the trends we predicted were con- for solubility in non-polar liquid, and a polar group that is capable
firmed experimentally, their actual rates deviated from the of dissociation (hydrophilic) on the other side. This polar group has
predicted values. We suspect that these discrepancies are due to a dipole moment, which may play and important role in ion
discrepancies in particle size distribution. And while we did mea- formation of various structures.
sure the particle size of these porous particles using acoustic spec- The earliest paper known to us on the subject of the interactions
troscopy, we were unable to resolve these differences using this of dipole moments of added molecules in non-polar liquids is by
technique alone. A more rigorous examination into the particle size Nelson and Pink in 1951 [28]. In that paper they wrote that
distribution of these porous particles would be useful for upcom- ‘‘. . .in a non-polar solvent, aggregation is clearly the result of a bal-
ing research in this field. ance between the solubilizing power of the hydrocarbon chains
Lastly, this research was initiated by studies conducted in 2005 and the attractive forces between polar parts of the soap [surfac-
by Dukhin and Goetz [6] pertaining to the of ionization of non- tant] molecules. . .” They also wrote that ‘‘. . .Like the coulombic
polar liquids via non-ionic surfactants. Parlia and Dukhin contin- force between ions, the attractive force due to the dipoles is
ued this work, publishing several additional papers on nonpolar exerted indiscriminately on all other polar molecules depending
liquids’ electrochemistry [12,13,15,17]. Furthermore, it was shown only on the distance of separation. . .”
that other amphiphilic substances such as alcohols can also induce This dipole-dipole interaction leads to the formation of ‘‘inverse
ionization of nonpolar liquids [14]. We have now expanded upon micelles” in non-polar liquids, which is supported by many
this work to include porous materials saturated with ionized detailed experimental studies. These ‘‘Inverse micelles” are neutral
non-polar liquids, and developed a theoretical model with which entities. However, it is possible that added molecular dipoles
to better study such systems. would affect the formation of ions in non-polar liquids.
According to Morrison [5], a paper by Mattoon and Mathews
Appendix A. Models of ion formation in non-polar liquids from 1949 [29] was the first to mention ‘‘charged inverse micelles”.
They discovered that AOT micelles behave as positively charged
There are two recently published reviews [17,18] on the elec- particles. The mechanism describing this charging of inverse
trochemistry of non-polar liquids. Both of them present an analysis micelles is called the ‘‘disproportionation model”. It assumes that
of the current understanding of the mechanisms by which ions two micelles exchange electric charge during collision.
form and are maintained in non-polar liquids. Both reviews are The disproportionation model is most widely used in Colloid
in agreement regarding the importance of ion-pair formation in Science. This field became involved into non-polar liquid electro-
non-polar liquids. On this subject they are in consensus with pre- chemistry because amphiphilic substances, such as surfactants,
vious studies of non-aqueous electrochemistry conducted in early are common.
20th century by Bjerrum, Onsager, and Fuoss [19–23]. They con- However, the disproportionation model is not the only mecha-
clude that the balance between free ions and ion-pairs is a control- nism used to describe these phenomena. The ‘‘Fluctuation model”
ling factor of ionic strength in non-polar liquids. reflects the potential role of water in the ionization process. The
Unfortunately, there is no such concurrence on the mechanism addition of water to a non-polar liquid that contains surfactant
that is responsible for ion formation in non-polar liquids in the first (with a certain HLB number) leads to the formation of microemul-
place. Below we present some comments regarding the variety of sion droplets. Such droplets are similar to inverse micelles to a
models on this subject. degree, except with a water core. And similarly to inverse micelles
Early works on non-aqueous electrochemistry stressed the these droplets can exchange electric charge during collisions and
importance of electrostatic interactions between cations and become ions. The fluctuation model theory had been derived and
anions due to the low dielectric constant of the non-polar liquid. discussed more thoroughly in the papers [30,31]. This model can
This leads to the understanding that ions should have a sufficiently be important even when water is present in very small amounts,
large size in order to exist as free entities in such liquids. That is as was shown in the paper [32].
why early pioneering works by Fuoss, Krauss, and Onsager were Lastly, the classical ‘‘dissociation model” described earlier can
conducted with large picrate molecules [21–23]. They assumed be extended to solutions that include amphiphilic substances.
that the dissociation of such molecules would yield large ions that The polar heads of these molecules can occasionally dissociate like,
could withstand electrostatic attraction. Effectively, they applied for instance, what happens with picrate, alcohols, and ionic surfac-
the classical ‘‘dissociation model” for explaining ion formation in tants. The nearby amphiphilic molecules, while still neutrally
non-polar liquids with one additional requirement: that ions must charged because they have not dissociated, are attracted to the
be sufficiently large. newly formed ions due to the dipole moment in their polar heads.
However, they neglected to discuss the solvation aspect of ion- This attraction leads to a buildup of a neutral molecule shell
ization. There is practically no discussion about the state of the around each of the charged (dissociated) ions. This, in turn, would
small counter-ions that split from the large picrate molecules. enlarge such ions, sterically protecting them from re-aggregation.
532 A. Dukhin, S. Parlia / Journal of Colloid and Interface Science 531 (2018) 523–532

In electrochemical terms, the added amphiphilic substance can [11] B.A. Yezer, A.S. Khair, P.J. Sides, D.C. Prieve, Determination of charge carrier
concentration in doped nonpolar liquids by impedance spectroscopy in
‘‘solvate” ions in non-polar liquids with molecules having no such
presence of charge adsorption, JCIS 469 (2016) 325–337.
capacity due to their lack of dipole moments. In contrast to aque- [12] A. Dukhin, S. Parlia, Ion-pair conductivity theory fitting measured data for
ous electrochemistry, the added substances serve as ‘‘solvating various alcohol-toluene mixtures across entire concentration range, J.
agents” and not necessarily as ‘‘electrolytes”. This ‘‘dissociation Electrochem. Soc. 162 (4) (2015) H1–H8.
[13] S. Parlia, A. Dukhin, P. Somasundaran, Ion-pair conductivity theory: mixtures
model with self-solvation” was suggested in the paper [17]. The of butanol with various non-polar liquids and water, J. Electrochem. Soc. 163
term ‘‘self-solvation” is derived from the amphiphilic ion being sol- (7) (2016) H1–H6.
vated by similar amphiphilic molecules, e.g. a dissociated ethanol [14] A. Bombard, A.S. Dukhin, Ionization of a nonpolar liquid with and Alcohol,
Langmuir 30 (15) (2014) 4517–4521.
ion would be solvated by other ethanol molecules. [15] S. Parlia, A.S. Dukhin, P. Somasundaran, Ion-pair conductivity theory: mixtures
The structure of an ion that is ‘‘solvated with a shell of amphi- of non-polar media with non-ionic surfactant, J. Electrochem. Soc. 164 (12)
philic molecules” is identical to the ‘‘charged inverse micelle”. In (2017) E1–E5.
[16] S. Parlia, P. Somasundaran, A.S. Dukhin, Ion-pair conductivity theory IV: SPAN
this sense ‘‘disproportionation model” and ‘‘dissociation model surfactants in toluene and role of viscosity, J. Electrochem. Soc. 165 (5) (2018)
with self-solvation” lead to the same result, due to the same ster- H1–H6.
ically stabilized ions. [17] A.S. Dukhin, S. Parlia, Ions, ion-pairs and inverse micelles in non-polar media,
Curr. Opin. Colloid Interf. Sci. 18 (2013) 93–115.
However, there is one important difference. ‘‘Dissociation [18] D.C. Prieve, B.A. Yezer, A.S. Khair, P.J. Sides, J.W. Schneider, Formation of charge
model with self-solvation” does not require the preliminary step carriers in liquids, Adv. Colloid Interf. Sci. (2016).
of neutral micelle formation. That is why it is applicable in cases [19] N. Bjerrum, A new form for the electrolyte dissociation theory, in: Proceedings
of the 7th International Congress of Applied Chemistry, Section X; 1929, pp.
when micelles do not form such as, for instance, solutions of alco-
55–60 [London].
hols in non-polar liquids [12]. It can also explain ionization in a [20] L. Onsager, Report on revision of the conductivity theory, Trans. Faraday Soc.
solution with surfactant concentration below the Critical Micelle 23 (1927) 341–349.
Concentration (CMC). [21] R.M. Fuoss, Ionic association. I. Derivation of constants from conductance data,
J. Am. Chem. Soc. 79 (13) (1957) 3301–3303.
[22] R.M. Fuoss, C.A. Kraus, Properties of electrolytic solutions. III. The dissociation
References constant, J. Am. Chem. Soc. 55 (1933) 1019–1028.
[23] C.A. Kraus, R.M. Fuoss, Properties of electrolytic solutions. 1. Conductance as
[1] J.B. Lawson, The adsorption of non-ionic and anionic surfactants on sandstone influenced by the dielectric constant of the solvent medium, J. Am. Chem. Soc.
and carbonate, in: SPE 7052, SPE Symposium on Improved Methods of Oil 55 (1) (1933) 21–36.
Recovery, 16–17 April, Tulsa, Oklahoma, USA, 1978. [24] ISO 20998 – Part 1, Measurement and characterization of particles by acoustic
[2] Xianmin Zhou, Ming Han, Alhasan B. Fuseni, Ali. A. Yousef, Adsorption- methods. Part 1 - Concepts and procedures in ultrasonic attenuation
desorption and relations of an amphoteric surfactant onto permeable spectroscopy, 2006.
carbonate rocks, in: SPE 153988, SPE Symposium on Improved Methods of [25] A.S. Dukhin, S. Swasey, M. Thommes, A method for pore size and porosity
Oil Recovery, 14–18 April 2012, Tulsa, Oklahoma, USA. analysis of porous materials using electroacoustics and high frequency
[3] M. Mirzaei, D.A. DiCarlo, G.A. Pope, Visualization and analysis of surfactant conductivity, Colloids Surf. (2013).
imbibition into oil-wet fractured cores, Soc. Petrol. Eng. J. 21 (1) (2016). [26] A.S. Dukhin, J.P. Goetz, Characterization of Liquids, Dispersions, Emulsions and
[4] G. Hirasaki, C.A. Miller, M. Puerto, Recent advances in surfactant EOR, Soc. Porous Materials using Ultrasound, Elsevier, 2017. 571 pages.
Petrol. Eng. J. 16 (4) (2011). [27] N.O. Mchedlov-Petrossyan, I.N. Palval, A.V. Lebed, E.M. Nikiforova, Association
[5] I.D. Morrison, Electrical charges in non-aqueous media, Colloids Surf. A 71 of the picrate ion with cation of various nature in solvents of medium and low
(1993) 1–37. relative permittivity. An UV/Vis spectroscopic and conductometric study, J.
[6] A.S. Dukhin, P.J. Goetz, How non-ionic ‘‘electrically neutral” surfactants Mol. Liq. 145 (2009) 158–166.
enhance electrical conductivity and ion stability in non-polar liquids, J. [28] S.M. Nelson, R.C. Pink, Solutions of metal soaps in organic solvents. Part III. The
Electroanal. Chem. 588 (2006) 44–50. aggregation of metal soaps in toluene, isobutyl alcohol, and pyridine, J. Chem.
[7] Q. Guo, V. Singh, S.H. Behrens, Electric charging in non-polar dispersions due to Soc. (1952) 1744–1750.
non-ionazable surfactants, Langmuir 26 (5) (2010) 3203–3207. [29] M.B. Mathews, R.W. Mattoon, J. Chem. Phys. 17 (1949) 496–497.
[8] C.E. Espinosa, Q. Guo, V. Singh, S.H. Behrens, Particle charging and charge [30] H.F. Eicke, M. Borkovec, B. Das-Gupta, Conductivity of water-in-oil
screening in nonpolar dispersions with non-ionic surfactants, Langmuir 26 microemulsions: a quantitative charge fluctuation model, J. Phys. Chem. 93
(22) (2010) 16941–16948. (1989) 314–317.
[9] S. Poovarodom, J.C. Berg, Effect of particle and surfactant acid-base properties [31] D.G. Hall, Conductivity of microemulsions: an improved fluctuation model, J.
on charging of colloids in apolar media, J. Colloids Interf. Sci. 346 (2010) 370– Phys. Chem. 94 (1990) 429–430.
377. [32] M. Gacek, D. Bergsman, E. Michor, J.C. Berg, Effects of trace water on charging
[10] M.M. Gacek, J.C. Berg, Effect of surfactant hydrophile–lipophile balance (HLB) of silica particles dispersed in nonpolar medium, Langmuir 28 (2012) 11633–
value on mineral oxide charging in apolar media, J. Colloids Interf. Sci. 449 11638.
(2015) 192–197.

You might also like