Colloids and Surfaces A: Physicochem. Eng.

Aspects 232 (2004) 229–237

Development of a method for measurement of relative

solubility of nonionic surfactants
Jiangying Wu, Yuming Xu∗ , Tadeusz Dabros, Hassan Hamza
CANMET Energy Technology Center-Devon, Natural Resources Canada, Suite A202, 1 Oil Patch Dr., Devon, Alta., Canada T9G 1A8

Received 9 June 2003; accepted 30 October 2003

Abstract

Relative solubility number (RSN) provides a practical alternative to the HLB method of assessing hydrophilic–lipophilic balance of
surfactants. Traditionally, RSN has been determined by titration of a surfactant in benzene/dioxane solution with water. Owing to the toxicity
of these solvents, a new procedure for determination of RSN, using less-toxic substances, toluene and ethylene glycol dimethyl ether (EGDE),
was developed. The relationship between RSN and classic HLB values for some well-characterized classes of surfactants was investigated.
Some factors that might affect RSN value were also studied.
The results showed that RSN values determined using toluene/EGDE were linearly correlated with those determined using benzene/dioxane,
indicating that toluene/EGDE solvent can be used to replace benzene/dioxane. It was found that, within the same surfactant family, RSN
values determined at certain molar concentration showed a good linear relationship with classic HLB values. The surfactant concentration in
RSN titration solvent affected the RSN values; the effect of salt concentration in the titration water was negligible. A generalized regression
model was used to correlate RSN values with the structure parameters of surfactants such as carbon number in hydrophobic groups, C–O
number and free OH number.
Crown Copyright © 2003 Published by Elsevier B.V. All rights reserved.

Keywords: Surfactant; Nonionic surfactant; Relative solubility number; HLB

1. Introduction It denotes the relative solubilities of a surfactant in water

and in oil. Higher numbers indicate a more water-soluble
RSN has been widely used in the surfactant chemical in- product and lower numbers indicate a more oil-soluble
dustry to assist formulation, quality control, and product se- product. Products with RSN < 13 are generally insoluble
lection. In emulsion research, RSN is useful in the selection in water. Products with RSN values between 13 and 17
of stabilizers and demulsifiers. For example, Márquez-Silva are dispersible in water at low concentrations and form
et al. used RSN as a demulsifier characteristic to study the gels at high concentrations. Products with RSN > 17 are
dehydration of crude oil [1]. In their work they investigated considered soluble in water.
the relationship between dehydration performance and the Surfactant manufacturers have attached considerable im-
characteristics of oil, associated water, and demulsifier. As portance to RSN because they think it can be used to reflect
a result, they proposed an empirical equation correlating the changes in the hydrophilic and lipophilic characteristics of
acidity number of crude oil, water salinity, and the RSN of nonionic surfactants. For example, Márquez-Silva et al. con-
demulsifier. sidered RSN as a measure of the combined affinity for the
The standard RSN value is empirically defined as the polar (hydrophilic) and nonpolar (lipophilic) components of
volume in milliliters of distilled water necessary to produce the surfactant [1]. It is known that the relative affinities of
persistent turbidity in a benzene/dioxane solvent system surfactant for the oil and water phases play an important
consisting of 1 g of surfactant sample and 30 ml solvent [2]. role in emulsion stability. Previous research demonstrates
that when the ratio of interaction of the surfactant with water
to its interactions with oil is near or equal to unity, Winsor
∗Corresponding author. type III systems are obtained [3], in which the rate of phase
E-mail address: yxu@nrcan.gc.ca (Y. Xu). separation of emulsion is at a maximum [4,5]. Therefore, a

0927-7757/$ – see front matter. Crown Copyright © 2003 Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.colsurfa.2003.10.028
230 J. Wu et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 232 (2004) 229–237

comprehensive understanding of RSN is helpful in selecting laboratory test results aimed at developing a procedure us-
the most effective demulsifier. ing two less-toxic compounds, toluene and ethylene glycol
Since the 1950s, hydrophile–lipophile balance (HLB) dimethyl ether (EGDE). We also attempted to establish a
value has been widely accepted as a unique parameter to relationship between RSN and HLB and correlate RSN
represent the affinity balance between hydrophilic and hy- with the molecular structure of the surfactants investigated.
drophobic groups of nonionic surfactants [6,7]. From the
definitions of RSN and HLB, it can be seen that the two
numbers measure similar properties. For instance, higher 2. Experimental
RSN or HLB values indicate higher solubilities of surfactant
in water. HLB can be determined by experimentation and 2.1. Materials
from a number of simple equations that permit calculations
Five surfactants supplied by Champion Technology Ltd.
for certain types of nonionic surfactant [8]. However, the
(Houston, Texas) were used as our RSN standards. The RSN
applicability of these equations is limited. Furthermore, in
values for these five surfactants were provided by Champion
many industrial applications it is difficult to obtain sufficient
Technology and were determined using traditional titration
structural information to enable the calculation using the
in benzene/dioxane solvent. The RSN values for the five
equations. Experimental determination of the HLB value
standards covered the range from 5.7 to 19.6. Ethylene glycol
for a surfactant is complicated and subjective. Although
dimethyl ether (analytical grade), toluene (analytical grade),
dozens of experimental methods for determining HLB val-
and surfactants, listed in Table 1, were all purchased from
ues have been tried, two have been commonly used: phase
Aldrich Chemicals and used as received.
inversion temperature (PIT) [9,10] and emulsion inversion
point (EIP) [11,12]. However, unlike the RSN procedure,
2.2. Procedures for RSN measurement
these two methods are time-consuming and difficult to
apply in day-to-day industrial operations. Therefore, the
RSN was determined by titration of a surfactant in
RSN measurement is often the method of choice for both
toluene/EGDE solution with deionized water. The titration
research and industry.
was conducted using an auto titrator (Titrino, model 751
Although RSN values have been measured by many sur-
GPD) in a titration beaker furnished with a ground lid that
factant manufacturers and provided to customers as one
has two holes for inserting a colorimeter probe and a solvent
of the parameters characterizing a surfactant, not much
delivery tube. A colorimeter (Brinkman, model PC-920)
information on the procedure for determining RSN has
was used to detect the end point. The titration procedure is
been reported in the open literature. It is known that ben-
as follows:
zene/dioxane was the most commonly used RSN solvent,
but the percentage of benzene in dioxane varied between (1) Premix toluene/EGDE at the desired volume ratio to
companies, which ultimately resulted in discrepancies be- obtain stock RSN titration solvent.
tween reported RSN values. Furthermore, dioxane and (2) Weigh the desired amount of surfactant sample into the
benzene are extremely toxic chemicals. Therefore, it is beaker. Add the desired amount of RSN titration solvent
desirable to develop a procedure that uses less-toxic com- to the beaker and mix with a magnetic stirring bar until
pounds for RSN determination. An improved understanding all surfactant is completely dissolved. (The amount of
of the fundamentals of RSN would also enable more ef- surfactant and solvent added will be discussed in detail
fective use of this parameter. In this paper, we report our in the following sections.)

Table 1
Surfactants used in this study
Surfactant Nominal structure HLBa MWa Surfactant Nominal structure HLBa MWa

Brij30 PEO(4) lauryl ether 9.7 362 Igepal CO 890 PEO(40) nonylphenyl ether 17.50 1982
Brij35 PEO(23) lauryl ether 16.9 1198 Igepal CO 990 PEO(100) nonylphenyl ether 18.90 4626
Brij52 PEO(2) cetyl ether 5.3 331 Tween 20 PEO(20) sorbitan monolaurate 16.7 1228
Brij56 PEO(10) cetyl ether 12.9 683 Tween 40 PEO(20) sorbitan monopalmitate 15.6 1284
Brij72 PEO(2) stearyl ether 4.9 359 Tween 60 PEO(20) sorbitan monostearate 14.9 1312
Brij76 PEO(10) stearyl ether 12.4 711 Tween 80 PEO(20) sorbitan monooleate 15 1310
Brij78 PEO(20) stearyl ether 15.3 1152 Tween 85 PEO(20) sorbitan trioleate 11 1839
Brij98 PEO(20) oleyl ether 15.3 1150 Span 20 sorbitan monolaurate 8.6 346.5
Igepal CA 210 PEO(2) octylphenyl ether 4.80 294.4 Span 40 sorbitan monopalmitate 6.7 402.6
Igepal CA 520 PEO(5) octylphenyl ether 10.00 426.6 Span 60 sorbitan monostearate 4.7 430.6
Igepal CA 720 PEO(12) octylphenyl ether 14.60 735 Span 80 sorbitan monooleate 4.3 428.6
Igepal CO 210 PEO(2) nonylphenyl ether 4.6 308.5 Span 85 sorbitan trioleate 1.8 957.5
Igepal CO 720 PEO(12) nonylphenyl ether 14.20 749
a HLB value and average molecular weight are provided by Aldrich.
 J. Wu et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 232 (2004) 229–237 231

(3) Titrate deionized water into the surfactant solution until

the solution becomes persistently turbid. The end point
is considered as the point at which the transmittance
is reduced to 80% of original signal and remains at or
below that value for 1 min or longer.
(4) Record the volume (in ml) of deionized water titrated
in at the end point. This volume is considered the RSN
value for that surfactant.

3. Results and discussion

3.1. Toluene/EGDE solvent system Fig. 1. RSN values measured with toluene/EGDE at different compositions
vs. RSN values measured with benzene/dioxane.
In order to be suitable for RSN titration, the selected
solvent system must produce titration values that reflect
the change in its hydrophilic–lipophilic property for vari- ues for all eight solutions are listed in Table 2. It can be
ous surfactants. To validate our new toluene/EGDE titra- noted from Fig. 1 and Table 2 that the slope of the regres-
tion system, five surfactants with known RSN values were sion line decreases monotonically with increasing toluene
used for titration in each trial solvent. The titration val- concentration, indicating that the solvent system is more
ues obtained from the new solvent system should not be sensitive to RSN values at very low toluene concentrations.
expected to be the same as those obtained from standard However, the RSN values obtained in these low-toluene sol-
solvent because of differences in solvent properties, how- vents are very different from those obtained using the stan-
ever, the RSN values from the two solvent systems must be dard method. In order to make the RSN data from the new
correlated. solvent system as comparable to the standard data as pos-
In the standard RSN titration procedure using ben- sible, a slope in Fig. 1 equal to or near unity would be
zene/dioxane solvent, 1 g surfactant was added to 30 ml ideal.
solvent. To make our procedure as consistent as possible Two criteria were used for selecting the best solvent com-
with the standard procedure, we adopted a similar concen- position: correlation coefficient (R2 ) of the linear regression
tration in our first series of experiments, i.e. 1 g surfactant in and the sum of the squares of the differences between the
30 ml solvent. For convenience, we use the term RSN(B/D) five
measured RSN values and the
standard RSN values,
in this work to represent RSN values obtained from the i.e. [RSN(T/E)−RSN(B/D)]2 or (RSN)2 . The former
dioxane/benzene system and RSN(T/E) for values obtained measures how well the new RSN(T/E) values are corre-
from our toluene/EGDE system. Since the toluene-to-EGDE lated to the standard RSN values, while the latter measures
ratio directly affects the RSN measurement results, several how RSN(T/E) values differ from the standard values. The
solvents with different toluene-to-EGDE ratios covering the regression equations, correlation coefficients

(R2 ), and the
range from 1 to 5% (v/v) were tested. sums of the RSN difference squares [ (RSN)2 ] for all
To validate the new RSN procedure, RSN(T/E) values eight solvent systems are listed in Table 2. Furthermore, we
were plotted against RSN(B/D) values. Although we tested plotted the regression correlation coefficient and sum of the
the solvents using eight different concentrations of toluene, difference squares as functions of toluene percentage in the
for the sake of clarity, only three sets of data are shown solvent, as shown in Fig. 2. As the graph shows, a transition
in Fig. 1, i.e. 5.0, 2.6, and 1.0% (v/v). Fairly good lin- point can be seen at a toluene concentration around 2.6%
ear relationships were obtained between the RSN(T/E) and (v/v). The R2 values increase almost linearly as the toluene
RSN(B/D) values. Linear equations correlating the RSN val- concentration in the solvent decreases from 5 to 2.6% (v/v).

Table 2
Fitting equations, R-squared values, and the sums of RSN difference squares


Toluene in RSN Fitting equation Correlation coefficient (R2 ) (RSN)2
solvent (%, v/v)
5 RSN(T/E) = 0.5591RSN(B/D) + 2.5684 0.8729 49.4
4 RSN(T/E) = 0.693RSN(B/D) + 1.849 0.8935 27.2
3 RSN(T/E) = 0.9209RSN(B/D) + 0.3344 0.9354 6.2
2.7 RSN(T/E) = 1.0343RSN(B/D) − 0.5099 0.9587 5.4
2.6 RSN(T/E) = 1.0895RSN(B/D) − 1.0569 0.9718 4.8
2.5 RSN(T/E) = 1.1842RSN(B/D) − 1.4359 0.9736 9.5
2.0 RSN(T/E) = 1.3997RSN(B/D) − 3.1108 0.9881 27.1
1.0 RSN(T/E) = 2.343RSN(B/D) −10.6 0.9895 277
232 J. Wu et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 232 (2004) 229–237

lower than RSN(B/D) provided by Akzo Nobel. It was also

noted that the differences between the RSN(T/E) values mea-
sured in this work and those obtained from benzene/dioxane
solvent for Akzo Nobel demulsifier are greater than those for
the Champion standard surfactants. This discrepancy is, in
our opinion, due to the difference in benzene concentrations
in the titration solvents used by Akzo Nobel and Champion.
Fig. 3 further shows that the RSN(T/E) determination pro-
cedure could be applied for industrial RSN measurements.
Based on findings discussed above, we chose solvent com-
prising 2.6% (v/v) toluene and 97.4% (v/v) EGDE as the
new standard solvent for all our RSN measurements. The
Fig. 2. Correlation coefficient and sum of RSN differences as a function RSN data reported below were all measured using this new
of toluene concentration in the titration solvent. standard solvent and from now on we will use RSN to de-
note RSN determined using the new solvent system.
At concentrations below 2.6% (v/v), the R2 values became
almost constant. It is interesting to note that, at concentra- 3.2. Mechanism of RSN test procedure
tions around 2.6% (v/v), the sum of the RSN difference
squares shows a minimum. Both the transition point and As mentioned earlier, RSN is a useful parameter related to
minimum point occur at similar solvent compositions. This the hydrophilic–hydrophobic balance of nonionic surfactant.
clearly demonstrates that there is an optimal solvent compo- It is well known that the appearance of turbidity from a clear
sition, which should be in the range between 2.6 and 2.7% solution indicates a phase separation. In order to understand
(v/v) toluene. the phenomenon of the appearance of turbidity, a series of
The reproducibility of this procedure was determined by experiments were conducted to investigate the mechanism
repeating the RSN measurement for a selected surfactant, of the complex phase separation behind the RSN test pro-
Tween 80 (PEO(20) sorbitan monooleate), at the same con- cedure. In these experiments a small amount of oil-soluble
dition, i.e. solvent consisting of 2.6% (v/v) toluene and dye powder was added to the system. After the turbid solu-
97.4% (v/v) EGDE, 1 g surfactant in 30 ml solvent. A to- tion had been allowed to stand motionless for a few hours,
tal of seven independent measurements were conducted on it separated into two clear phases. On the top was a thin
different dates giving an average of 13.28 and standard de- layer with darker color, containing an elevated concentra-
viation of 0.059. As the data show, the reproducibility of the tion of the dye, and, therefore, more oil (toluene). This phe-
measurements was better than 1%. nomenon was observed for all surfactants tested regardless
In order to test the applicability of the new RSN proce- of the RSN values.
dure, we measured the RSN values for a number of demulsi- To gain a better understanding of phase separation behav-
fiers supplied by Akzo Nobel Surface Chemistry LLC. They ior, we conducted five additional titration tests in which the
also provided the RSN values for these demulsifiers. Their compositions in the two phases were analyzed. At the end
RSN values were measured using the benzene/dioxane sol- of titration the turbid system was allowed to separate into
vent system, but the benzene concentration in their solvent two phases, either by long-time settling or centrifugation,
was lower than that used by Champion Technology. Fig. 3 and samples from the top and bottom phases were collected
compares the RSN(T/E) values obtained using our new pro- using a micro-syringe. The organic composition was deter-
cedure and those provided by Akzo Nobel. The graphs in- mined by gas chromatography and water content by Karl Fis-
dicate a reasonably good correlation between the two sets cher titration. Because the surfactant molecules did not elute
of RSN values, although RSN(T/E) values are consistently from the GC column, the surfactant concentration could not
be determined using GC. We also attempted other methods,
however, no reliable data were obtained. Therefore, only the
ratios of toluene/EGDE are reported. The results are listed
in Table 3.
Because the volume of the top phase is very small
(0.1–0.6 ml) compared to the volume of the bottom phase
(35–50 ml) the toluene-to-EGDE ratio in the bottom phase
remained almost constant, which was also confirmed by GC
analysis. As shown in Table 3, the toluene concentration
in the top layer is always higher than that in the bottom
phase, and depends upon the characteristics of the surfac-
Fig. 3. RSN values for demulsifiers supplied by Akzo Nobel Surface tant; more specifically, it increases with the RSN value of
Chemistry LLC. the surfactant. At the same time, the water content in the
 J. Wu et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 232 (2004) 229–237 233

Table 3
Composition of two phases from RSN titration
Surfactant H2 O titrated Toluene-to-EGDE Water content
(ml) or RSN ratio (wt.%)
Top Bottom Top Bottom
phase phase phase phase
None 20.7 0.104 ∼
=0.026 7.3 44.5
Tween 20 16.3 0.080 ∼
=0.026 13.6 37.7
Tween 80 13.3 0.064 ∼
=0.026 13.4 33.6
Span 20 9.8 0.055 ∼
=0.026 6.8 27.2
Span 85 4.2 0.046 ∼
=0.026 4.3 14.6

top phase is much lower than that in the bottom phase. The
water contents in both phases also increase with the RSN Fig. 4. Relationship between HLB and RSN values determined in 1 g per
30 ml solution.
value of the surfactant.
Bearing in mind the compositions of the two phases, we
can speculate on the mechanism of the RSN titration. It is Brij (PEO alkyl ether), Igepal (PEO alkylphenyl ether), Span
known that EGDE is miscible with both water and toluene, (sorbitan ester), and Tween (PEO sorbitan ester) families. In
but water and toluene are immiscible. When a small amount the first series of RSN measurements we applied the proce-
of water is titrated into the RSN solvent (toluene and EGDE) dure that uses 1 g surfactant in 30 ml solvent. The HLB and
without surfactant, EGDE dissolves both water and toluene. RSN values are shown in Fig. 4. It can be seen that there
However, when the water addition exceeds a certain amount, is no an overall correlation between RSN and HLB values.
the EGDE can no longer hold the toluene and water together, However, considerable correlations can be seen within in-
and new phase rich in toluene and low in water precipitates dividual surfactant families, especially for the Tween and
out. At this point the solution becomes turbid. When the Span families, in which very good linear relationships exist
turbid solution is allowed to stand, the precipitated low-water with R2 values around 0.97. The correlations for the Igepal
phase separates slowly from the bulk phase to the top due and Brij families are weaker.
to the density difference. It is believed that the deviation from a linear relationship
When the solvent contains surfactant, the hydrophobic was caused by the large differences between the molecular
group in the surfactant enhances the hydrophobicity of the weights of the surfactants. From the data in Table 1 we can
system, while the hydrophilic groups are hydrated with wa- see that the molecular weight for the Tween family varied
ter molecules. The ability of EGDE to dissolve water and from 1200 to 1800, and for the Span family, from 350 to 950.
toluene is altered by the presence of the surfactant. The However, molecular weight changed from 330 to 1200 for
amount of water that the solvent can tolerate depends on the Brij family and from 300 to 4600 for the Igepal family.
the hydrophobic–hydrophilic characteristics of the surfac- When the surfactant solutions contained the same mass (1 g)
tant. If the surfactant is more hydrophobic, the presence of of surfactant, the large variations in molecular weight re-
surfactant increases the hydrophobicity of the solvent and, sulted in significant differences in the number of molecules
therefore, less water is needed to cause the phase separation, in the solutions. Therefore, the use of molar concentration
resulting in a smaller RSN value. For the hydrophilic sur- appears to be called for instead of mass concentration in
factant, the surfactant molecules themselves bond to some the RSN determination. In the next series of measurements,
water molecules because of hydration and increase the vol- RSN values were determined at surfactant concentrations of
ume of water dissolved in EGDE. The more hydrophilic the 0.02 and 0.05 mol/l in 30 ml RSN solvent. The titration re-
surfactant, the more water is needed for hydration and, there- sults, shown in Figs. 5 and 6, confirm that when molar con-
fore, the higher the RSN value. From the explanation above, centration of surfactant is used, the RSN values show very
RSN can indeed be considered as a parameter that reflects good linear relationships with HLB values. The R2 values
the relative affinities of hydrophilic and hydrophobic groups for linear regressions between HLB and RSN for all four
in a nonionic surfactant. families of surfactants are over 0.96, except for the Brij fam-
ily at a concentration of 0.02 mol/l, where it is 0.896. These
3.3. Relationship between RSN and HLB results indicate that RSN values determined at the same
molar concentration represent the hydrophobic–hydrophilic
Since the HLB is a well known and widely used prop- balance more accurately. An empirical relationship between
erty and relative solubility numbers are easily measured, it RSN and HLB was established:
would be useful to find a relationship between them. Four HLB = A × RSN + B (1)
families of surfactants with known HLB values and known
chemical structures were selected to investigate the relation- where A and B are empirical constants varying with surfac-
ship between RSN and HLB. The surfactants included the tant family. Their values depend on characteristics of the
234 J. Wu et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 232 (2004) 229–237

cmc of nonionic surfactants with more general structure

parameters [16]. The correlations were not limited to a
homologous series but extended to a variety of nonionic
surfactants. In their work they used topological descriptors
that represent contributions from the size of the hydrophobic
group, the size of the hydrophilic group, and the structural
complexity of the hydrophobic group. A similar correlation
was performed on cloud points of surfactants by Huibers
et al. [17].
In this work we also attempted to correlate the RSN val-
ues to surfactant structure by adopting a scheme similar to
that used by Becher [14] and Ravey et al. [18]. We had RSN
Fig. 5. Relationship between HLB and RSN values determined in
0.05 mol/l solution.
values for 25 surfactants covering four surfactant families,
i.e. PEO alkyl ether, PEO alkylphenyl ether, sorbitan ester,
and PEO sorbitan ester. Unlike Becher and Ravey, we used
the following three structure parameters: carbon number in
hydrophobic group (C#), C–O number (CO#), and free OH
number (OH#). The C–O number includes all ethylene ox-
ide, C–O in ether, and double bond C=O in ester. The rea-
son for separating the free OH as an independent parameter
is that we believe the contribution of the free OH group to
the hydrophilic property is much greater than that of oxy-
gen in ether or ethylene oxide. This is also true when using
the group contribution method to calculate the HLB number,
in which the assigned number for free OH group is much
higher than that for the ethylene oxide group. The first pa-
rameter, C#, represents the hydrophobicity and the second
Fig. 6. Relationship between HLB and RSN values determined in two parameters, C–O# and OH#, measure the hydrophilicity
0.02 mol/l solution. of the nonionic surfactants.
The RSN values were correlated to the structure parame-
hydrophobic and hydrophilic groups. The constants A and B ters by multiple parameter polynomial regression using STA-
for the four families of surfactants studied in this work are TISTICA 6.0. In the regression all possible combinations of
summarized in Table 4. By using Eq. (1) and the constants the parameters were generated by the program and only the
reported in Table 4, HLB values of nonionic surfactants of statistically significant terms were maintained to achieve the
these families can be estimated from their measured RSN best regression. The following equation was obtained from
values using the procedures recommended in this paper. the regression for RSN values (0.02 M surfactant solution in
toluene/EGDE solvent):
3.4. Prediction of RSN values from molecular structure
RSN = A + B × CO# + C(CO#)2 + D × OH#
It is well known that surfactant properties such as critical + E(OH#)2 + F × C# (2)
micelle concentration (cmc), cloud point, HLB, etc., depend
on molecular structure. In previous research several empir- The regression coefficients and standard errors are listed in
ical relationships were developed to relate the surfactant Table 5. The coefficient A, 20.7, represents the RSN num-
properties to structure parameters such as ethylene oxide ber when the other parameters are all zero, which is the
number and alkane carbon number for some homologous experimentally observed volume of water titrated to blank
compounds [13–15]. Recently, Huibers et al. correlated toluene/EGDE solvent. Positive numbers for coefficients B

Table 4
Constants for empirical linear correlation between RSN and HLB (Eq. (1)) for four families of nonionic surfactants
0.05 mol/l 0.02 mol/l

A B R2 A B R2

Brij (PEO alkyl ether) 0.568 −2.0271 0.964 1.1019 −11.357 0.896
Tween (PEO sorbitan ester) 0.6029 6.7624 0.971 0.6495 5.9761 0.961
Span (sorbitan ester) 1.0236 −2.4167 0.967 0.8978 −2.7488 0.965
Igpel (PEO alkylphenyl ether) 0.8596 −9.3896 0.968 2.5189 −43.11 0.964
 J. Wu et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 232 (2004) 229–237 235

Table 5
Regression coefficients and S.E. for Eq. (2)
Coefficient Value S.E.

A 20.7 –
B 0.344 0.04
C −0.00287 0.0004
D 7.00 1.04
E −2.94 0.31
F −0.448 0.03

and D indicate that the RSN values increase with ethylene

oxide group and free OH group. The negative numbers for
Fig. 8. RSN values as functions of concentration in titration solution for
coefficients C and E (second-order parameter) are correc- the Brij family.
tions for the first-order parameters. Coefficient F is also neg-
ative, indicating the RSN value decreases with hydrophobic
chain length.
Eq. (2) was obtained from regression of all four fami-
lies of surfactants. It is not limited to a specific family and,
therefore, is a generalized model for at least these four fami-
lies. Using Eq. (2) one can predict the RSN number of other
surfactants in these families. Whether the equation is appli-
cable to other families requires further investigation. Fig. 7
shows the scatter plot for the calculated RSN values ver-
sus the experimentally measured RSN values. As the plot
shows, a reasonable prediction can be made using Eq. (2).

3.5. Effect of surfactant concentration Fig. 9. RSN values as functions of concentration in titration solution for
the Tween family.
It was noted earlier that the RSN changed when the sur-
factant concentrations in titration solution varied from 1 g
per 30 ml to 0.05 mol/l as seen in Figs. 4–6. This indicates
that the RSN values are also dependent on surfactant con-
centration in the titration solution. We investigated the ef-
fect of surfactant concentration in the titration solution on
the RSN values for all four surfactant families; the titration
results are presented in Figs. 8–11. The surfactant concen-
trations typically ranged from 0.01 to 0.1 mol/l. From the
graphs in Figs. 8–11 it can be observed that the degree of the
effect of surfactant concentration on RSN is dependent upon

Fig. 10. RSN values as functions of concentration in titration solution for

the Igepal family.

Fig. 7. Scatter plot for predicted vs. measured RSN values obtained in Fig. 11. RSN values as functions of concentration in titration solution for
0.02 M surfactant in toluene/EGDE solution. the Span family.
236 J. Wu et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 232 (2004) 229–237

the RSN value of the surfactant. For surfactants with RSN Table 6
greater than about 20, RSN increases with surfactant con- Effect of salt concentration in water on RSN value
centration. The greater the RSN value of the surfactant, the DI water 0.05 M NaCl 0.1 M NaCl
more rapidly RSN increases with concentration. For surfac- Brij30 19.30 18.50 18.19
tants with RSN less than about 20, the RSN value decreases Tween 20 16.85 16.40 16.34
when the surfactant concentration increases. The smaller the Igepol CO210 18.71 18.07 17.73
RSN value of a surfactant, the more rapidly it decreases.
The graphs also show that when the surfactant concentra- 3.6. Effect of salt concentration in titration water
tion approaches zero the RSN values approach about 21, as
indicated before. In the petroleum industry, the application of surfactant
The dependence of RSN on surfactant concentration can commonly involves salty water. Previous research has al-
be explained by the surfactant properties. As mentioned ready demonstrated that the salt concentration in water phase
earlier, the RSN value characterizes the hydrophobic and has very little effect on the performance of nonionic surfac-
hydrophilic properties of the surfactant. The higher the tants [19]. In order to confirm that the RSN values measured
RSN values, the more hydrophilic the surfactant. There- using this new procedure can be also applied to salt wa-
fore, more water is needed for the hydrophilic group to be ter systems, we conducted several RSN measurements us-
hydrated. On the other hand, the low-RSN surfactants are ing two NaCl solutions, 0.05 and 0.1 M, as titration water.
more hydrophobic. The increase in the surfactant concen- Three surfactants were chosen for the salty water titration at
tration increases the overall hydrophobicity of the system a surfactant concentration of 0.02 mol/l. The titration results,
and, therefore, reduces the volume of water required to displayed in Table 6, show that salinity has a very small ef-
titrate to phase separation. This result provides a further fect on measured RSN values. The RSN value decreases by
confirmation of the mechanism of the RSN test procedure about 1 ml when water salinity increases from 0 to 0.1 M.
as described in the previous section. This result agrees with the previous findings that the perfor-
The results shown in Figs. 8–11 also explain why the cor- mance of nonionic surfactant is hardly affected by the salt
relations between HLB and RSN for the Igepal and Brij fam- concentration in water.
ilies are less satisfactory than those for the Tween and Span
families when mass concentration is used in the titration, as
shown in Fig. 4. In the first two families, some surfactants 4. Conclusions
have high RSN values that are sensitive to the concentra-
tion. Different concentrations result in different RSN val- 1. The new RSN solvent system (EGDE and toluene) was
ues. For example, in the Igepal series, there are two points developed to replace the standard solvent system (ben-
(HLB 17.5 and 18.9) that are far from the regression line. zene and dioxane). The optimal toluene concentration in
These two points are actually for the two samples having the solvent is 2.6% (v/v). The RSN values measured us-
much larger molecular weights than the rest samples. At the ing the new solvent (2.6% (v/v) toluene and 97.4% (v/v)
same mass concentration their molar concentrations would EGDE) system correlated very well with those measured
be much lower than those of the other points. These two using benzene/dioxane.
points are also in the high-RSN range and are sensitive to 2. A generalized regression model was obtained for predic-
concentration. Therefore, these two factors (lower concen- tion of the RSN values for nonionic surfactants based on
tration and higher sensitivity to concentration) resulted in their structure parameters: carbon number in hydropho-
lower RSN values and led to deviations from the regression bic group, C–O number, and OH number.
line. If their concentrations were adjusted to molar concen- 3. RSN values measured using molar concentration of sur-
trations similar to those of the other points the RSN values factant show good linear correlation with HLB values.
would be much higher and would therefore much closer to Although there is no universal correlation between RSN
the regression line. For the Tween and Span families, the and HLB that covers all kinds of surfactants, good linear
RSN values for all surfactants tested in this work are smaller correlations exist within families of surfactants.
than 20 and the concentrations used are in the range that 4. Surfactant concentration in the titration solution af-
is insensitive to small changes in concentration. Therefore, fects measured RSN values. However, the degree of
both mass concentration and molar concentration gave sim- the concentration dependence varies with the surfactant
ilar correlation results, as shown in Figs. 4 and 5. property. For high-RSN (>20) surfactants, RSN values
Figs. 8–11 also imply that it is important to choose an increase with surfactant concentration. The higher the
appropriate concentration for RSN measurement. If the con- RSN value, the more rapidly the RSN changes with
centration is too low, the RSN values measured are not suf- the concentration. For low-RSN (<20) surfactants, RSN
ficiently sensitive to the characteristics of the surfactant. values decrease sharply with increasing surfactant con-
Therefore, sufficiently high concentration of surfactant in centration in the initial low-concentration range, but the
RSN solvent is recommended. However, the solubility of RSN values decrease only slightly with further increase
surfactant in the solvent may limit the concentration. in surfactant concentration.
 J. Wu et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 232 (2004) 229–237 237

5. For nonionic surfactant, salt concentration in titration [4] P.D. Berger, C. Hsu, J.P. Arendell, 1987, SPE 16285.
water has little effect on RSN determination. [5] M.A. Krawczyk, D.T. Wasan, C.S. Shetty, Ind. Eng. Chem. Res. 30
(1991) 367.
[6] W.C. Griffin, J. Soc. Cosm. Chem. 1 (1949) 311.
[7] W.C. Griffin, J. Soc. Cosm. Chem. 5 (1954) 249.
Acknowledgements [8] J.T. Davies, E.K. Rideal, Interfacial Phenomena, Academic Press,
New York, 1963.
The authors thank Champion Technology Ltd. for provid- [9] K. Shinoda, H. Saito, J. Coll. Interface Sci. 64 (1978) 68.
[10] H. Kunieda, N. Ishikawa, J. Coll. Interface Sci. 107 (1985) 122.
ing the surfactant samples used as our RSN standards and [11] L. Marszall, Coll. Polym. Sci. 254 (1976) 674.
Akzo Nobel Surface Chemistry LLC for supplying many [12] L. Marszall, Fette Seifen Anstrichm. 80 (1978) 289.
demulsifier samples that were also used in this work. [13] M.J. Rosen, J. Coll. Interface Sci. 56 (1976) 320.
[14] P.J. Becher, Dispersion Sci. Tech. 5 (1984) 81.
[15] M.J. Rosen, Surfactant and Interfacial Phenomena, 2nd ed., Wiley,
New York, 1989.
References [16] P.D. Huibers, V.S. Lobanov, A.R. Katritzky, D.O. Shah, M. Karelson,
Langmuir 12 (1996) 1462.
[1] R.L. Márquez-Silva, S. Key, J. Marino, C. Guzmán, S. Buitriago, [17] P.D. Huibers, D.O. Shah, A.R. Katritzky, J. Coll. Interface Sci. 193
1997, SPE 37271. (1997) 132.
[2] H.L. Greenwald, G.L. Brown, M.N. Fineman, Anal. Chem. 28 (1956) [18] J.C. Ravey, A. Gherbi, M. Stebe, J. Prog. Coll. Polym. Sci. 76 (1988)
1693. 234.
[3] P.A. Winsor, Solvent Properties of Amphiphilic Compounds, Butter- [19] L. Marszall, in: M.J. Schick (Ed.), Nonionic Surfactants—Physical
worths, London, 1954. Chemistry, Marcel Dekker, New York, 1987.