Solubilization power of surfactant-free microemulsions

Sebastian Schöttl a, Nobuyuki Matubayasi b and Dominik Horinek *ab
aInstitut für Physikalische und Theoretische Chemie, Universität Regensburg, D-93040 Regensburg, Germany. E-mail:
bDivision of Chemical Engineering, Graduate School of Engineering Science, Osaka University, Toyonaka, Osaka 560-8531, Japan

Received 1st June 2020 , Accepted 6th August 2020

First published on 6th August 2020

Hydrophobe solubilization concepts rely on the shielding of solutes from water in nonpolar domains comprising traditional surfactants. We show how an octanol/ethanol/water surfactant-free microemulsion solvates hydrophobic propane in small octanol/ethanol aggregates similar to traditional micelles. These aggregates have the comparable solvent quality as bulk octanol/ethanol with the same composition.

Solubilization in water is frequently a key problem in science and technology. Biological processes frequently involve poorly water-soluble matter while still occurring in a water-based environment. In chemistry, water offers the potential of being an environmentally friendly reaction medium, but this requires that the reactants are soluble in water. Typical colloidal systems like microemulsions are well known to serve this purpose, but come with several disadvantages. Traditional ionic or nonionic surfactants can be hard to remove at the stage of purification, and environmental implications are typically problematic. Aside from surfactants, so-called hydrotropes like aromatic sulfonic acid salts are frequently used to mediate solubility.1 In this context, it has become increasingly evident in recent years that one key strategy to solubilization is the formation of a microheterogeneous environment.2 Contrary to traditional colloidal structures that are formed by surfactant molecules, these solutions involve weak aggregates.3 A type of such weakly structuring solutions are surfactant-free (or ultraflexilble) microemulsions (SFMEs).4 SFMEs are mixtures of relatively simple compounds that show structures in close similarity to microemulsions, but are free of traditional surfactants. In this communication, we discuss a strategy for hydrophobe solubilization in aqueous media based on SFMEs5 and illustrate the underlying solution thermodynamics using propane as a testcase hydrophobic species.

Propane is poorly water-soluble: our simulation setup predicts a very unfavorable hydration free energy, which is a measure for the free energy change during a transfer from vacuum, of 13.8 kJ mol−1 for propane. This somewhat overestimates the experimental value of 8.3 kJ mol−1;6 see the methods part for more details. The hydrophobic effect as the origin of this poor solubility is well established. A better solvent for propane comprises molecules with nonpolar parts: solvation of propane in octanol is favorable with a predicted solvation free energy of −10.6 kJ mol−1, which again overestimates the experimental result of −5.0 kJ mol−1. Using octanol for mediating solubility of propane in an aqueous solution by acting as a hydrotrope or surfactant is not directly possible, because octanol itself has a very poor solubility in water. Despite its amphiphilic molecular architecture, it does not form micelles or micelle-like aggregates in water. Very different are surfactant-free microemulsions that are formed in ternary systems that contain ethanol in addition to water and octanol. Ethanol facilitates the mutual solubility, thus showing hydrotropic action for facilitating octanol solubility in water. In this SFME, it has been recently shown that micelle-like aggregates are present.7 An in-depth analysis of the aggregates shows the existence of an octanol-rich core and an interfacial layer, where ethanol is enriched.8 Thus, these aggregates are good candidates for hosting hydrophobic compounds to be solubilized. Therefore, we choose a composition where a direct, water-continuous SFME is present. Using Tanford's pseudo-phase concept, the system separates into a polar pseudo-phase with mole fractions xethanol = 0.17 and xwater = 0.83 and a nonpolar pseudo-phase with xethanol = 0.84 and xoctanol = 0.16.

The solvation free energy of propane in pure ethanol is calculated to moderately favorable −3.7 kJ mol−1. Yet, solubilization by plain addition of ethanol does not present a viable approach: a bulk system with the composition of the polar pseudo-phase is still a poor solvent for propane with a solvation free energy of 7.5 kJ mol−1. In contrast, a bulk system with the composition of the nonpolar pseudo-phase is a good solvent for propane with a solvation free energy of −6.7 kJ mol−1.

Here, we address how the solvation free energy is affected by the local environment in the microheterogeneous SFME, where octanol aggregates of different size are present. This question is difficult to tackle. The usual umbrella sampling approach would rely on distance restraints to the center of some aggregate. However, the timescale of the aggregation dynamics is very fast. In such situations, the restraint is difficult – though not impossible – to introduce. Since the aggregates are transient, an additional bias may be needed to prohibit the decay of the aggregate. We avoid this difficulty by performing unbiased simulations of the propane/SFME system and analyzing them by using the energy representation method (ERmod).9–11 This method has been successfully applied to analyze solubilization in a sodium dodecyl sulfate micelle12 and is – due to its fast convergence – applicable to the flickering SFME aggregates; the free-energy estimation is possible only with the conditional sampling in terms of the propane-aggregate distance. Fig. 1a shows the total solvation free energy and the contributions of the three solvent components of propane relative to the center of mass of an aggregate comprising 25 octanol molecules. The radial distributions of the three components of the SFME in Fig. 1b meanwhile give an indication of the local variations in composition. The free-energy calculations were performed in 9 radially resolved concentric regions of 0.5 nm width. The solvation free energy shows a favorable solvation free energy inside the aggregate, −5 kJ mol−1, which is in fact only slightly smaller than the propane solvation free energy in the similarly composed binary system. Outside of the aggregate, the solvation free energy is approximately 7 kJ mol−1, which is again close to the solvation free energy in the corresponding binary system. Fig. 1 also shows that the solvation free energy varies in parallel with the water contribution, which vanishes toward the core of the aggregate. This energetic feature is consistent with the structural insight discussed next.

image file: d0cp02933e-f1.tif
Fig. 1 (a) Solvation free energy of propane in an octanol/ethanol/water surfactant-free microemulsion as a function of the distance to the center of mass of a nonpolar aggregate comprising 25 octanol molecules. We stress that the error bars for the points in the bin closest to the center are very large, because the sampling efficiency of this region is very low. (b) Distribution of the main system components with respect to this coordinate, where the density ρ refers to the center of mass of octanol, ethanol, or water. The reference values ρ0 are the values of ρ at the center of the aggregate for octanol and the values of ρ in the region separated from the aggregate for ethanol and water.

The simulations demonstrate that (i) the presence of the propane molecule does not interfere with the SFME's molecular-scale aggregation and that (ii) the free-energy change that occurs during the solvation of the hydrophobic molecule is largely determined by its local environment. We further clarify this by analyzing the solvation shell of the propane molecule as a function of the position relative to the aggregate center of geometry. Fig. 2 shows that inside the aggregate the solvation shell is dominated by octanol, and solvation shell water molecules are absent. The water contribution to the solvation free energy is insignificant in this case. Outside the aggregate, the propane solvation shell is dominated by water and ethanol. However, a small fraction of solvating octanol is still present, and accordingly, the solvation free energy contribution of octanol is very small, but not zero. This, however, is also in part a consequence of our analysis, in which the fact that the propane is outside the analyzed aggregate does not exclude that the molecule might be close to another aggregate. Since this effect is small, it does not hamper the conclusions we can draw from our study.

image file: d0cp02933e-f2.tif
Fig. 2 Composition of the first solvation shell of a propane molecule in the octanol/ethanol/water surfactant-free microemulsion as a function of the distance to the center of mass of a nonpolar aggregate comprising 25 octanol molecules.

A main conclusion of the analysis of the solvation free energy is that the primary effect of the aggregate is the protection of propane from the unfavorable hydrophobic solvation. While the ethanol and octanol contributions are stronger inside the aggregate than outside, the strength of the hydrophobic solvation effects leaves the direct octanol and ethanol contributions as secondary contributions. The poorest solubility of propane is to be assumed at a separation of 2–3 nm from the aggregate, where consistently sampling is deficient. This leads to substantial error margins that mask actual position and magnitude of the conjectural maximum. This region of unfavorable solubility coincides with the zone of lowest octanol content that originates from the natural average spacing of aggregates, leaving mostly monomers as interaction partners.

The actual interfacial plane of aggregates of this size was determined to be located at a distance of approximately 1 nm from the geometrical center.8 Similar to the formation of a bilayer at an octanol–water interface between bulk phases, molecules in the aggregates are distinctly ordered with the polar headgroups oriented towards the aqueous pseudo-phase. However, octanol aggregates in the SFME are not big enough to form bilayer-like structures, and the interface is highly disordered and flexible and has some similarity to traditional micelles.13 A comparison of the planar interface and the aggregate structure is shown in Fig. 3. It might be considered that this outer shell of hydroxyl groups entails a clear response in the solvation free energy profile, yet no singular signal is observed close to the surface. Instead, the graph increases steadily towards the octanol-depleted region, which can be attributed to an accumulation of ethanol occurring at the interface5 that lowers the overall hydrophilic character of the boundary.

image file: d0cp02933e-f3.tif
Fig. 3 Octanol/water interface (a) between two bulk phases, (b) of an aggregate in a direct SFME. The images show 1 nm thick slices through the simulation boxes, where octanol alkyl tails are colored blue, the hydroxyl groups are accentuated as orange beads and the surrounding medium is sketched in line representation. A bilayer-like structure is present in (a), but not in (b).

The free energy profile in the SFME closely resembles free energies of transfer from bulk water to micelles, which was used to argue that micellar hydrophobic cores are free of water.12,14

The MD simulations were performed with GROMACS Pure octanol, pure ethanol, pure water, their three binary mixtures, and the ternary SFME were studied by MD simulations. Octanol/water forms a two-phase slab system, and a rectangular box with a cross section of 6.3 nm × 6.3 nm was chosen, all other systems are macroscopically single-phase systems that were simulated in cubic boxes. The ternary mixture forms an SFME, which is heterogeneous because of aggregate formation. The large system with 11.2 nm edge length, which is in line with previous simulations of the pre-Ouzo effect,7,16–18 ensures that the system is large enough to accommodate the SFME structures. Octanol molecules are described with the L-OPLS19 force field, ethanol molecules by standard all-atom OPLS,20 and water molecules by TIP4P/2005.21 Nonpolar hydrogens of the CH2 and CH3 groups were treated as virtual sites22 and all remaining bond lengths were fixed. Thus, the simulations are stable with an efficient time step of 5 fs. Electrostatic interactions were accounted for by smooth particle mesh Ewald summation (PME)23 with a k-space lattice spacing of 0.12 nm−1. The van der Waals forces were smoothly truncated between 0.9 nm and 1 nm. The temperature was set to 300 K via stochastic velocity rescaling and a Parrinello–Rahman barostat24 was used for pressure coupling at 1.0 bar. The time constant for both methods was 1.0 ps. The system was initially prepared in a randomly mixed state and simulated for at least 200 ns (see Table 1). All post-simulation analysis was made after neglecting the first 100 ns as relaxation time, which is long enough to allow initial relaxion even for the large SFME.

Table 1 Summary of used simulation system compositions, the simulation times t0/t1 (solvent/solution) in ns, and the solvation free energy (sfe) from ERmod in kJ mol−1
System N (oct) N (eth) N (wat) t0 t1 sfe
Wat 2228 200 200 13.8
Eth 1213 200 200 −3.7
Oct 368 200 200 −10.6
Eth/Wat 304 1500 200 200 7.5
Oct/Eth 100 525 200 200 −6.7
Oct/Wat 1351 6469 258
SFME 224 6366 25[thin space (1/6-em)]380 1000 1895

Aggregation of octanol in the SFME was judged based on a distance criterion as implemented in the GROMACS tool g_clustsize. Two molecules are considered to be part of the same cluster if the distance between two of their atoms (excluding the nonpolar hydrogens) is below 0.478 nm, similar to our previous work.7 In contrast to the binary solutions, the ternary SFME forms aggregates within the initial 100 ns. An aggregate size of 25 was chosen, as this size is close to the local maximum in the cluster size distribution that represents micelle-like aggregates.7 This implies that aggregates of this size appear frequently and that they contribute crucially to the properties of the SFME. Due to the short-lived nature of the aggregates, their presence had to be detected for every frame independently. Frames that contain an aggregate of size 25 ± 2 octanol molecules were chosen from the trajectory. Those frames were used for the ERmod analysis as described later.

The separation of the system into polar and nonpolar pseudo-phases requires the separation of the ethanol molecules in a fraction that belongs to the nonpolar aggregates and a fraction that belongs to the surrounding water molecules. This separation was realized by means of an interaction energy criterion, as previously described in ref. 7. For every pair of ethanol/octanol in the system, the Coulomb and Lennard-Jones interaction energies between these molecules were determined. Histograms were constructed for both interaction energies. Both the van der Waals and the electrostatic interaction histograms exhibit contributions from two populations of ethanol molecules: molecules with only weak interaction energies with octanols that do not belong to the nonpolar pseudo-phase and strongly interacting molecules that belong to the nonpolar pseudo-phase. Both populations are separated by a minimum in the histogram. This minimum value was chosen, as it allows for a natural differentiation by means of a cutoff criterion, which was determined as EvdW = −1.6 kJ mol−1 and ECoulomb = −11.0 kJ mol−1, respectively. Ethanol molecules displaying interactions stronger than the respective cutoff with any octanol molecule are classified as “bound” and are part of the nonpolar domain. The compositions of the two pseudo-phases were determined from the resulting numbers of Ethanol in the polar and nonpolar pseudophase.

The radial distributions of any subset of atoms around the geometric center of a cluster of given size, averaged over the trajectory, were determined with a self-written program. The space around the center was divided into concentric spherical shells of 0.5 nm width for each of which the number of atoms belonging to the considered group is evaluated.

The solvation free energy refers to the free-energy change accompanying the transfer of the solute from vacuum to the solution of interest or a specified region within the solution. This quantity was computed with the energy-representation method (ERmod)9–11,25 for a propane molecule in the five solvents mentioned above: water, ethanol, octanol, and binary ethanol/water and octanol/ethanol mixtures. For the sixth solvent, the SFME, the solvation free energy was determined as a function of the distance to the center of mass of a nonpolar aggregate comprising 25 octanol molecules. ERmod requires the simulation of an isolated propane, of the solutions of a propane molecule in the respective solvents, and of the solvents only without propane. Table 1 summarizes the used compositions and simulation times of the liquid systems. The simulation of the isolated propane was done for 1 ns. In addition, an interface between water and octanol was studied. In this simulation, no ethanol was present.

The frames in the MD trajectory of the SFME system were sorted into subsets according to the distance of the propane molecule from the aggregate's center such that each subset covers a 0.5 nm wide part of the propane-aggregate distance. ERmod requires two MD simulations in addition to the simulation of the solution: the solution system with the propane solute absent and the isolated solute. Using these trajectories, test particle insertions were done such that the desired 0.5 nm wide regions of propane-aggregate separation can be sampled.

The energy-representation method is a density-functional method to approach the solvation free energy approximately in combination with MD simulation. It is an endpoint method and does not refer to the intermediate states, which are necessary in free-energy perturbation and thermodynamic-integration methods. The method is thus a fast scheme for free energy, while the error due to the use of approximate functional is typically not larger than the error from the force field.25 See ref. 25 for the detailed formulation and representative applications. When solvation is examined in a mixed solvent, it is insightful to separate the effect of each solvent species (octanol, ethanol, or water). In the energy-representation formalism, the solvation free energy is formally written as a sum of the contributions from the respective solvents and the contribution from a solvent species is expressed as a functional of correlation functions of the solute with that solvent species. This can be a useful feature to discuss the separated roles of the solvent species, and the role of water was highlighted in Fig. 1(a).

During the ERmod calculations, no long-range corrections were applied. Using the correction, the ERmod result for the solvation free energy in water improves to 12.8 kJ mol−1. However, such a correction is not appropriate for inhomogeneous systems like an SFME. Since the inclusion of this term would consistently lower all results by a similar shift, the conclusions of this work are not affected.

The value of 12.9 kJ mol−1 is still higher than the experimental value of 8.3 kJ mol−1, but is a reasonable MD prediction. More accurate thermodynamic integration MD data based on different force fields yielded results between 7.9 kJ mol−1 and 13.0 kJ mol−1.26 Our ERmod approximation lies within this range.


In this work we show the feasibility of SFMEs for solubilization of hydrophobic compounds. This is similar to standard hydrotropic action that facilitates hydrophobic solvation in binary solvents like t-butanol/water mixtures,27 yet octanol with its longer nonpolar chain provides an even better environment for hydrophobes. Aggregates in water-continuous SFMEs resemble micelles formed by traditional surfactants and shield the nonpolar solute efficiently from the polar, water-rich pseudo-phase. However, octanol, in contrast to traditional surfactants, does not form micelles in binary aqueous solutions, and has to be solubilized itself by another hydrotrope, which is ethanol in our case. This mechanism, which has previously been labeled as “facilitated hydrotropy”,28 can only occur in the ternary SFME. The micelle-like aggregates offer an octanol-rich local environment that is favorable for hydrophobic solvation. Their environmental impact is lower than the one of traditional surfactant systems, and thus SFMEs have a vast potential in cases where hydrophobic solubilization is crucial. The solubilization mechanism can be seen as a type of facilitated hydrotropy, where one hydrotrope is solubilized in water by a second, more polar one. The solubilization enhancement of the dye sudane red observed in water/dimethyl isosorbide/benzyl alcohol29 or of vitamin B2 in water/caffeine/nicotineamide30 is likely an example of our discussed mechanism.

Conflicts of interest

There are no conflicts to declare.


DH thanks JSPS for financial support through an invitation fellowship (fellowship number L17523). NM is grateful to the Grant-in-Aid for Scientific Research (No. JP19H04206) from the Japan Society for the Promotion of Science and to the Elements Strategy Initiative for Catalysts and Batteries (No. JPMXP0112101003) and the Fugaku Supercomputer Project from the Ministry of Education, Culture, Sports, Science, and Technology.

Notes and references

  1. C. Neuberg, Biochem. Z., 1916, 76, 107 CAS .
  2. J. J. Booth, S. Abbott and S. Shimizu, J. Phys. Chem. B, 2012, 116, 14915 CrossRef CAS .
  3. W. Kunz, K. Holmberg and T. Zemb, Curr. Opin. Colloid Interface Sci., 2016, 22, 99 CrossRef CAS .
  4. S. Schöttl and D. Horinek, Curr. Opin. Colloid Interface Sci., 2016, 22, 8 CrossRef .
  5. M. L. Klossek, D. Touraud and W. Kunz, Phys. Chem. Chem. Phys., 2013, 15, 10971 RSC .
  6. R. Sander, Atmos. Chem. Phys., 2015, 15, 4399 CrossRef CAS .
  7. S. Schöttl, J. Marcus, O. Diat, D. Touraud, W. Kunz, T. Zemb and D. Horinek, Chem. Sci., 2014, 5, 2949 RSC .
  8. S. Schöttl, D. Touraud, W. Kunz, T. Zemb and D. Horinek, Colloids Surf., A, 2015, 480, 222 CrossRef .
  9. N. Matubayasi and M. Nakahara, J. Chem. Phys., 2002, 117, 3605 CrossRef CAS .
  10. N. Matubayasi and M. Nakahara, J. Chem. Phys., 2003, 118, 2446 CrossRef CAS .
  11. S. Sakuraba and N. Matubayasi, J. Comput. Chem., 2014, 35, 1592 CrossRef CAS PubMed .
  12. N. Matubayasi, K. K. Liang and M. Nakahara, J. Chem. Phys., 2006, 124, 154908 CrossRef PubMed .
  13. H. Wennerström and B. Lindman, Phys. Rep., 1979, 52, 1 CrossRef .
  14. Y. Chevalier and T. Zemb, Rep. Prog. Phys., 1990, 53, 279 CrossRef CAS .
  15. B. Hess, C. Kutzner, D. van der Spoel and E. Lindahl, J. Chem. Theory Comput., 2008, 4, 435 CrossRef CAS PubMed .
  16. T. Lopian, S. Schöttl, S. F. Prévost, S. Pellet-Rostaing, D. Horinek, W. Kunz and T. Zemb, ACS Cent. Sci., 2016, 2, 467 CrossRef CAS PubMed .
  17. T. N. Zemb, M. Klossek, T. Lopian, J. Marcus, S. Schöttl, D. Horinek, S. F. Prévost, D. Touraud, O. Diat, S. Marčelja and W. Kunz, Proc. Natl. Acad. Sci. U. S. A., 2016, 113, 4260 CrossRef CAS PubMed .
  18. S. Schöttl and D. Horinek, J. Chem. Phys., 2018, 148, 222818 CrossRef PubMed .
  19. S. W. I. Siu, K. Pluhackova and R. A. Böckmann, J. Chem. Theory Comput., 2012, 8, 1459 CrossRef CAS PubMed .
  20. W. L. Jorgensen, D. S. Maxwell and J. Tirado-Rives, J. Am. Chem. Soc., 1996, 118, 11225 CrossRef CAS .
  21. J. L. F. Abascal and C. Vega, J. Chem. Phys., 2005, 123, 234505 CrossRef CAS PubMed .
  22. K. A. Feenstra, B. Hess and H. J. C. Berendsen, J. Comput. Chem., 1999, 20, 786 CrossRef CAS .
  23. U. Essmann, L. Perera, M. L. Berkowitz, T. Darden, H. Lee and L. G. Pedersen, J. Chem. Phys., 1995, 103, 8577 CrossRef CAS .
  24. M. Parrinello and A. Rahman, J. Appl. Phys., 1981, 52, 7182 CrossRef CAS .
  25. N. Matubayasi, Bull. Chem. Soc. Jpn., 2019, 92, 1910 CrossRef CAS .
  26. N. Garrido and A. Queimada, J. Chem. Theory Comput., 2009, 9, 3033 Search PubMed .
  27. M.-E. Lee and N. F. A. van der Vegt, J. Chem. Theory Comput., 2007, 3, 194 CrossRef CAS PubMed .
  28. P. Simamora, J. M. Alvarez and S. H. Yalkowsky, Int. J. Pharm., 2001, 213, 25 CrossRef CAS PubMed .
  29. M. Durand, A. Stoppa, V. Molinier, D. Touraud and J. M. Aubry, J. Solution Chem., 2012, 41, 555 CrossRef CAS .
  30. M. P. Evstigneev, V. P. Evstigneev, A. A. H. Santiago and D. B. Davies, Eur. J. Pharm. Sci., 2006, 28, 59 CrossRef CAS PubMed .

This journal is © the Owner Societies 2020