Pressure-Responsive, Surfactant-Free CO2-Based Nanostructured Fluids.

Microemulsions are extensively used in advanced material and chemical processing. However, considerable amounts of surfactant are needed for their formulation, which is a drawback due to both economic and ecological reasons. Here, we describe the nanostructuration of recently discovered surfactant-free, carbon dioxide (CO2)-based microemulsion-like systems in a water/organic-solvent/CO2 pressurized ternary mixture. "Water-rich" nanodomains embedded into a "water-depleted" matrix have been observed and characterized by the combination of Raman spectroscopy, molecular dynamics simulations, and small-angle neutron scattering. These single-phase fluids show a reversible, pressure-responsive nanostructuration; the "water-rich" nanodomains at a given pressure can be instantaneously degraded/expanded by increasing/decreasing the pressure, resulting in a reversible, rapid, and homogeneous mixing/demixing of their content. This pressure-triggered responsiveness, together with other inherent features of these fluids, such as the absence of any contaminant in the ternary mixture (e.g., surfactant), their spontaneous formation, and their solvation capability (enabling the dissolution of both hydrophobic and hydrophilic molecules), make them appealing complex fluid systems to be used in molecular material processing and in chemical engineering.

Microemulsions are macroscopically homogeneous, isotropic, and thermodynamically stable systems, containing at least three compounds: a polar one, usually water, a nonpolar one, usually oil, and an amphiphilic compound, usually a surfactant.[1] These macroscopically homogeneous systems are nanostructured in “oil-rich” and “water-rich” domains, which can show various size and shape, depending on the composition and the environmental conditions.[2] The structuration at the nanoscale entails other specific properties of microemulsions, such as ultralow interfacial tension,[3,4] large interfacial area,[5] and the ability to solubilize otherwise immiscible compounds.[1,2] As they contain both a polar and a nonpolar solvent, microemulsions have been considered as universal solvents. Accordingly, they are very appealing systems either as commercial products or as reaction media for many processes, ranging from nanoparticle templates to preparative confined organic chemistry.[1,6] Furthermore, as they form spontaneously, their preparation process is considered facile and of low cost. However, a significant disadvantage of conventional microemulsions is their reliance on the use of—sometimes highly concentrated and environmentally malign—surfactants, which are required in order to thermodynamically stabilize the system. This in turn, affects their sustainability both in terms of costs and environmental impact.

CO2-based microemulsions and surfactant-less or surfactant-free microemulsions have been extensively investigated as valuable greener alternatives to conventional microemulsions since the late 1970s, but only recently their structure has been investigated at the mesoscopic scale.[7−11] In CO2-based microemulsions, the oil phase is substituted by compressed CO2 (cCO2). Compared to organic solvents, cCO2 is low cost, nonflammable, environmentally benign, bio- and food-compatible, and naturally abundant. Generally, CO2 is a poor solvent, in particular for polar and/or high molecular weight solutes,[12] but when dissolved in water in the presence of specific surfactants, a water/CO2-based microemulsion is formed. Mainly, fluorinated or partly fluorinated surfactants have been used in order to form CO2-based microemulsions. However, they are expensive and environmentally unviable, and nowadays many efforts are made in the design of small-cheap-safe molecules able to stabilize the water/CO2 interface.[12−17] Water/CO2-based microemulsions have the attractive characteristics of cCO2 and the solvation properties of bulk water, being a universal solvent for many applications. Furthermore, because of the presence of cCO2, the properties of these unconventional microemulsions, such as density, and hence solvation power, can be strongly modified just by pressure variation.[11,500] The latter feature is very attractive both from the fundamental and technical point of view. This pressure responsiveness of the emulsion’s properties is significantly less pronounced in conventional microemulsions,[18] where in order to change the microemulsion’s properties, composition and/or temperature are the more appropriate tuning parameters.

In this context, we recently reported on the structuration of the single-phase mixture water/acetone/CO2 into “water-rich” and “water-depleted” regions at 10 MPa and 308 K. It has to be underlined that the structuration was observed in the single-phase region, far from the critical point of the ternary mixture.[19] Moreover, it has been already reported that this pressurized mixture shows the capability to solubilize non-water and non-CO2 soluble compounds, which are soluble in acetone and in CO2-expanded acetone, such as ibuprofen. Indeed, the addition of CO2 over an equimolar water/acetone mixture saturated with ibuprofen, where a solid phase and a liquid phase coexist at 10 MPa and 308 K, induces the solubilization of the precipitated CO2-phobic drug. This behavior can be explained if the addition of CO2 causes a structuration at the nanoscale in “water-rich” regions and in “water-depleted” domains, where the ibuprofen is dissolved (see SI for more details).[19,20]

The here studied structured single-phase system is referred to as “microemulsion-like system” rather than microemulsion,[19] as it is not clear yet whether or not the acetone takes over the role of the surfactant. In this paper, we report on the nanostructuration of differently composed water/acetone/CO2 mixtures, at various temperatures (298, 308, and 328 K) and pressures (10, 14, 18, and 22 MPa), by means of Raman spectroscopy. Moreover, we also describe the nanostructuration of pressurized mixtures of water/acetonitrile/CO2. Molecular dynamics (MD) simulations were performed to study the molecular nanostructuration with atomistic detail. The size of the “water-rich” domains in these pressurized mixtures was experimentally characterized by high-pressure small-angle neutron scattering (hp-SANS). By using hp-SANS we have also observed that the nanostructure is pressure responsive, in a reversible way. These single-phase, nanostructured fluids have several of the favorable properties of conventional microemulsions in terms of interfacial properties, solvent capability, and ability to act as templates (nanocontainers), with the advantages of being “green” and showing a reversible, pressure-responsive nanostructure. We expect that by pressure changes, we can instantaneously tune the extent of the fluid’s nanostructure, resulting in a rapid and homogeneous exchange of species between the multiple nanodomains, that is, mixing/demixing of the fluid. This can open not yet anticipated pathways in process engineering, making these nanostructured pressure-responsive fluids appealing systems to be used in molecular material processing and in chemical engineering.

Results/Discussion

The tendency of water molecules to aggregate in the different pressurized ternary mixtures of water/acetone/CO2 and water/acetonitrile/CO2 and to form a thermodynamically stable single-phase fluid with “water-rich” nanodomains embedded into a “water-depleted” matrix has been explored, using a range of complementary experimental and computational techniques.

Figure shows schematic Gibbs phase diagrams of the different ternary systems water/organic-solvent/CO2 at 10 MPa studied.

Figure 1

Schematic Gibbs diagrams of the ternary systems water/organic-solvent/CO2 based on data from literature (refs (21 and 22)) and data reported in Table . Solid colored lines show qualitatively the binodals of the system water/acetone/CO2 (10 MPa) at 298 K (blue), 308 K (green), and 328 K (red). For the system water/acetonitrile/CO2, the binodal has been measured experimentally at 308 K and 10 MPa and is shown by the dotted line (see SI). The gray arrows show the CO2 dilution paths of the initial “water/acetone” systems when CO2 is added to the mixture until the binodal. The solid black arrow shows the organic solvent dilution path when the initial mixtures are diluted by the addition of more organic solvent.

Table 1

CO2 Molar Fraction xCO (± 0.001) at the Binodal of the Ternary Mixture Water/Acetone/CO2a

		xCO2
P (MPa)	initial water/acetone molar ratio	298 K	308 K	328 K
10	50/50	0.158	0.170	0.192
	60/40	0.115	0.121	0.126
	75/25	0.051	0.056	0.061
14	50/50	–	0.182	–
18	50/50	–	0.190	–
22	50/50	–	0.198	–

When water/acetone mixtures of different initial molar ratios (50/50, 60/40, and 75/25) are diluted with CO2 either at 10 MPa but at different temperatures or at 308 K but at different pressures.

The colored solid curves were extracted from refs (21 and 22) and were partly measured (see Table ), showing qualitatively the binodals of the system water/acetone/CO2 at three different temperatures, which separate the single-phase region from the two-phase region. The dotted curve corresponds to the binodal of the system water/acetonitrile/CO2 at 308 K and 10 MPa, and it has been experimentally determined (see SI). The solid black arrow on the water/organic solvent (acetone or acetonitrile) axis shows the dilution path when an initial binary water/organic-solvent mixture (marked with a star) is diluted with additional acetone (or acetonitrile). The gray arrows show for 298 K the CO2 dilution paths when an initial binary water/acetone mixture is diluted with CO2 in the single-phase region (not exceeding the binodal). Depending on the nature of the organic solvent (acetone vs acetonitrile) as well as on the starting binary mixture composition or the temperature or pressure, the length of the CO2 dilution path in the single-phase region (before intersection with the binodal) varies.

Table lists the experimentally determined locations where the CO2-dilution path (gray arrows in Figure ), starting with different initial water/acetone mixture compositions, crosses the binodal, for the ternary mixture water/acetone/CO2 at various pressures and temperatures. With increasing temperature, the molar fraction of CO2 at the crossing point of the binodal and the CO2-dilution-path increases, while the fractions of acetone and water slightly decrease. All dilution paths were realized isothermally and isobarically in a variable-volume high-pressure chamber. With dilution, the total amount of water molecules in the variable-volume high-pressure chamber is constant, while the water concentration in “moles per liters” decreases when the volume of the chamber has to be varied in order to dilute isobarically.

Intuitively, if the water concentration decreases homogeneously, less hydrogen bonds between water molecules are formed. For example, the number of hydrogen bonds decreases monotonically, if an initial binary water/acetone mixture is diluted with additional acetone at 10 MPa and 308 K (black arrow mixing path in Figure ), as it has been demonstrated in Hankel et al.[19] and as it is shown in the Supporting Information (SI). On the contrary, if the dilution induces a nanostructuration into “water-concentrated” regions, containing many water molecules, and “water-less-concentrated” regions, containing few water molecules, the number of hydrogen bonds increases. Experimentally, this can be studied by the use of Raman spectroscopy; while diluting, we analyzed the frequency (energy) of the symmetric water stretch vibration band. From the shape of this Raman band, the development of hydrogen bonds can be extracted. In particular, we use the Raman shift v̅water in wavenumbers (cm–1) of the centroid of the symmetric water stretch vibration Raman broad band for the characterization of the development of hydrogen bonds, as described in the SI. The smaller v̅water is, the more hydrogen bonds are developed in the mixture.[24−27]

The Ternary System Water/Acetone/CO2

Figure shows the Raman shift of the centroid of the symmetric water stretch vibration band v̅water as a function of the water concentration cwater for the dilution of three different binary mixtures of water/acetone (molar ratio) 50/50, 60/40, and 75/25 (columns in Figure ) upon the addition of CO2 at 298, 308, and 328 K (lines in Figure ) at a pressure of 10 MPa. In each subfigure of Figure , the dilution of the initially binary water/acetone mixtures (right data points) with CO2 goes from large to small water concentration values (from right to left). The left data points are the last measurements made in the single-phase region. A further dilution with CO2 drove the mixture across the binodal into the two-phase region. Thus, all the data points shown in the figure correspond to mixture compositions in the single-phase region of the ternary system, before crossing the binodal. Each data point is the average of 200 single Raman measurements. The acquisition of one Raman spectrum took 5 s. The standard deviation of v̅water is given by the error bars. While the ordinate of each subfigure covers consistently a range of six wavenumbers, the absolute values of v̅water vary significantly from subfigure to subfigure. The smallest values of v̅water can be found at 298 K for the CO2 dilution of the initial 75/25 water/acetone mixture, which implies that at the lower temperatures and the larger water concentrations, many hydrogen bonds are developed. The largest values of v̅water can be found at 328 K for the CO2 dilution of the initial 50/50 water/acetone mixture, which implies that at the higher temperatures and the smaller water concentrations, less hydrogen bonds are developed.

Figure 2

v̅water as a function of the water concentration cwater when three initially binary water/acetone mixtures are diluted with CO2 at 298, 308, and 328 K at 10 MPa. Only one temperature (308 K) was analyzed for the initial 60/40 water/acetone binary mixture.

The data measured for 328 K show the behavior expected for a system that does not develop a nanostructuration. The Raman shift of the centroid of the symmetric water stretch vibration band v̅water increases monotonically with decreasing cwater, irrespectively of the composition of the initial water/acetone mixture. Therefore, at 328 K the development of hydrogen bonds gets less pronounced with increasing CO2 dilution. For 308 K and 298 K, the data rows show a maximum of v̅water. Starting from the initial binary mixtures, v̅water first increases with CO2 dilution, until the maximum of v̅water is reached, and then decreases with further CO2 dilution.

Therefore, at 308 K and 298 K, the development of hydrogen bonds with increasing CO2 dilution first declines to a minimum, which is the maximum of v̅water, and then rises. The rise of the development of hydrogen bonds with increasing CO2 dilution is due to the emergence of a nanostructuration of the ternary mixture in “water-concentrated” regions containing many water molecules and “water-depleted” regions containing few water molecules. This hypothesis will be further underlined by small-angle neutron scattering (SANS) experiments and by MD simulations. The overall change of v̅water during one CO2 dilution experiment increases with increasing temperature from 298 K to 308 K and with decreasing water content in the initial binary water/acetone mixture. For the CO2 dilution of the 75/25 initial binary water/acetone-mixture at 308 K, a maximum of v̅water is hardly detectable. Finally we want to underline that the CO2 dilution path starting from the initial 75/25 binary water/acetone mixture is the closest to the critical point of the ternary system. For this CO2 dilution path, we find the smallest change of v̅water as a function of the water concentration. Therefore, we can conclude that the nanostructuration we observed is different from critical fluctuations, which would reach their maximum at the mixture critical point.[26]

Figure shows the Raman shift of the centroid of the symmetric water stretch vibration band v̅water as a function of the CO2 molar fraction xCO for the dilution of initial 50/50 water/acetone mixtures at 308 K for pressures of 10, 14, 18, and 22 MPa. In contrast to Figure , in Figure we scale on the abscissa the CO2 molar fraction xCO instead of the water concentration, cwater. This assures a better comparability with the hp-SANS results that follow. Nevertheless, the CO2 molar fractions xCO are correlated with the corresponding water concentrations cwater on the left ordinate. As the dilution is realized by the addition of CO2, the CO2 dilution in Figure proceeds from left to right, meaning that the very left data points correspond to the binary water/acetone mixtures at 308 K at the different pressures before the addition of CO2. The very right data points represent the last measurement made in the single-phase region. A further addition of CO2 drove the system across the binodal into the two-phase region. As already mentioned above, the experimentally determined compositions of the binodal are shown in Table .

Figure 3

v̅water (right ordinate) as a function of the CO2 molar fraction xCO of the ternary mixture water/acetone/CO2 when binary water/acetone mixtures of an initial 50/50 molar relation are diluted with CO2 at 308 K and at 10, 14, 18, and 22 MPa. The water concentration as a function of the CO2 molar fraction is also given (left ordinate). Regions of nanostructuration are highlighted in blue.

In each subfigure in Figure the ordinate quantifying v̅water comprises six wavenumbers (cm–1). Irrespective of pressure (within the analyzed range), v̅water first increases when the system is diluted by the addition of CO2 (from left to right). Therefore, the development of hydrogen bonds first decreases until the maximum of v̅water is reached. With further addition of CO2 the system undergoes a nanostructuration in “water-concentrated” regions containing many water molecules and “water-depleted” regions containing few water molecules, which is expressed by the decrease of v̅water. The higher the pressure is, the larger CO2 molar fractions are required for the initialization of the nanostructuration (this is highlighted in Figure by blue shadows). A ternary mixture with a CO2 molar fraction of xCO = 0.15 at 10 MPa and 308 K (upper subfigure) is further in the region of nanostructuration than a mixture with the same composition but at higher pressure and the same temperature (lower subfigures). With increasing pressure, the minimum xCO value, for which nanostructuration is detectable (maximum of v̅water), moves to larger xCO values and at the same time also the xCO values that correspond to the binodal increase (see Table ).

The Ternary System Water/Acetonitrile/CO2

Figure shows the Raman shift of the centroid of the symmetric water stretch vibration band v̅water as a function of the water concentration for the dilution of two different binary mixtures of water/acetonitrile with 50/50 and 40/60 molar relationships. At 328 K v̅water increases monotonically with CO2 dilution, irrespective of the initial mixture composition. This indicates that with CO2 dilution, the hydrogen bonds develop less. At 298 K, v̅water decreases from the very beginning of the CO2 dilution, irrespective of the initial binary mixture composition indicating that initially more hydrogen bonds develop. The maximum development of hydrogen bonds is found for 298 K at the minimum of v̅water.

Figure 4

v̅water as a function of the water concentration cwater when two initially binary water/acetonitrile mixtures with 40/60 and 50/50 molar relationships are diluted with CO2 at 298, 308, and 323 K, at 10 MPa.

For the ternary systems water/acetone/CO2 and water/acetonitrile/CO2 at 10 MPa, we can summarize that the behavior of v̅water as a function of the water concentration cwater significantly depends on the temperature. At the highest analyzed temperature of 328 K, v̅water increased monotonically with CO2 dilution for each initial binary mixture composition. Whereas at the lowest analyzed temperature of 298 K, v̅water decreased with CO2 dilution either from the very beginning or after reaching a maximum. At the intermediate temperature of 308 K, the behavior of v̅water as a function of cwater is in between the behaviors identified for 298 and 328 K. With this observation together with the temperature sensitivity of the binodal curves shown in Figure and Table , we can deduce that, for a given initial binary mixture composition, the behavior of v̅water as a function of cwater depends on the length of the CO2 dilution path in the single-phase region, before phase separation, or in other words on the solubility (quantified as CO2 molar fraction) of CO2 in the ternary mixture. This statement is also underlined by the results shown in the context of the pressure variation (Figure ). Again we can deduct from Table that for the 50/50 initial binary water/acetone mixture, the behavior of v̅water as a function of xCO depends on the length of the CO2 dilution path in the single-phase region, before phase separation, or in other words on the solubility (quantified as CO2 molar fraction) of CO2 in the ternary mixture at various pressures. The lower the solubility of CO2 in the ternary water/organic solvent/CO2 mixtures is, the more pronounced is the extent of nanostructuration in “water-concentrated” regions and “water-depleted” regions (see Table ).

Dimensions and Shape of “Water-Rich” Nanodomains

SANS succeeds in probing structural heterogeneities over spatial scales on the order of several Å up to 100 nm and, as such, has been successfully exploited by several teams to characterize microemulsion systems over mesoscopic spatial scales.[2,27] Here, in order to get direct structural insight into the complex mesoscopic organization of the pressurized water/organic solvent/CO2 systems, we used a specially designed high-pressure SANS cell.[28]

In Figure , the hp-SANS data are presented for different mixtures water/deuterated acetone/CO2 at 308 K and 10 MPa, as a function of the dissolved amount of CO2 and as a function of the water concentration cwater. These data sets are plotted together with the pattern of the binary equimolar mixture water/deuterated acetone at the same experimental conditions (pressure, temperature). While the latter data set shows a small excess scattering with respect to the flat incoherent scattering due to hydrogen, in agreement with previous determinations,[29,30] upon CO2 addition a distinct increase of the scattering amplitude can be observed in the low Q range that is presently accessed (0.01 ≤ Q (Å) ≤ 0.1). This fingerprints the progressive development of structural heterogeneities with spatial extent of the order of a few nanometers. Related scattering studies that were focused on other surfactant-free microemulsion systems (e.g., water/ethanol/octanol)[7−10,31] highlighted comparable diffraction features in the low Q portion of the SANS patterns. The present hp-SANS data sets have been modeled in terms of the Ornstein–Zernike (O-Z) model that allows to estimate the characteristic size of the structural heterogeneities responsible for the excess low Q scattering, namely:where I(0) represents the normalized scattering intensity at zero angle and ξ is the spatial extent of the structural heterogeneities.[7]

Figure 5

SANS curves of water/deuterated acetone/CO2 mixtures, with xwater = xacetone = (1 – xCO)/2 and xCO = 0, 0.07, 0.12, and 0.14 (cwater = 11.50, 10.47, 9.91, 9.68 mol L–1) at 308 K and 10 MPa. The lines refer to fits of experimental data to the Ornstein–Zernike model, excluding the Q portion below 0.01 Å. The upturn might reflect larger structural heterogeneities that the present hp-SANS experiment does not provide access to. In the inset, the dependence of the characteristic size for the structural heterogeneities as a function of the CO2 molar fraction responsible for the low Q excess scattering is plotted.

In the inset of Figure , the CO2 content dependence of ξ is shown, indicating that when adding CO2 to the water/acetone equimolar mixture, the formation of segregated clusters (nanostructuration) occurs, whose spatial extent grows with CO2 content and reaches a size of approximately 2.5 nm at xCO = 0.14 (cwater = 9.682 mol L–1). This finding correlates well with the decrease of the centroid of the Raman band v̅water with CO2 dilution for water concentrations between 10.5 and 9.5 mol L–1 for the ternary system water/acetone/CO2 at 308 K and 10 MPa reported in Figure .

Molecular Dynamics Simulations

Using MD simulations, we can provide an atomistic interpretation of the experimental results. To this end, we performed all-atomic MD simulations of water/acetone and water/acetone/CO2 mixtures with eight different water concentrations at 10 MPa and 308 K (see Methods for technical details). In each MD simulation we started with binary mixtures of water and acetone (xwater = xacetone = 0.5). For the dilution with acetone we added more acetone molecules to the initial mixture, obtaining binary mixtures with decreasing molar fractions of water (black arrow in Figure ). For the CO2 dilution path, we added more CO2 molecules (gray arrow in Figure ). In this way we obtained simulations of ternary water/acetone/CO2 mixtures with different amounts of CO2, keeping constant the number of acetone and water molecules. Note that all simulations contain the same number of water molecules. In the first set of simulations, the water concentration decreases due to the addition of acetone, while in the second simulation set, the water concentration decreases due to the addition of CO2.

For all simulations we computed the number Nw–w of hydrogen bonds of each water molecule with other water molecules (water–water hydrogen bonds), and we divided this quantity by the result obtained for the initial water/acetone 50/50 mixture, Nw–w(0). The resulting ratio, RHbonds = Nw–w/Nw–w(0) is shown in Figure . As it can be seen in Figure , the dilution with acetone induces a decrease of RHbonds as the concentration of water decreases. On the contrary, in the case of dilution with CO2, we observe a nearly constant RHbonds, though the concentration of water decreases. In fact, a careful look at Figure shows a very weak dependence of RHbonds with water concentration in the ternary system, with a slight decrease of RHbonds to a minimum and a subsequent slight increase. The simulation results reported in Figure are in complete agreement with the interpretation of Raman experiments provided in the previous section and in our previous work.[19]

Figure 6

Evolution of the number of water–water hydrogen bonds in MD simulations at 308 K and 10 MPa after dilution of an initial water/acetone 50/50 mixture. We consider dilution by adding acetone (circles) or CO2 (triangles). RHbonds is defined as the ratio between the number of water–water hydrogen bonds observed in a particular simulation divided by the number of water–water hydrogen bonds obtained in the initial water/acetone 50/50 mixture which contains the same number of water molecules.

The nanostructuration postulated by the analysis of Raman and SANS experimental results can also be seen in our MD simulations. In Figure , we show simulation images for a ternary mixture with xCO = 0.15 and xwater = xacetone = 0.425. As it can be seen in Figure a, acetone molecules are homogeneously distributed over the whole system, but water is localized in pockets containing hydrogen-bonded water molecules. These water pockets induce inhomogeneities in the water density profiles (see Figure S7 in the SI), which have a typical length scale of ∼2 nm. Inside these water pockets, water molecules form hydrogen-bonded chain structures, as illustrated in Figure b (see also the pair correlation functions reported in Figure S8 in the SI). The average number of hydrogen bonded neighbors of a water molecule is 2. On average, we found that in these chain-like structures, 48% of the water molecules have two hydrogen bonds, and with other water molecules, 28% had only one and 24% more than 2 hydrogen bonds. Acetone also makes hydrogen bonds with water through the carbonyl group (Figure b), as demonstrated by the sharp peak in the acetone–water pair correlation function (Figure S8 in the SI). CO2 molecules avoid contact with water-concentrated pockets. The broad peaks in the CO2 correlation functions (Figure S8 in the SI indicates that CO2 molecules tend to be coordinated both with other CO2 molecules and acetone molecules). In Figure c we also show a typical configuration contributing to these broad peaks in the CO2–CO2 and CO2–acetone correlation functions. Typically, we found two or three CO2 molecules (with C–C distance about 4 Å) surrounded by acetone molecules (with a peak at a carbon–carbon separation of 4.4 Å). Water also tends to be anticorrelated with (effectively repelled by) CO2, up to distances of about 1 nm.

Figure 7

Images from MD simulations of the water/acetone/CO2 ternary mixture (10 MPa and 308 K, xCO = 0.15, xwater = 0.425). (a) Simulation snapshot. Water and CO2 molecules are shown as spheres with van der Waals radius. Acetone molecules (which form a network across the system) are shown as a surface calculated using the molecular surface solver of VMD28 with a 1.4 Å probe radius. The employed color code is red for O, white for H, and cyan for C. (b) A water-acetone hydrogen bond and a water hydrogen-bonded chain configuration extracted from (a) (all molecules are shown in CPK representation). (c) Detail of a pair of CO2 molecules (in van der Waals representation) surrounded by acetone molecules (shown in bonds representation) in a configuration extracted from (a). Hydrogen-bond distances are given in angstroms.

Pressure-Responsiveness of Nanostructured, Single-Phase Liquid

The hp-SANS measurements were performed at different pressures. For this study the ternary mixture water/deuterated acetone/CO2 with xCO = 0.15 (xwater = xacetone = 0.425) at 10 MPa and 308 K was used. The effect of the increase of pressure, from 10 to 30 MPa, was determined by a stepwise pressure increase in increments of 4 MPa. After that, the pressure is reduced back from 30 to 10 MPa, in order to rule-out any hysteresis phenomenon. The data are presented in Figure as a function of pressure. In particular, the description of the scattering data in terms of the O-Z model allows detecting a progressive decrease of the scattering amplitude upon pressure increase. The isothermal compressibility of the ternary water/acetone/CO2 mixture (xwater = xacetone = 0.43 and xCO = 0.14) as a function of pressure was measured and is shown in Figure S2a in the SI, aiming at addressing the source of scattering amplitude decrease. The ternary mixture has a lower isothermal compressibility than that of pure CO2[32] (4.1 × 10–2 to 5.5 × 10–3 MPa–1 for pure CO2 and 6.8 × 10–4 to 6.9 × 10–4 MPa–1 for the ternary mixture, respectively). In fact, the values of compressibility we measured for the ternary mixture at elevated pressure are between the corresponding compressibility values of pure water and pure acetone at ambient conditions. Moreover, the compressibility of the ternary mixture slightly increases with pressure (<2%), showing an opposite trend to that observed for pure CO2 (pronounced decrease, ca. 80%, Figure S2b in the SI). According to the relationship I(0)= nKBTκ[33] (where n is the number density of scatterers (heterogeneities), KB is Boltzmann constant, T is temperature, and κ is the isothermal compressibility), the low Q limit of the normalized scattering intensity I(0) should increase with increasing pressure, following the trend of the compressibility of the ternary mixture. In Figure , the opposite trend is observed. We can then rule out the effect of compressibility in affecting the amplitude of the low Q scattering and directly relate it to the decreasing average size of the segregated domains with increasing pressure (see inset of Figure ), which is proportional to the scattering domains’ volume. All of these observations reveal that these single-phase fluids feature a pressure-responsive extent of nanostructuration. This finding allows proposing pressure as an effective tool to affect the segregated morphology in these systems. Similar conclusions were drawn from the results shown in Figure , where we saw that the higher the pressure is, the more CO2 is needed for the appearance of the nanostructuraction. This, in other words, means that at equal composition (i.e., xwater = xacetone = 0.425 and xCO = 0.15) and temperature, the extent of the nanostructuraction is tunable by pressure.

Figure 8

SANS curves for the system water/deuterated acetone/CO2, with xwater = xacetone = 0.425 and xCO = 0.15 at 308 K and different values of pressure in the range between 10 and 30 MPa. The lines refer to fits of experimental data to the Ornstein–Zernike model, excluding the Q range below 0.01 Å–1. In the inset, the pressure dependence of scattering amplitude and characteristic size for the structural heterogeneities are shown.

Furthermore, hp-SANS was used to monitor the reversibility of the formation of the nanostructuration upon cyclic pressure jumps. In Figure , we show two different data sets at 14 and 30 MPa that were obtained by raising the pressure (14 → 30 MPa in several steps), then decreasing (30 → 14 MPa in several steps) it, and then raising (14 → 30 MPa in one step) it again. As it can be seen, the two data sets obtained at the same nominal pressure nicely lay on top of each other, thus providing experimental support to the reversibility of the extent of nanostructuration upon pressure modulation.

Figure 9

SANS curves for system water/deuterated acetone/CO2, with xwater = xacetone = 0.425 and xCO = 0.15 at 308 K and 14 or 30 MPa. The different data sets at a given pressure refer to different routes used to reach these pressure values, aiming at confirming the lack of process hysteresis in the formation/decomposition of nanoaggregates.

Conclusions and Future Perspectives

The combination of Raman spectroscopy, hp-SANS, and MD simulations provides valuable complementary insights into the appearance of “water-rich” nanodomains within the pressurized, single-phase mixtures of water/organic solvent/CO2, both in the case of acetone and acetonitrile being the organic solvent. Raman spectroscopy reveals the occurrence of water molecules aggregation, that is, the formation of “water-concentrated” regions, and clarifies how the thermodynamic conditions, that is, composition, temperature, and pressure, affect this phenomenon. hp-SANS measurements provide information about the characteristic size of these “water-rich” nanodomains, and MD simulations clarify how the molecules are localized within the ternary mixture and supports experimental findings. The final outcome is that these ternary mixtures water/organic solvent/CO2 can turn into a surfactant-free microemulsion-like system, where “water-concentrated” nanodomains, with a characteristic size of the order of 2 nm, are embedded in a “water-depleted” matrix. The aggregation of water molecules occurs when CO2 is added to a binary mixture of water/organic solvent. hp-SANS measurements show that the nanostructure of these surfactant-free fluids can be reversibly tuned by pressure variations. We can distinguish between an “ON” state, where the nanostructuraction is occurring at a large extent, and an “OFF” state, where by progressively increasing the pressure (above 10 MPa), the medium remains macroscopically homogeneous, but the structural heterogeneities start to drastically decrease in their extent (see the abstract graphic). This reversible, pressure responsiveness is completely missed in conventional microemulsions, where changes in their nanoscopic structure require temperature and/or compositional variations, which propagate much slower through the mixture, thus making jumps in the activating parameters (temperature or concentration) less efficient and/or less homogeneous. We expect to exploit the herein presented pressure responsiveness together with the already proved solvent capability of these nanostructured green fluids[19] (see Figure S9 in the SI for more details) in order to use them in chemical and process engineering. In fact, if we think to apply the revealed “water-rich” nanodomains embedded in a water-depleted matrix as nanocontainers, when in “ON” mode, then these nanocontainers can instantaneously be destroyed by pressure changes, resulting in a rapid and homogeneous mixing/demixing of the containers’ contents. From the technological point of view, by using these fluids we can imagine to create innovative pathways in process engineering and for the manufacturing of products with a high degree of control over material properties.

Methods

Microemulsion Preparation and Materials

Nanostructured systems have been prepared in a homemade, high-pressure visual variable volume cell (HPVVVC), which is described in the SI (Figure S1). Acetone (Uvasol) and acetonitrile (LiChrosolv) purchased from Merck Millipore (Germany) both with a purity of 99.9% as well as CO2 4.5 purchased from Linde (Germany) with a purity of 99.995% were used without further purification. Water was highly purified (conductivity <18 μS cm–1) in house by an ion-exchange water system (SG 2800 SK, SG Wasseraufbereitung and Regenerierstation, Germany). For hp-SANS measurements, acetone (Uvasol) has been replaced by its perdeuterated form (99.9 atom % D) purchased from Sigma-Aldrich and used without further purification.

Experimental Characterization

The miscibility gap of the ternary mixture was identified using the cloud-point method as described in detail in the SI. Data have been collected for water/organic solvent/CO2 mixtures at various compositions and temperatures. The Raman spectra of water, acetone, CO2, and their mixtures have been acquired using the setup described in the SI (Figure S3) and are reported in Figure S4. The water symmetric stretch vibration band has been isolated from the Raman spectrum of the mixture as described in SI. The Raman shift (wavenumber position) of the centroid of the isolated symmetric water stretch vibration signal was computed and considered as an indicator for the development of hydrogen bonds. Small Raman shift values of the centroid indicate that many hydrogen bonds are developed. Large Raman shift values of the centroid indicate that few hydrogen bonds are developed. High-pressure small-angle neutron scattering (hp-SANS) measurements were performed at the Institut Laue-Langevin (Grenoble, FR) (beamline D11). Data were collected using an incident neutron wavelength of 6 Å with 9% fwhm and working at two different sample–detector distances (1.75 and 10 m); this setup allows reaching a momentum transfer range between 0.007 and 3.5 Å–1. A high-pressure cell, previously described by Pütz et al., was utilized for the measurements.[34,35] Further technical details are provided in the SI.

Computer Simulations

The all atomic molecular dynamics (MD) simulations reported here were performed using the NAMD 2.9 software[36] with a thermostat at 308 K and a barostat at 10 MPa. In our simulations, water was described employing the TIP4P/2005 model,[37] which is known to correctly reproduce the phase diagram of water, both at low and high pressures.[38] This model is also the best available in reproducing the hydrogen-bonding features of liquid water at all pressures.[39] CO2 and acetone were described using the CGenFF version of the CHARMM force field.[40] We performed four simulations of a binary water/acetone mixture and also four simulations of the ternary water/acetone/CO2 mixture, all corresponding to different compositions but always at 308 K and 10 MPa. The first simulation was performed with a binary mixture with initially the same number of acetone and water molecules (xwater = xacetone). This simulation was the starting point for two different simulations series, one series corresponding to dilution with acetone (binary water/acetone mixtures) and the second series corresponding to dilution with CO2 (water/acetone/CO2 ternary mixtures). Note that the ternary water/acetone/CO2 mixture simulations correspond to different amounts of CO2 but always with xwater = xacetone. Further details of the simulation protocol are described in the SI.