Morphological Transition in Fatty Acid Self-Assemblies: A Process Driven by the Interplay between the Chain-Melting and Surface-Melting Process of the Hydrogen Bonds.

In surfactant systems, the major role of the nature of the counterion in the surfactant behavior is well-known. However, the effect of the molar ratio between the surfactant and its counterion is less explored in the literature. We investigated the effect of the molar ratio (R) between 12-hydroxystearic acid (12-HSA) and various alkanolamines as a function of the temperature in aqueous solution from the molecular scale to the mesoscale. By coupling microscopy techniques and small-angle neutron scattering, we showed that 12-HSA self-assembled into multilamellar tubes and transitioned into micelles at a precise temperature. This temperature transition depended on both the molar ratio and the alkyl chain length of the counterion and could be precisely tuned from 20 to 75 °C. This thermal behavior was investigated by differential scanning calorimetry and wide-angle X-ray scattering. We highlighted that the transition at the supramolecular scale between tubes and micelles came from two different mechanisms at the molecular scale as a function of the molar ratio. At low R, with an excess of counterion, the transition came from the chain-melting phenomenon. At high R, with an excess of 12-HSA, the transition came from both the chain-melting process and the surface-melting process of the hydrogen bonds. At the mesoscale, this transition of supramolecular assemblies from tubes to micelles delimited a regime of high bulk viscosity, with a regime of low viscosity.


■ INTRODUCTION
In surfactant science, the main parameter known to dictate surfactant properties is the nature of the surfactant itself: cationic, anionic, or nonionic. 1 A slight change in the molecular structure of the surfactant can affect the self-assembled structure in water and the interfacial activity, which can, in turn, tune the properties at the macroscopic scale, such as the rheological behavior or the ability to stabilize emulsions and foams. 2 However, not only the nature of the surfactant is important, since the presence of electrolytes and both the chemical structure and the charge of the counterion can also drastically change the properties of the solution. For a given ionic surfactant, a change of the counterion can modify the behavior of the surfactant: its solubility, its critical aggregation concentration (cac), and its foaming and emulsifying properties. 3−8 These changes come from the interactions between these two species, which depend on the chemical structure of the counterion. 9,10 Surfactant counterions may be either inorganic (sodium, potassium, etc.) or organic (tetramethylammonium, amino acids, etc.). In the case of fluorosurfactants, James and Eastoe have summarized in a review the effects of various counterions on the physical and chemical properties of such fluoro-based molecules. 11 For fatty acids, numerous examples are described in the literature showing how a change of counterion will have an effect on the Krafft temperature, the self-assembly, and the interfacial properties. 12−18 Therefore, by changing the counterion size and hydrophobicity, it is possible to easily tune the properties of a given surfactant.
Another way to modify the surfactant's properties is to tune the molar ratio between the surfactant and the counterion. It is an efficient way, but less explored in the literature. 14 Regarding fatty acids as anionic surfactants, the interest in modifying the molar ratio has increased in the last 5 years in order to improve their dispersion in aqueous solution. 14,19−21 For example, for the system based on myristic acid in the presence of choline hydroxide as counterion, it has been described that the myristic acid displays a broad polymorphism from faceted vesicles, multilamellar vesicles, and lamellar phases to spherical micelles as a function of the molar ratio. 22 The interfacial properties of the myristic acid are also directly linked to the quantity of counterions, leading to various foaming and emulsifying behaviors.
In past studies, we focused on the 12-hydroxystearic acid , which is an inexpensive molecular surfactant available in large quantities and at low cost derived by the hydrogenation of a sustainable materialricinoleic acid from castor plantsbecause it can self-assemble in unusual morphologies. 23−26 Indeed, the 12-HSA is known to selfassemble into multilamellar micron-size tubes in the presence of alkanolamine as counterion. 25 Upon heating, the tubes transform into spherical micelles due to a change in temperature, which in turn changes the packing parameter of the fatty acid assemblies. These tubes can be used for drug delivery, as illustrated in the literature. 27 More interest in this system comes from the outstanding stability over months of the foams produced from these 12-HSA tubes dispersions, which are even described as "ultrastable". 28 The phase transition of tubes into micelles upon heating leads to fast foam destabilization. By using the temperature response of this selfassembly, the production of responsive foams has been achieved. 28 Previously, we have demonstrated that the temperature transition between tubes and micelles is tuned by the alkyl chain length of the alkanolamines. 25 For example, at an equimolar 12-HSA/alkanolamine ratio, the transition is around 75°C for ethanolamine and 60°C for hexanolamine. In the case of ethanolamine, we have shown that the molar ratio modifies the transition from 75 to 43°C in an excess of ethanolamine in bulk. 14 Our aim was to understand the effect of the molar ratio between 12-HSA and various alkanolamines on the tube/ micelle transition from the molecular scale to the supramolecular scale in order to tune it for a wide range of temperatures. At the microscopic scale, we investigated these systems by coupling microscopy techniques, DSC, SANS, and WAXS. Combining these techniques allows for a comprehensive overview of the changes occurring at the supramolecular and molecular scale for 12-HSA. We showed how the molar ratio and the nature of the counterion are the key parameters to tune the transition between tubes and micelle, leading to drastic changes of the solution viscosity.

■ MATERIALS AND METHODS
Sample Preparation. 12-Hydroxystearic acid (12-hydroxyoctadecanoic acid, Sigma-Aldrich, 99% purity) was weighed in a sample tube into which Milli-Q water was added to obtain the desired concentration. Next, we mixed in the desired volume of a 1 M stock solution of the counterion to obtain the desired molar ratio (R), defined as R = n 12-HSA /(n 12-HSA + n counterion ), where n the molar concentration in mol L −1 . All the counterions used are listed in Table 1 and were purchased from Sigma-Aldrich with the highest purity available. The mixture was heated at 80°C for 15 min until all fatty acid solids were dispersed. The samples were then vigorously vortexed and cooled to room temperature. Prior to use, each sample was heated at 80°C for 15 min and cooled to room temperature.
Microscopy observations were carried out as a function of temperature (10−80°C within ±0.2°C ) at 20× magnification using an optical microscope in the phasecontrast mode (Nikon Eclipse E-400, Tokyo, Japan) equipped with a 3-CCD JVC camera, allowing digital images (768 × 512 pixels) to be collected. A drop of the lipid dispersion (about 20 μL) was deposited on the glass-slide surface (76 × 26 × 1.1 mm, RS France) and covered with a cover slide (22 × 22 mm, Menzel−Glaser). The glass slides were previously cleaned with ethanol.
Transmission Electronic Microscopy (TEM). A drop of each sample was placed on a carbon-coated TEM copper grid. They were negatively stained with uranyl acetate. Then, the grid was air-dried before observation. The samples were mounted in a Gatan 910 specimen holder that was inserted in the microscope using a CT-3500cryotransfer system. TEM images were then obtained by using a JEM 1230 cryo-microscope (JEOL) operated at 80 kV and equipped with a LaB6 filament.
Differential Scanning Calorimetry (DSC). The phase transition temperatures were measured on a microcalorimeter (Micro-DSC 7, Setaram). Two stainless steel cells were used, one containing ca. 0.75 g of sample and the other filled with the same amount of water used as reference. The heating and cooling ramps were from 10 to 85°C at a rate of 1°C/min. The data analysis was performed with Calisto processing software.
Small-Angle Neutron Scattering (SANS). Small-angle neutron scattering experiments were performed at Laboratoire León-Brillouin (Saclay, France) on the spectrometer PAXY. We used three configurations to get a Q-range lying between 0.005 and 0.4 Å −1 (respectively 4.5 Å at 1 m, 4.5 Å at 4.5 m, and 13.5 Å at 4.5 m). The neutron beam was collimated by appropriately chosen neutron guides and circular apertures, with a beam diameter at the sample position of 7.6 mm. The neutron wavelength was set to the desired value with a mechanical velocity selector (Δλ/λ ≈ 0.1). The samples, prepared with deuterated water, were held in flat quartz cells (Hellma) with a 2 mm optical path and temperature-controlled by a circulating fluid to within ±0.2°C. All samples were freshly made. The azimuthally averaged spectra were corrected for solvent, cell, and incoherent scattering, as well as for background noise, by using the PASINET software package provided at the beamline. 29 The fitting procedure with all the form and structure factors used is described in the Supporting Information (SI).
Wide-Angle X-ray Scattering (WAXS). WAXS spectra were recorded during 2 h on a Bruker D8 Discover diffractometer. Cu Ka 1 radiation (Cu Ka 1 = 1.5405 Å), produced in a sealed tube at 40 kV and 40 mA, was selected and parallelized using a Gobel mirror parallel optics system and collimated to produce a 0.5 cm beam diameter. Samples were prepared using the previous procedure with a 12-HSA concentration of 2% put in thin capillaries 1.5 mm in diameter, which were then flame-sealed immediately. The temperature was varied from 15 to 75°C and controlled by a HFS 91-CAP platine (Linkam).
Rheometry. An Anton Paar rheometer (MCR 301) was used to study the viscoelastic properties of the solutions of 12-HSA and alkanoamine. Experiments were performed with a cone−plate setup, well-suited for these solutions and for steady-shear measurements. The rheometer was equipped with a Peltier system so that we could accurately set the temperature from 10 to 80°C.
Tests were made with other geometrical setups to be sure of the reproducibility of the results and the independence on the setup used.

■ RESULTS
Phase Behavior in the 12-HSA/Counterion System at the Supramolecular Scale. In this study, five counterions have been used in a mixture with 12-HSA. The abbreviation of the counterions corresponds to their number of carbons. The mixture of 12-HSA with counterions are denoted in the rest of the paper with the abbreviation of the counterions, that is to say C2−C6 ( Table 1). The 12-HSA concentration was fixed at 10 g L −1 , and the molar ratio, as defined previously, was R = n 12-HSA / (n 12-HSA + n counterion ).
For R > 0.66, 12-HSA crystals were observed by eye, indicating that the amount of counterion was not enough to Langmuir Article disperse the 12-HSA. Therefore, we limit our study to the range between 0.18 and 0.66. By using phase-contrast microscopy, we determined the phase diagram as a function of both R and temperature for each counterion. Two different cases were observed by phase contrast microscopy as a function of the temperature: either tubes leading to a turbid solution or a limpid solution (Figure SI.1, SI). The temperature at which tubes transitioned into a limpid solution was denoted as the temperature transition (T t ) and is represented in Figure 1 and Table SI.1 (SI). Three regimes were observed as a function of R. For example, in the case of C5, at low molar ratio (0.18 ≤ R ≤ 0.33), T t was constant around 30°C ± 0.5. From R = 0.35 to 0.55, T t increased sharply to reach a value of 72 ± 1°C. For R > 0.55, T t did not evolve anymore and remained constant around 72 ± 1°C. For each counterion, the same behavior was observed, but T t was shifted to lower temperatures by increasing the number of carbons of the counterion. From C2 to C6, at low molar ratio, T t decreased by around 23°C. At high molar ratio (R > 0.55), T t varied by 10°C between C2 and C6. The intermediate regime of R, for which T t increased sharply, was shifted to higher R by increasing the number of carbons of the counterion. We suppose that this result could be explained by a difference in pK a between the counterions. 13 By modifying both R and the counterions, T t could be finely tuned from 20 to 78°C. In order to characterize the structure of tubes in solution on the supramolecular scale in various regions shown in Figure 1, TEM experiments were performed for each counterion below T t at R = 0.25 and 0.60. When we compared, for a given counterion, the tubes' aspect at the two molar ratios, we observed that the tubes were similar (Figures 2 and SI.2a−d). For example, for C5, the tubes' length was around 10 μm, and the tubes' diameter was around 0.6 μm, whatever the R value ( Figure 2). Below T t , tubes of a given counterion had similar characteristics on the supramolecular scale.
As described in the Introduction, the 12-HSA tubes are multilamellar. 24 In order to determine the effect of R on the multilamellar structure of these tubes, we performed SANS experiments at various R values for each counterion below T t . Figure 3a displays the scattering profiles for C5 at R = 0.33, 0.45, 0.50, and 0.55 at 20°C. The scattering profiles for C2, C3, C4, and C6 are shown in Figure SI.3 (SI).
In the low-Q region, we observed two to four intense sharp peaks. Their positions were exactly in a ratio of 1:2:3:4 (Q 0 , 2Q 0 , 3Q 0 , 4Q 0 ). The presence of a strong correlation peak followed by its harmonics indicated the presence of stacked bilayers. This result was similar to the ones previously obtained, confirming that tubes remained multilamellar whatever the value of R and the nature of the counterion when T < T t . 25 The interlayer spacing (d), corresponding to the repeat distance to one fatty acid bilayer and one water layer in the stack, was estimated from the first peak position (d = 2π/Q 0 ). The data are displayed as a function of both R and the nature of the counterion at 20°C (Figure 4a). For C2, the interlayer spacing increased from 314 Å for R = 0.20 to 349 Å for R = 0.55. For C6, the interlayer spacing increased from 203 Å for R = 0.45 to 255 Å for R = 0.55. For each counterion, the interlayer spacing increased upon increasing R. This result can be explained by the decrease of the counterion amount when increasing R. This decrease leads to a decrease of the screening effect of the counterion and an increase of interlayer spacing between the negatively charged bilayers. At a fixed R, we observed that the interlayer spacing increased by decreasing the carbon numbers of the counterion. For counterions with higher carbon numbers, it is possible that they are more condensed onto the surfactant layers, reducing the effective charge of the bilayers. Therefore, they could decrease the double-layer repulsion and lead to a closer approach between the bilayers.
Above T t , the scattering spectra were completely different [Figures 3b and SI.3 (SI)]. At large Q, the spectra were fitted with a form factor of a sphere. From the oscillation of the form factor, we determined the radius of the sphere. By fitting the data, we showed the presence of negatively charged spherical micelles, and the radius slightly varied around 22 ± 2 Å as a   (Table SI.2, SI). These results show that the absence of micron-sized objects as viewed by phase-contrast microscopy inside the limpid solution above T t comes from the presence of spherical micelles, as previously shown in the literature. 25 At low Q, the scattered intensity decreased due to the low isothermal compressibility of the system. Micelles were mainly composed of the ionized 12-HSA molecules and were negatively charged. Micelles repelled over large distances due to electrostatic repulsions, giving rise to a broad correlation peak, which corresponds in the direct space to the average distance between spherical micelles. For C5 at R = 0.45, the broad correlation peak was centered on Q = 0.055 Å, which corresponded to a distance between micelles of around 114 Å. For C5 at R = 0.50, it was located at Q = 0.052 Å. The distance between micelles was around 121 Å. The distance between micelles decreased by decreasing R. We observed in the same time a slight decrease in the micelles size by decreasing R, which means that the aggregation number also decreased (Table SI.2, SI). Whatever the R value, the number of 12-HSA molecules remained the same. Therefore, the number of micelles increased, leading to a decrease of the distance between them in solution. To quantify the evolution of the aggregation number and the micelle charge, the position of the counterion regarding the micelles and the 12-HSA needs to be known in order to perfectly fit the data. Other experiments such as SANS with contrast variation are needed. 30 In conclusion, these observations confirm the results previously obtained for various alkanolamines at an equimolar ratio and show that in all cases, whatever the counterion and the molar ratio, when T < T t , multilamellar tubes were present, and when T > T t , tubes transitioned into spherical micelles ( Figure SI.5, SI).
Thermal Behavior of the Bilayers at the Molecular Scale. The 12-HSA tubes are made of multilayers. In the system of 12-HSA tubes based on C2, it is known that the alkyl chains of the fatty acid embedded in the bilayers may be present either in the gel state (Lβ) or the fluid state (Lα), depending on the temperature. 26 In the literature, the transition between the two states is called the melting transition and leads to a change in the fluidity and thickness of the bilayers. 31 We studied the state of the bilayers for each system in order to characterize the melting transition as a function of both R and the nature of the counterion.
To evaluate the evolution of the bilayer thickness, we used the SANS spectra previously described at large Q [Figures 3a and SI.3 and SI.4 (SI)]. From the oscillation of the form factor, we determined the bilayer thickness. The evolution of the bilayer thickness at 20°C as a function of R and the counterion is represented in Figure 4b. For C2 at 20°C, from R = 0.2 to 0.33, the bilayer thickness was constant around 25 ± 1 Å. Then, for 0.33 < R < 0.40, the bilayer thickness sharply increased to reach 42 ±1 Å. For R > 0.40, the bilayer thickness remained constant, around 42 ±1 Å. At high R, the value of 42 Å suggested that 12-HSA molecules were embedded in a gel bilayer phase, since it corresponds to exactly twice the length of the 12-HSA chain length in its extended conformation (21 Å). For the other counterions (C3, C4, C5, and C6), from R = 0.33 to 0.55 at 20°C, the bilayer thickness remained constant, around 25 ±1 Å. For R > 0.55, the bilayer thickness began to increase sharply for C3, C4, and C5 and slowly for C6. By  increasing R for each counterion, the bilayer thickness evolved from 25 to 28 to 42 Å. When we compared the bilayer thickness at a fixed R = 0.55 at 20°C, we observed that the bilayer thickness increased from 28 Å for C6 to 42 Å for C2. The bilayer thickness evolved as a function of R and the counterion at a given temperature. When the bilayer thickness was around 25 Å, the thickness was lower than twice the 12-HSA in extended conformation. This suggested that either the bilayer was in the fluid phase (Lα) or in the gel phase (Lβ) with its alkyl chains interdigitated. 32 To discriminate between these two hypotheses and to understand the thermal behavior of the 12-HSA alkyl chains inside the bilayer, we coupled two techniques: wide-angle X-ray scattering (WAXS) and differential scanning calorimetry (DSC).
We first performed DSC measurements as a function of R for each counterion [Figures 5a and SI.6 (SI)]. As a function of R, one or two endothermic peaks were observed. The global shape of the curves and the position of the transition temperatures were dependent on R. For example, for C5, for R = 0.33, only one endothermic peak was observed and the maximum of the peak was around 27°C. This temperature corresponds exactly to the temperature of the transition between the tubes and the micelles (Figure 1). From R = 0.46 to 0.55, two endothermic peaks were observed (Figure 5a). The second peak of the temperature corresponds here to the temperature of the transition between the tubes and the micelles (Figure 1). The same trend was observed for all the counterions (Figure SI.6, SI). For each counterion, we determined the peak with the highest temperature on the enthalpogram, and we plotted the change of the maximum temperature of this peak as a function of R (Figure 5b). By comparing Figures 1 and 5b, we observed exactly the same behavior. We can conclude that the peak at the highest temperature in DSC endotherms corresponds to the tubes-to-micelles transition. For all counterions, the same behavior with three regimes was observed (Figure 5b). At low R, from R = 0.20 to 0.40, for C3, C4, C5, and C6, only one endothermic peak was observed and the temperature of the peak remained almost constant. The temperature of the peak was 32, 27, 24.5, and 19.5°C, for C3, C4, C5, and C6, respectively. For C2, again one endothermic peak was observed with a maximum temperature around 37°C, but from R = 0.20 to 0.33. At low R, only one endothermic peak was observed with a constant temperature for all the counterions. We observed that the temperature of the peak increased by decreasing the number of carbons of the counterion (Figure  5b). At intermediate R, constituting the second regime, for all counterions, two endothermic peaks were present and the temperature of the highest peak increased sharply by increasing R (Figure 5a). For example, for C5, with 0.40 < R < 0.55, two endothermic peaks were noted and the temperature of the highest peak increased from 26 to 67°C (Figure 5a). At high R (R > 0.55), two endothermic peaks were present and the temperature of the peak was almost constant (Figure 5b). For example, the temperature was around 78 and 68°C for C2 and C5, respectively. We noticed that the temperature increased upon decreasing the carbon numbers of the counterion, as previously observed by microscopy.
To understand the nature of these transitions, we performed WAXS experiments. For C5 at R = 0.33, two diffraction peaks were observed at 1.50 and 1.58 Å −1 from 10 to 20°C ( Figure  6a). Then, the two peaks completely disappeared above 25°C. However, at R = 0.55, three sharp peaks were observed at 1.40, 1.58, and 1.62 Å −1 from 10 to 20°C with the same intensity ( Figure 6b). However, from 22 to 25°C the peak intensity decreased progressively to disappear completely at 30°C. The presence of these peaks shows that the 12-HSA alkyl chains are in the Lβ state. When the peaks disappear, the bilayer is in the Lα state. The presence of two peaks at low R and three peaks at high R showed that the molecular organization inside the bilayers was not the same. By comparing the DSC and SANS results, we can conclude that below the first peak in the DSC curve for C5 at R = 0.33, the bilayer thickness of 27 ±1 Å at 20°C corresponded to interdigitated alkyl chain chains in the gel state. We can suppose that for all counterions below the first peak in the DSC curve the alkyl chains are in the gel state. When the carbon number of the counterion is high enough, it may enter inside the bilayer, leading to interdigitated bilayers. Above the first peak in the DSC curve, the alkyl chains were in the fluid phase (Lα).
Effect of R and of the Counterion at the Mesoscale: Bulk Viscosity. In a previous study, we showed that the transition between tubes and micelles leads to drastic changes of viscosity for C2 at an equimolar ratio R = 0.50. 33 Here, we determined the evolution of the dynamic viscosity for each counterion at various R as a function of the temperature. In Figure 7a, the results obtained for R = 0.50 are shown. The shear rate was fixed at 0.5 s −1 . All the curves present a drastic fall, where the viscosity decreased by more than 3 orders of magnitude within a few degrees around a critical temperature, which actually turns out to be close to the T t discussed previously for molecular and supramolecular scales. Below T t , the solutions were always highly viscous for each counterion (viscosity >1000 mPa s) and with a strong shear-thinning behavior. At T ≈ T t , the viscosity sharply decreased to reach low viscosity values around 1 mPa s, close to the value of pure water, and became Newtonian. In agreement with the previous observations, the temperature at which the viscosity decreased was shifted to lower values by increasing the number of carbons of the counterion. We also determined the evolution of the temperature at which the viscosity changed as a function of R. The results for C2 are shown (Figure 7b). We observed that this temperature increased by increasing R and was again close to T t , but it was less easy to distinguish the three transition regimes. Therefore, there is a direct correlation between the presence of tubes and the high viscosity of the solution, and the abrupt fall of viscosity is a direct consequence of the tube−micelle transition. These results are complementary to our previous work based only on C2 at an equimolar ratio and make clear that the same qualitative features are recovered for the five counterions tested here and for all R.
Discussion: Link between the Transition on the Supramolecular Scale and the Melting Phenomena on the Molecular Scale. We observed that the transition between tubes and micelles was tuned by R. For all counterions, we identified three different regimes according to R: low R, intermediate R, and high R.
At low R, T t was constant and only one endothermic peak was observed by DSC. In this regime, there was an excess of counterion in solution. 12-HSA and alkanolamines are a weak base and acid, respectively. There is a coexistence of fatty acids under their carboxylic (COOH) and carboxylate (COO − ) forms. The ratio between the two forms depends on the pH of the solution, which is governed by the molar ratio. At low R, the pH of the solution for all counterions was relatively high around 10.75 ± 0.25 at 20°C and almost constant. We suppose that when R is low, the amount of the carboxylate form is markedly higher than that of the carboxylic form. In these conditions, the headgroup area is large. There are few hydrogen bonds and the interactions between ionized headgroups are repulsive. Below T t , the aliphatic chains were in a rigid and ordered gel state, leading to the formation of multilamellar tubes. The interdigitated bilayers observed for some counterions can be attributed to the electrostatic repulsion between the negatively charged fatty acid molecules and/or the insertion of the counterion inside the bilayer. Upon heating at T t , the tubes transformed into spherical micelles due to the alkyl-chainmelting process. The aliphatic chains change from the gel state to a fluid state, in which bilayers become more flexible. The molecules become disordered and the repulsions between the ionized headgroups lead to a decrease of the packing parameter, and micelles, which are flexible and self-assemble with higher curvature, are formed. In this low R regime, T t is mainly dependent on the hydrophobic interaction. The amount of COO − is high and almost constant. The interactions between headgroups then do not vary. Therefore, the chain-melting process happens at the same temperature for a given counterion. At low R, T t has a much lower value than at high

Langmuir
Article R, because it depends only on the chain melting, which happens at low temperatures. Moreover, the chain-melting temperature depends on the alkyl chain length of the counterion: the higher the alkyl chain length, the lower the chain-melting temperature.
At intermediate R, T t increased sharply according to R. Two endothermic peaks were observed by DSC. The pH of the solution for all counterions decreased sharply from 10.75 ± 0.25 down to 9.75 ± 0.25 at 20°C. In this intermediate regime, by increasing R, the excess of counterion decreases progressively to reach an excess of 12-HSA in solution. The quantity of fatty acid molecules neutralized by the counterion decreases and the quantity of protonated 12-HSA molecules increases. There is a coexistence of the carboxylic and carboxylate forms in solution. Hydrogen bonds can be obtained between the two types of forms at the headgroup level. We identified that the first endothermic peak in the DSC curve corresponds to the chain-melting process and it is almost constant, such as in the low R regime. Above the chain-melting process, multilamellar tubes are still observed, even if the bilayers are in the fluid disordered state due to the hydrogen bonds, which help to stabilize and maintain the bilayer structure. 32,34 However, when the hydrogen bonds disappear, the tubes-to-micelles transition is observed. This phenomenon is called surface melting. The second peak in the DSC curve corresponds to the disappearance of the hydrogen bonds at the headgroup level. 32,34 The headgroup area becomes larger, leading to a lower packing parameter and the formation of micelles. For this regime, T t is linked to the surface melting and not to the chain-melting process. T t is dependent on the headgroups' interactions, and as a result, T t increases with the number of hydrogen bonds between headgroups. The increase of R leads to an increase of the carboxylic forms, giving more hydrogen bonds to stabilize the bilayers, which can resist the surface melting at higher temperatures. Thus, by increasing R, the sharp increase observed for T t is correlated to the increase of the surface-melting temperature.
At high R, T t was almost constant but markedly higher than for the low R regime. The pH of the solution for all counterions was almost constant, around 9.75 ± 0.25 at 20°C. It was lower than for the low R regime. In such a high R regime, there is an excess of 12-HSA in solution. The quantity of protonated 12-HSA molecules is high and almost constant up to a threshold R for which the amount of counterion is too low to disperse the 12-HSA, and crystals are observed (for R > 0.66). At high R, hydrogen bonds are still present at the headgroup level, but the number remains almost constant. Two endothermic peaks were observed by DSC. In the same way as for the intermediate regime, the alkyl-chain-melting and the surface-melting processes occur. Since the number of hydrogen bonds remains constant, the surface-melting process occurs at a constant temperature and T t is constant. We observed that the surfacemelting temperature depends on the alkyl chain length of the counterion: the higher the alkyl chain length, the lower the surface-melting temperature. We can suppose that it is linked to the pK a of the counterion and the pH of the solution modifying the ionization state of the 12-HSA, which is slightly higher when the alkyl chain length of the counterion increases.
To summarize, thanks to complementary studies on the molecular and supramolecular scale, we have illustrated many features of this tube−micelle transition and identified its origins. The temperature T t at which tubes transit toward micelles can be easily tuned by R, which governs both the surface-melting and the alkyl-chain-melting transitions. For all the alkanolamines tested here, two extreme limits are found for low and high R, where T t is constant and respectively corresponding to an excess of counterion or an excess of 12-HSA. In between these two limits, a intermediary regime is found where T t varies linearly with R. Similar observations have been described in the literature for salt-free catanionics systems based on fatty acid and cationic surfactant in its hydroxide form. 31,35 The transition between tubes and micelles was reversible due to the reversibility of the surface and alkyl-chainmelting phenomena.

■ CONCLUSION
The effect of the molar ratio between 12-HSA and various alkanolamines as counterion has been studied in bulk. The 12-HSA self-assembled into multilamellar tubes that transitioned into micelles at a precise temperature, which depended on both the molar ratio and the alkyl chain length of the counterion. This transition delimited a regime of high bulk viscosity, with a regime of low viscosity. This temperature transition could be precisely tuned from 20 to 75°C. We demonstrated that this transition between tubes to micelles at the supramolecular scale had two origins at the molecular scale as a function of the molar ratio. At low R, with an excess of counterion, the transition came from the chain-melting phenomenon. At high R, with an excess of 12-HSA, two phenomena at play were observed leading to the transition between tubes into micelles. The first one was the chain-melting process and the second one was the disappearance of the hydrogen bonds. This study confirms that the molar ratio is a crucial parameter to take into account in surfactant systems and the approach used in this study could be extended to other surfactant systems. Moreover, from our results, we can see that the effects of the counterion and molar ratio are not only important to the transition between tubes and micelles, but they seem to have a role in the tubes' structure and formation mechanisms. Further studies are needed to determine precisely the position of the counterion as a function of its alkyl chain length and the molar ratio regarding the fatty acid bilayers and the micelles. These results could be compared with the different theories about selfassembled molecules and formation of lipid tubes in order to determine precisely how these tubes are formed. 36−41 Our system based on 12-HSA as surfactant in combination with alkanolamine is known to exhibit excellent foamability and stability due to the presence of 12-HSA tubes in the foam liquid channels and their adsorption at the air/water interface. The transformation of the tubes into micelles upon heating above the transition temperature led to the complete destruction of the foam, leading to responsive foams controllable by the temperature as stimulus. Moreover, bulk and interfacial rheological properties are tuned by the 12-HSA changes of the supramolecular assembly. The temperature at which all of these properties are modified is directly linked to the temperature transition between tubes and micelles. From applied perspectives, the modification of both the counterion and the molar ratio could be a simple but effective way to choose the temperature transition at any given temperatures between 20 and 75°C. Thus, the 12-HSA properties at the macroscopic scale, such as foaming, bulk and interfacial rheological properties, and drug delivery properties, could be manipulated precisely in a wide range of temperatures.