The bicontinuous microemulsions investigated in this work were prepared according to the composition given by the quaternary phase diagram shown in Supplementary Figure S1 in the supporting material. At a constant cyclohexane to water ratio of α = 0.5 the single phase domain is depicted as a function of the C_8/10G_1.3 (Glucopon 220) weight fraction γ and the total pentanol weight fraction δ. A more detailled discussion of the phase behavior can be found in the literature Wellert et al. (2011). At a surfactant concentration of γ = 0.124 the bicontinuous region appears. With increasing surfactant content above γ = 0.25 the one phase region passes into the lamellar phase. For the experiments a bicontinuous sample at γ = 0.21, safely far from the phase boundaries, was chosen.

The silicon blocks were cleaned with RCA solution (5:1:1 of water, NH₄OH at 29%, H₂O₂ at 30%) at 72 ^◦C for 10 min, then rinsed several times with water and dried in a nitrogen stream.

For functionalization the silicon blocks or wafers were transferred from water to chloroform by subsequent ultrasonication in mixtures of CHCl₃/CH₃OH = 1:1, 3:1 and pure CHCl₃ for 10 min. Afterwards, the silicon substrates were put in an upright position in a desiccator with a residuum of approximately 6 μL) of HMDS in a vacuum desiccator at ambient temperature for 24 h. Then the substrates were transferred from chloroform to methanol by subsequent ultrasonication for 5 min in solvent mixtures of CHCl₃: CH₃OH from 1:0 → 0:1. The substrates were rinsed several times with water and dried in a nitrogen stream.

RCA 1 cleaned silicon blocks (static water contact angle ≤10 ^◦), were used as hydrophilic surface, whereas HMDS-modified silicon blocks (static water contact angle (80 ± 10) ^◦) were used as hydrophobic surface. The RCA cleaning protocol was applied just before sealing the substrate inside the solid/liquid cell dedicated to NR and GINSES measurements. In the remainder of this paper, hydrophilic and hydrophobic surfaces are termed h-Si respectively hp-Si.

The surface tension of the pure MilliQ-type water, cyclohexane, aqueous sugar surfactant solution (c (C_8/10G_1.3) in H₂O of 20 mmol/ml) and the microemulsion were determined with a force tensiometer DCAT 11 (Dataphysics, Germany) applying the du Noüy ring method. Prior to each measurement, the platinum ring was flamed to remove any organic residuals. For each sample, 10 measurements were carried out and averaged at 25^◦C.

Contact angles were measured with a drop contour analysis of sessile drops on the planar hydrophilic (h-Si) and hydrophobic (hp-Si) substrates using the OCA20 contact angle goniometer (Dataphysics, Germany). Prior to the measurements, the sample surfaces and the sessile droplets were placed in a quartz sample cell in a saturated vapor phase using the investigated liquids. Measurements of the surface energy components of h-Si and hp-Si substrates are shown in Supplementary Figure S2 and described in the supporting information.

Neutron reflectivity was measured at the neutron reflectometer V6 located at the NL4 neutron guide of the medium flux research reactor at the Helmholtz Zentrum Berlin (Germany). The neutron wavelength of 4.66 Å was selected by a graphite monochromator and filtered by a liquid nitrogen cooled Be filter. The beam profile was 1 × 40 mm which sets the instrument resolution to ΔQ = 2 × 10^–3 Å⁻¹. The sample cell consisted of a Teflon trough closed with the single crystal silicon block serving as the solid interface in a thermostated aluminium housing. This cell was placed horizontally in the collimated neutron beam. Both front ends of the cell were covered with cadmium shielding to prevent bulk phase scattering.

Reflectivity curves were measured in θ/2θ-geometry and recorded using a position sensitive detector. The scattering length densitites (SLDs) of the microemulsion components are 6.38 × 10^–6 Å⁻², 6.72 × 10^–6 Å⁻², 0.34 × 10^–6 Å⁻², and −0.32 × 10^–6 Å⁻² for D₂O, deuterated cyclohexane (C₆D₁₂), C_8/10G_1.3 and pentanol (C₅H₁₁OH). Additionally, water with a SLD of −0.53 × 10⁻⁶Å⁻² was used in a D₂O/H₂O mixture to adjust the bulk phase SLD. To obtain reliable fit results, the critical edge of total reflection of a reflectivity curve has to be within the probed Q _z range. To achieve this, the contrast between oil phase and water phase was accordingly tuned, while still maintaining a SLD step towards higher SLD from Si (2.1 × 10^–6 Å⁻²) to the average SLD of the microemulsion. Raw data were normalized and footprint corrected with the software provided by the HZB.

The measured reflectivity curves were analyzed with the fitting routine Parratt 32. This approach determines the neutron reflectivity from planar surfaces using a calculation based on Parratt’s recursion scheme for stratified media. An oscillating SLD profile of the form ρ z = A e − z / ξ z ⁡ cos 2 π z D z + ϕ (1)

was used in this approach. The SLD model assumes alternating layers of oil and water with a domain size D _z perpendicular to the confining substrate. The ordered structure vanishes according to a correlation length ξ _z . Based on Ginzburg–Landau theory this order parameter profile was obtained by including additional surface fields in the free energy functional Lee et al. (1995); Gompper and Schick (1990); Chernov and Mikheev (1988). Similar theoretical concepts are used for the description of planar amphiphilic membranes adhered to a planar solid substrate Charitat et al. (2008).

All NSE measurements were carried out on the J-NSE instrument at the FRM II neutron source in Garching, Germany Holderer et al. (2008). A neutron wavelength of 8 Å was used. All samples were measured at 25^◦ C.

The preferred contrast for NSE is the “film contrast”, where only the interface layer contains hydrogenated materials and oil and water phase are deuterated with as little hydrogen as possible. Interface fluctuations can be best observed in this contrast. The measurements on the bulk sample were done in transmission geometry using standard quartz sample cells. The neutron path length in the Hellma quartz cells was 2 mm. The bulk sample was measured at Q = 0.08 Å⁻¹ for reference.

Neutron spin echo spectroscopy under grazing incidence requires a well defined angle of incidence α _i to provide a defined depth of neutron penetration Λ into the sample Frielinghaus et al. (2012); Holderer et al. (2014). Therefore the entrance aperture is reduced to 2 mm. Figure 1 illustrates the GINSES scattering geometry. The neutron beam enters the sample from the silicon block and is reflected at the interface. The critical angle of total reflection α c = λ Δ ρ / π depends on the chosen neutron wavelength λ and the contrast Δρ between the silicon block and the average SLD of the microemulsion. At an angle of incidence α _i ≤ α _c an evanescent neutron wave of intensity I _ev (z) exponentially decays into the sample over a characteristic depth Λ with Λ = 1 4 π Δ ρ 1 − α i 2 / α c 2 . (2)

Plot b) in Figure 1 shows the spin coherent and incoherent scattered intensities at Q = 0.08 Å⁻¹ as a function of α _i for both surfaces. In the spin-up configuration (all polarization flippers off) no spin manipulation occurs and mainly the coherent contribution is measured. Contrary, the spin-down configuration (180^◦-polarization flip at the sample position) is dominated by incoherent scattering. At increasing angle of incidence α _i the scattered intensities slightly increase until α _c is reached. From the SLDs of deuterated cyclohexane and D₂O, constituting the bulk phases of the bicontinuous microemulsion, a value of α _c ≈ 0.15^◦ was estimated. However, already below α _c the coherent scattered intensities from the h-Si and hp-Si interfaces increase steeply. An angle of incidence α _i = 0.11^◦ close to the onset of this increase was chosen in the grazing incidence measurements. For this combination of α _c and chosen α _i , a neutron penetration Λ ≈ 70 nm was estimated using Eq. 2. This corresponds to a penetration into the sample over a distance of 4–5 domains. The scattered intensity was measured at Q = 0.08 Å⁻¹.

To study the influence of the chemical composition of a confining substrate on the near surface dynamics of bicontinuous microemulsions, freshly cleaned silicon surfaces (h-Si) and HMDS-modified silicon surfaces (hp-Si) were used as hydrophilic and hydrophobic surfaces, respectively. For h-Si the measured surface energy was γ ^SE=(69 ± 3) mN/m with γ p SE =(48 ± 2) mN/m and γ d SE =(20 ± 2) mN/m being the polar and unpolar components of the surface energy. In the case of hp-Si, the surface energy was γ ^SE=(30 ± 1) mN/m with γ p SE =(8 ± 1) mN/m and γ d SE =(22 ± 1) mN/m. Details of this measurements can be found in the supporting information in Supplementary Figure S3. The vapor deposition chemically changes the hydroxylated silicon surface to a trimethylsilyl terminated surface with ((CH₃)₃Si-O) groups at the outermost layer. This is reflected in the drastic change of the wetting behavior of water. Since water spreads over the freshly cleaned silicon surface without a detectable contact angle, at the trimethylsilyl terminated silicon surface values between 70^◦ an 90^◦ were observed, depending on the coverage of the surface with trimethylsilyl groups. In case of the bicontinous microemulsion, wetting with a very low contact angle below the detection limit of the optical contact angle goniometer was observed. This is illustrated by the images in Figure 2. Additionally, an aqueous solution of Glucopon 220 with a concentration of approximately 20 mmol/L was studied. On hp-Si a contact angle of (26 ± 2)^◦ was detected while h-Si was fully wetted by the surfactant solution. The results for the static contact angles were compiled in Table 1. The static contact angle data are shown in Supplementary Figure S2. In addition to the static contact angles, also the measured surface tensions of pure water, the surfactant solution, cyclohexane and the bicontinuous microemulsion are shown in Table 1. These values suggests that due to its low surface tension, a significant amount of cyclohexane is present at the air-liquid interface of the microemulsion which is the reason for the same wetting behavior of the microemulsion on both surfaces, hp-Si and h-Si.

The bicontinuous microemulsions show the same wetting behavior at the hp-Si and h-Si surfaces, as discussed in Section 4.1. This raises the question of whether there are differences in the microemulsion structure in the immediate proximity of the solid-liquid interfaces. Figure 3 shows the reflectivity curves of the investigated bicontinuous microemulsion in contact with the hydrophilic and hydrophobic surfaces. The critical edge of total reflection was shifted into the Q _z range accessible for the instrument mixing H₂O/D₂O to tune the SLD of the aqueous phase. The reflectivity curves show a broad but distinct Bragg peak which is slightly more pronounced and shifted to lower Q _z in case of hp-Si. The Bragg peak originates from the structuring of the microemulsion at the solid/liquid interface. The reflectivity data can be described by using a oscillating variation of the SLD related to alternating oil and water rich regions of the bicontinuous structure given by Eq. 1.

The inset in Figure 3 shows the SLD profiles for the bicontinuous microemulsion on both, h-Si and hp-Si substrates. Starting from negative z, the profile corresponds to the silicon substrate, followed by the silicon oxide (thickness of 15 Å) and, in the case of the hp-Si, a negative SLD value for the trimethylsilyl surface layer. A similar oscillation behavior was observable for both measurements. The SLD oscillations decay over two oscillations to the bulk value at z → ∞.

From the interpretation of these results, one can say, that the bicontinous bulk structure remains unchanged up to the immediate vicinity of the h-Si and hp-Si substrate. For the layer in direct contact with the substrate, a very similar SLD profile was obtained, which suggests that the structure of the microemulsion in the vicinity of either a hydrophilic or a hydrophobic surface is governed by the same ordering of the bulk phases. This corresponds well to the results of the contact angle measurements.

However, since these measurements determine the reflectivity without additional information of the phase of the complex reflection coefficient, in the kinematical approxiamation the near surface structure could only be determined to a remaining ambiguity. Hence, no reliable conclusion can be drawn about the existence and composition of an adsorbed surfactant layer at the surface. The “phase problem” in neutron reflectometry, i.e., the loss of phase information due to the measurement of reflected intensities, could be addressed in principal by measuring the same sample in different contrasts by isotope substitution or by introducing a known reference layer at the interface and varying the interfacial contrast by polarization analysis.

However, the use of the oscillating SLD profile given by Eq. 1 enables the estimation of the domain size D _z and the correlation length ξ _z perpendicular to the confining substrate. These results and the corresponding values of the bulk microemulsion, d _TS and ξ _TS , measured with SANS are compiled in Table 2. The sizes of the oil and water domains from the measurements at the h-Si and the hp-Si surfaces are similar. Compared to the bulk phase the correlation lengths are larger in case of the microemulsions in contact with the h-Si and the hp-Si surfaces. This difference suggests, that a small effect of the planar interface on the inner structure of the microemulsion exists. The SLD profiles of both measurements have a very similar course. This suggests a similar composition of the microemulsions in the near surface region of the two solid liquid interfaces.

The differences between the results from the bulk measurements and those at the interface must be interpreted with caution. Due to the pronounced maximum in the scattering curve in bulk, the peak position can be found more accurately in the fit than in the analysis of the reflectivity data where the maximum is strongly broadened. It can be deduced from the results that the domain sizes are modified, if at all, only to a small extent by the presence of the interfaces.

In a SANS measurement, the renormalized and the bare bending elasticity constants κ _SANS and κ _bare of the surfactant membrane in the microemulsion Gompper et al. (2001); Gompper and Kroll (1998), κ SANS k B T = 10 3 π 64 ξ T S d T S (3) and κ bare k B T = κ SANS k B T + 3 4 π ln d T S 2 l c (4)

can be calculated from the structural parameters (domain size d _TS and correlation length ξ _TS ) obtained from the Teubner–Strey analysis of the SANS data. l _c represents the thickness of the surfactant membrane. The same approach was applied to estimate the values of the renormalized and bare bending elasticity constants from the neutron reflectometry data. From the domain size D _z and correlation length ξ _z from reflectometry, the bendig rigidity has been calculated with Eq. 3. In particular the comparison between the hydrophilic and hydrophobic interface is possible with this approach, and it also gives a strong hint that the membrane elasticity is not significantly altered by the interface layer, as mentioned above.

The renormalized bending modulus κ _SANS and the bare bending modulus κ _bare characterize the elastic properties of the amphiphilic film at different length scales. On the length scale of the oil and water domains, κ _SANS is the effective bending modulus of the membrane. The bending motion of the membrane is superimposed with thermal fluctuations at smaller scales. Strong thermal fluctuations result in a larger deformation of the membrane Gompper et al. (2001); Pieruschka et al. (1995). The effective thickness of the surfactant membrane between the oil and water domains l _c determines the bare bending elasticity κ _bare Gompper and Kroll (1998); Morse (1994). The values of κ _SANS are rather small, between 0.54 k _B T and 0.42 k _B T and support the conclusion of a highly flexible and very soft interface. These values are smaller compared to the results from previous experiments with bicontinuous microemulsions from the ternary phase system decane/H₂O/C₁₀E₄ at hydrophilic surfaces. For these microemulsions, the formation of a lamellar ordering in the vicinity of the surface was found. The spatial expansion of this ordering away from the substrate extended up to a few domain sizes.

The membrane dynamics as measured with NSE spectroscopy can be described by calculating the intermediate scattering function S (Q,τ _NSE ) of fluctuating membrane patches Zilman and Granek (1996); Mihailescu et al. (2001). Before applying often used approximations for large bending rigidities, the intermediate scattering function is: S Q ⃗ , τ NSE ∝ ⟨ ∫ d 2 r ∫ d 2 r ′ ⁡ exp i Q ⃗ x y r ⃗ − r ⃗ ′ × exp − k B T 4 π 2 κ bare Q z 2 ∫ k min k max d k k 4 1 − e i k ⃗ r ⃗ − r ⃗ ′ − ω k τ α ⟩ (5) This can be approximated by a stretched exponential function which shows typically a rate Γ ∝ Q ³ and a stretching exponent β and yields by this information important hints if membrane dynamics is observed and what the direct effects of the interfacial confinement on the membrane dynamics are.

In Figure 4 three plots of normalized intermediate scattering functions (ISF) are shown. The ISF at the top was measured from the bulk sample in transmission geometry measured at a momentum transfer Q = 0.08 Å⁻¹. Below this, two ISF’s were measured under grazing incidence at Q _GINSES ≈ Q _z = 0.08 Å⁻¹. The data from the bicontinuous microemulsion in contact with the hydrophilic (h-Si) sample are given in the middle graph. Plotted in the bottom graph is the ISF of the microemulsion adhered to the trimethylsilyl terminated surface (hp-Si). All three curves show a continuous, smooth decay within the Fourier time range τ _NSE .

As a first step in the analysis, a stretched exponential with a stretching exponent β = 2/3, S Q , τ NSE / S Q , 0 = A e x p − Γ τ NSE β . (6)

was fitted to the data. The decay rate Γ measures the relaxation of thermally excited fluctuations of the surfactant film. The resulting relaxation rates Γ are given in the plots. A comparison shows, that the relaxations close to the surface are slightly faster than in bulk but are the same for both types of surface within the precision of the fit.

A detailed analysis was achieved by applying the full integral evaluation, required for small bending rigidities of the order of 1 k _B T, usually observed for bicontinuous microemulsions. In this model, the bending rigidity κ is the only fitting parameter.

Bulk microemulsions have a dispersion relation ω ( k ) = κ bare k 3 4 η with the undulation wave number k and the solvent viscosity η. For near surface dynamics, long wavelength undulations show a different dispersion relation ω(k) ∝ k ² due to interactions with the rigid wall Frielinghaus et al. (2012) Seifert (1994); Kraus and Seifert (1994). The interaction potential between the microemulsion and the confining wall introduces another length scale, which is related to l ̄ , the distance between the first membrane and the wall. GINSES experiments allow to fit this length if using the full integral version of S (Q, τ _NSE ).

This has been measured with GINSES for bicontinuous microemulsions Frielinghaus et al. (2012); Lipfert et al. (2014) and requires a fit with the full Zilman-Granek theory involving the numerical integration of all undulation waves of the membrane patches.

First, the bulk microemulsion has been fitted to determine the required parameters of the fitting function, especially the integration limit for long wavelength undulations, which is four times the correlation length determined from SANS. The bending rigidity has been fixed to κ _bare = 0.82 k _B T to reduce the number of fitting parameters. This corresponds to the renormalized value of κ SANS = κ bare − α 4 π ln ( d T S / 2 l c ) with the molecular length l _c of about 12 Å and the distance d _TS between domains in the microemulsion and α ≃ 3. Additional corrections would be necessary from the influence of the saddle splay modulus κ ̄ which can be neglected for the purpose of fixing the fit parameters for the interface layer Holderer et al. (2013).

In the next step, the fit of the microemulsion dynamics at both types of interfaces, the parameters were fixed by the bulk microemulsion except the characteristic distance l ̄ between first layer and wall. The integration limit for the long wavelength undulations determined in the bulk by ξ is set to infinity. In this case, l ̄ , measuring the wall-membrane interaction remains as the only fitting parameter to describe the difference between bulk and interface dynamics. Details can be found in literature Frielinghaus et al. (2012). The fit yields l ̄ = (24 ± 4) Å and (33 ± 4) Å for the h-Si and hp-Si interface, respectively. Within the precision of the experiment, both values are the same. The fits to the data are shown in Supplementary Figure S4 in the supporting information. Additionally, the fit parameters are summarized in Supplementary Figure S5.

These results agree with the findings from the neutron reflectometry measurements and suggest minor differences in the coupling of the investigated bicontinuous microemulsion to both surfaces. Probably, the microemulsion bulk structure adheres to both solid interfaces without a distinct extended lamellar transition region. In a previous experiment with ternary bicontinuous microemulsions in the decane/H₂O/C₁₀E₄ phase system in contact with a hydrophilic silicon substrate (h-Si) faster relaxation was found in the vicinity of the interface. On the other hand, for the contact between clay particle surfaces and bicontinuous microemulsion of the same phase system no difference between bulk and near surface relaxations in the microemulsions were found, which also was attributed to the absence of a lamellar transition region Frielinghaus et al. (2012).

The results of these GINSES experiments indicate a significant influence of the interaction between surfactant molecules and confining surface on the formation of an adjacent lamellar structure. Comparative studies on the adsorption of nonionic ethoxylated surfactants and of sugar surfactants on a variety of substrates revealed significant differences. The adsorption of a surfacatant is influenced by the chemical composition of the adsorbent. Moreover, although in both cases hydrogen bonding mediates the adhesion of the surfactant, at silicate surfaces ethoxylated surfactants were found to adsorb stronger than sugar surfactants Zhang et al. (2006); Zhang and Somasundaran (2006) which might support the formation of lamellar ordering in a C _i E _j based bicontinuous microemulsion.