ARTICLE pubs.acs.org/Langmuir
Adsorption of Aerosol-OT to Sapphire: Lamellar Structures Studied with Neutrons Maja S. Hellsing* and Adrian R. Rennie Materials Physics, Department of Physics and Astronomy, Ångstr€om Laboratory, Uppsala University, Box 516, 751 20, Uppsala, Sweden
Arwel V. Hughes ISIS Facility, Rutherford Appleton Laboratory, Didcot OX11 0QX, United Kingdom
bS Supporting Information ABSTRACT: The adsorption of sodium bis 2-ethylhexyl sulfosuccinate, NaAOT, to a sapphire surface from aqueous solution has been studied by neutron reflection at concentrations above the critical micelle concentration (cmc). Complementary measurements of the bulk structure were made with small-angle neutron scattering and grazing incidence small-angle neutron scattering. At a concentration of about 1% wt (10 cmc), lamellar phase NaAOT was observed both at the surface and in the bulk. The structure seen at the interface for a solution of 2% wt NaAOT is a 35 ( 2 Å thick bilayer adsorbed to the sapphire surface at maximum packing density, followed by an aligned stack of fluctuating bilayers of thickness 51 ( 2 Å and with an area per molecule of 40 ( 2 Å2. Each bilayer is separated by a water: at 25 °C, this layer is 148 ( 2 Å. A simple model for the reflectivity from fluctuating layers is presented, and for 2.0% wt NaAOT the fluctuations were found to have an amplitude of 25 ( 5 Å. The temperature sensitivity of the structure at the surface was investigated in the range 1530 °C. The effect of temperature was pronounced, with the solvent layer becoming thinner and the volume occupied by the NaAOT molecules in a bilayer increasing with temperature. The amplitude of the fluctuations, however, is approximately temperature independent in this range. The adsorption of NaAOT at the sapphire surface resembles that previously found at hydrophilic and hydrophobic silica surfaces. The coexisting bulk lamellar phase has a spacing of layers similar to that observed at the surface. These observations are an indication that the major driving force for adsorption is self-assembly, independent of the chemical nature of the interface.
’ INTRODUCTION The use of surfactants for industrial applications, in detergents, and as personal care products is often at concentrations significantly above the critical micelle concentration (cmc). Developing a deeper understanding of the behavior of surfactants at high concentrations such as in lamellar and other highly ordered phases will enable more effective use of these materials. The phase behavior of a wide range of surfactants has been reported,1 and this depends on the chemical structure of the surfactants as well as interactions with solvent and other components in a solution. In many cases, simple ideas about the relative size and shape of hydrophilic and hydrophobic moieties allow prediction of the packing of molecules that leads to different micellar shapes and to different ordered structures such as lamellar, cubic, and hexagonal phases.2 Sodium bis 2-ethylhexyl sulfosuccinate, NaAOT, shown in Figure 1, is an anionic surfactant with two branched hydrophobic tails. The hydrophobic and hydrophilic regions of NaAOT have approximately equal cross-sectional areas, and thus it packs efficiently in planar structures such as a lamellar phase and readily forms cosurfactant-free microemulsions.3 NaAOT is widely used in industry as an important component in wetting agents, as a dispersant for pigments and minerals in aqueous r 2011 American Chemical Society
solutions and organic liquids, and in formulations that require microemulsions or reverse micelles in organic liquids. The unusual efficiency in diverse applications has led to extensive research on several aspects of the surfactant properties: the phase behavior has been described in several papers that are based on the original work of Rogers and Winsor.4 In other previous studies, NaAOT has been found to adsorb to both hydrophilic5 and hydrophobic6 silica surfaces. In the present study, the adsorption of the lamellar phase of NaAOT to a (0001) sapphire surface has been investigated. This alumina, Al2O3, substrate has an isoelectric point79 of about pH 67 and therefore, unlike silica, is not expected to be strongly negatively charged. The adsorption of an ionic surfactant may be favored by a surface of opposite charge. When surfactants form lamellar phases at low concentrations, the lamellae will be separated by large amounts of water as shown in Figure 2. This structure is quite different from higher concentration lamellar phases or from bicontinuous phases and other Received: December 9, 2010 Revised: March 10, 2011 Published: March 28, 2011 4669
dx.doi.org/10.1021/la1048985 | Langmuir 2011, 27, 4669–4678
Langmuir
ARTICLE
compositional information about the adsorbed material to be obtained at molecular length scales by providing a profile of scattering density normal to the reflecting surface. In specular neutron reflection, the ratio of the intensity of the reflected beam to that of the incident beam is measured as a function of the momentum transfer normal to the reflecting surface, Q, which is given by Q ¼ ð4π=λÞ sin θ
Figure 1. The chemical structure of NaAOT.
Figure 2. Relevant structural parameters indicated in a very simple schematic illustration of surface adsorbed lamellar phase NaAOT.
mixed phases formed at higher surfactant concentrations10,11 where the thickness of the separating solvent layer may be of the same order as the thickness of the surfactant layer. The effect of temperature on lamellar phase 2.0% wt NaAOT, which corresponds to a concentration of 20 cmc, where the cmc for NaAOT has been established12 to be 2.5 103 M, has been investigated at the solid silica/liquid interface. It was found that the lamellar d-spacing decreased with a temperature increase.13 In previous bulk studies of NaAOT, the same trend was observed.14,15 A similar effect of temperature has also been reported for cationic surfactants.16 In this study, we investigate the structure and effects of small changes of temperature in the range 1530 °C on the adsorbed lamellar phase NaAOT at the solid sapphire/liquid interface. In modeling the reflectivity data, allowance for the fluctuations of the bilayers had to be introduced. The theory of fluctuations of bilayers and lamellae has been reported previously, describing the role of fluctuations in experimental systems,17 the elastic properties of lipid bilayers,18 and steric interactions in multilayer systems.19 Studies that use smallangle scattering to look at bulk phases of surfactants have highlighted the importance of including fluctuations in models of fluid membrane systems.20,21 Fluctuations of bilayers that are either adsorbed or physically deposited at interfaces have been described for a range of materials2224 and observed by X-ray or neutron scattering. In this Article, we present a study of self-assembled, multiple bilayers of NaAOT at the sapphire/solution interface. The study uses neutron reflection, and we describe a simple model of the structure that includes fluctuations of lamellae at the interface.
’ INTERPRETATION OF NEUTRON REFLECTION FROM MULTILAYER STRUCTURES Neutron reflectometry is widely used to study molecules adsorbed at interfaces.25 It allows quantitative, structural, and
where λ is the incident neutron wavelength and θ is the angle of incidence. In specular reflection, the angle of the incident beam is equal to the angle of the reflected beam. The scattering is a nuclear interaction, and information is derived via the scattering length density of the material, given by F¼
∑n i b i
where ni is the number density of the element, and bi is its scattering length. The scattering length density, F, determines the neutron refractive index. Different molecules and isotopes have different values of bi. For example, hydrogen and deuterium have very different scattering lengths as shown in Table 1. Matching F of the solvent with that of the substrate, one can obtain a reflection signal that depends only on the interfacial layer. Making additional measurements with different hydrogen and deuterium composition in the solvent allows one to verify the composition of surface layers as F is related to the volume fraction of each component in the layer by F ¼ js Fs þ jw Fw and
js þ jw ¼ 1
where js and jw are the volume fractions of surfactant and water, and Fs and Fw are their scattering length densities. If a surfactant molecule has a total scattering length bs (from the sum of b for all component atoms), the area per molecule, A, in an interfacial layer of thickness, tL, is 1=A ¼ tL js Fs =bs As the wavelength of neutrons is on molecular length scales, repeating structures from, for example, uniform lamellae will appear as equally spaced Bragg diffraction peaks corresponding to the lamellar spacing. The momentum transfer at Bragg peaks, Qp, gives the interlayer spacing, d, by Qp ¼ 2πn=d where n is an integer and d is the distance between the midpoints of two layers as shown in Figure 2. The first order Bragg peak, where n = 1, gives the interlayer spacing for a lamellar structure. For bilayers of surfactants, the layer thickness, tL, is the sum of different lengths given by tL ¼ 2th þ tt where th is the thickness of the headgroup regions in the bilayer, and tt is the thickness of the tail region, which includes the tails from surfactant molecules that are oriented in both directions in the bilayer and may be intermixed. The overall spacing d is d ¼ tsol þ tL where tsol is the thickness of the solvent layer. Off-specular scattering can also appear, and, in broad terms, it arises from lateral inhomogeneity in interfacial layers, for 4670
dx.doi.org/10.1021/la1048985 |Langmuir 2011, 27, 4669–4678
Langmuir
ARTICLE
Table 1. Materials Used and Values of Molecular Parameters material
formula
RMM
F/106 Å2
density/g cm3
volume/Å3
b/fm
40.60
sapphire
Al2O3
102.0
5.75
4.0
NaAOT
C20H37NaO7S
444.6
0.64
1.16
636
heavy water
D2O
20.0
6.35
1.11
30
19.05
water
H2O
18.0
0.56
1.00
30
1.68
50% heavy water
DHO
19.0
2.90
1.05
30
8.69
NaAOT tail
C8H15
111.0
0.14
0.92
200
2.76
NaAOT head
C4H7NaO7S
222.6
2.03
1.57
236
47.85
example, patches of surfactant bilayers or fluctuations in a bilayer, the so-called Helfrich undulations.18,19 Quantitative interpretation of off-specular scattering has been made for a few systems,26,27 but this generally involves scattering theories such as the distorted wave Born approximation and is computationally intensive even for simple models such as rough surfaces.26 Modeling Multilayers. Reflectivity curves for a given scattering length density profile can be calculated exactly using the optical matrix method of Abeles.28 Imposing the constraints of stoichiometry of head and tail groups has a significant advantage in constraining the fitted curves to realistic models. A further important constraint is to fit simultaneously multiple isotopic contrasts for the same structure: different water contrasts obtained by mixing H2O and D2O thus unambiguously allow the determination of the composition of regions comprised of mixtures of surfactant and solvent. The structural parameters used for deducing the reflectivity profile in this study were the thickness of the surfactant layers, the area per molecule, and the interlayer distance occupied by solvent. Computer programs that exploit this approach have been developed for a single bilayer at an interface29 and for multiple layers.30 Such models will correctly give rise to Bragg peaks in the calculated reflectivity if there are several repeating layers. The intensity and width of the Bragg peaks depend on the number of repeating layers and the contrast within the repeating structure. In the case of models for adsorbed bilayers, it is straightforward to include gradients in the composition or even small variations in the interlayer spacing between adjacent layers. Modeling the data in this way gives a good indication of certain parameters of the structure such as the spacing between lamellar layers and the thickness of the hydrocarbon region. These programs allow the known instrumental resolution to be convoluted with the models in the fitting process. Thermal fluctuations of bilayers are expected, and because these fluctuations will change the average density distribution in the various layers parallel to the interface, it is necessary to include these in models of the data. The smearing of the composition profile in the direction perpendicular to the surface is distinctively different from simple intermixing or roughness of adjacent layers that can be included in optical calculations using the methods of Nevot and Croce.31 A single bilayer is expected to remain substantially unperturbed except in local curvature with an unchanged thickness of tails and separation of the two regions of head groups. These three layers can be considered as fluctuating coherently. If the individual bilayers are separated by a large thickness of solvent, fluctuations of adjacent lamellae will be uncorrelated, and the intensity of the Bragg peaks is reduced. An important constraint of the optical matrix method in the modeling of roughness is that its only valid if the magnitude of the roughness is less than the thickness of the layer.
An alternative route to model reflectivity is to use an approach that is analogous to that used in crystallography. This approach is similar in concept to that described by Tidswell et al.32 and Penfold et al.33 and has the advantage that there are aspects of the scattering from periodic structures such as uncorrelated thermal fluctuations that are easy to incorporate in calculations that treat the scattered intensity as a Fourier transform of a density distribution. However, the reflection at low Q, near the critical angle for total reflection, is high, and the approximation of weak scattering that is implicit in the interpretation of data as a Fourier transform is no longer valid. A simple means to avoid this difficulty that has been used previously32,33 is to consider data that are divided by the calculated reflectivity of an ideal sharp interface between the two bulk materials known as the Fresnel reflectivity, RF(Q). It is then possible to consider the contribution of surface layers to the reflectivity as a small perturbation and model RS(Q) = R(Q)/RF(Q). Our approach is slightly different: we can consider, in place of RF(Q), a model for reflectivity, RI(Q), that includes surface roughness and possible hydration of the solid substrate as well as a dense, near-surface bilayer of adsorbed surfactant. We have used this as a divisor to provide a data set that is then modeled using a simple model of scattering from a periodic structure. The criterion for choosing RI(Q) is that it should reproduce the features of the observed reflectivity apart from Bragg diffraction peaks. This is equivalent to requiring that “background” for RC(Q) = R(Q)/RI(Q) is close to zero and flat. RC(Q) can then be simulated by simple crystallographic models. Disorder such as that caused by thermal fluctuations can be included in the model with a DebyeWaller factor34 that reduces the intensity of the Bragg peaks by a factor of e2W where 2W = (Q2ξ2)/3 and ξ is the root-mean-square displacement amplitude of the repeating structures. The diffraction from a perfect lamellar structure is straightforward to calculate as it consists of equally spaced peaks. The intensity of the peaks is dependent on the contrast in the structure (scattering length density difference) and the number of layers in the crystal as well as being modulated by a function that describes the shape or structure of an individual lamellar sheet. The width of the Bragg peaks depends on the size of the crystal (number of layers) and the experimental resolution. RC(Q) can therefore be modeled directly with physical parameters that would include a DebyeWaller factor and hence root-mean-square amplitude of fluctuations. An alternative to fitting a model to RC(Q) is to recognize that the arguments presented in the preceding two paragraphs suggest that the reflectivity may be dominated by Bragg peaks when there is a multilayer structure near the surface. The relative intensity of the Bragg peaks is modified by a DebyeWaller function that can be included directly. This is simply an extra multiplicative term in 4671
dx.doi.org/10.1021/la1048985 |Langmuir 2011, 27, 4669–4678
Langmuir
Figure 3. Assembly of the cell indicating its ability of use in neutron reflection and small-angle scattering geometry. The sample is held in place by a PTFE insert sandwiched between two crystals of choice. For horizontal geometry, the cell is turned through 90°.
the results of an optical matrix calculation that reduces the calculated reflectivity depending on the amplitude of the incoherent fluctuations between layers. Other parameters for the multilayer structure, including roughness and lamellar structure, can be modeled in the usual manner. This approach used in the program byban35 would be valid for reflectivity curves in the ranges of Q that are dominated by Bragg peaks.
’ EXPERIMENTAL SECTION Reflectivity measurements were made on the neutron reflectometer SURF at the ISIS Facility, Rutherford Appleton Laboratory, Chilton, U.K.36 The time-of-flight data were measured with a single detector for two or three angles, 0.35° and 0.8° or 0.25°, 0.7°, and 1.5 degrees, respectively. Complementary small-angle and grazing incidence scattering measurements were made with the D22 small-angle scattering instrument at the Institut Laue Langevin, Grenoble, France.37 The distance to the position sensitive 1 m 1 m area detector was 10 m, and each pixel has an area of 8 8 mm2. These complementary measurements were made to establish the relationship between the adsorbed material at the interface and the bulk material, which coexist in an equilibrium relationship. The measurements at SURF and D22 were made in the same cell. For reflectivity measurements on SURF, the total acquisition time for the full range of Q was about 2.5 h. On D22, scans of angle were made, and the acquisition time ranged from 10 s at the lowest angle to 10 min at the largest measured angle. Each scan with about 50 points took approximately 2 h including the time to move the sample. The smallangle scattering in transmission geometry was recorded for 10 min. A sample holder has been developed that allows measurement of both neutron reflection from a solid/liquid interface and small-angle scattering from the bulk of the liquid. This is shown in Figure 3. The surface (50 50 10 mm) to be studied is clamped in an aluminum frame with a piece of PTFE between it and another solid that is chosen to be
ARTICLE
transparent to neutrons. In these experiments, the back surface was an unpolished silica crystal. In grazing incidence experiments, the beam enters through one of the small faces of the sapphire crystal and is reflected out of the sample cell through the opposite edge. For smallangle scattering experiments in transmission geometry, the sample mount allows rotation through 90° so that the incident beam is normal to both solid surfaces. The large holes in the aluminum frame allow scattering to be measured over a wide range of angles. The PTFE spacer was chosen to have a thickness of 2 mm in the area between the crystals so as to provide an appropriate path length for small-angle scattering experiments. The sample cell has a number of special features: there is a circuit for water from a circulating bath to allow the temperature of the sample to be controlled. The temperature is monitored with a Pt100 resistance thermometer that is screwed in to the aluminum frame. Small isolating spacers made of polyoxymethylene reduce the thermal contact between the frame holding the crystals and the support. Extra masks of cadmium that are stuck to aluminum supports can be mounted and adjusted to reduce background scattering. The mounting system is designed so that the surface to be studied can be either horizontal or vertical depending on the geometry of the instrument. The PTFE between the crystals is machined so that tube connectors from a liquid chromatography pump can be attached. The pump can be programmed to inject specific mixtures from appropriate stock solutions. At SURF the pump was a Hitachi L-7100 LaChrom and at D22 a Knauer Smartline 1000 HPLC pump with Smartline Manager 5000 degasser unit. The volume of the cell is about 3 mL, and in our experiments we flushed 10 mL of solution through the cell with a flow-rate of 2 mL min1 to ensure complete exchange of the sample. The choice of the “back surface” for the cell is necessarily a compromise. Silicon provides a high transparency for neutrons and good thermal conductivity. It can also be cleaned readily. An unpolished, rough surface was chosen so that there was negligible reflection from the back interface. This increases the background scattering for the small-angle scattering measurements. For the present experiments, this was not a problem as the small-angle scattering signal from the solution was significantly stronger. Sodium bis 2-ethylhexyl sulfosuccinate (98% purity) NaAOT, from Sigma-Aldrich, was purified by liquidliquid extraction using the procedure of Li et al.38 Briefly, NaAOT was dissolved in pure water, and impurities were removed by extraction with heptane. The product was obtained by freeze-drying from the aqueous phase. Pure water (18 MΩ cm) was obtained from a Millipore system, and D2O was from Euriso-top and from Sigma-Aldrich. The (0001) sapphire surface was purchased from PiKem and cleaned with a “dilute piranha” solution39 consisting of pure water, concentrated sulfuric acid, and 30% hydrogen peroxide solution in the ratio 5:4:1 at a temperature of about 80 °C for 15 min, followed by extensive rinsing by Millipore water.40 The other parts of the sample cell and connecting tubing were cleaned with Decon 90 followed by extensive rinsing with Millipore water.
’ RESULTS AND DISCUSSION Structure and Adsorption of NaAOT above cmc in Pure Water at 25 °C. The neutron specular reflectivity profile shown
in Figure 4 was measured for 2% wt NaAOT (20 cmc) in D2O. The reflectivity curve contains three sharp peaks that fulfill the Bragg condition for a lamellar structure with peaks appearing at Qp, 2Qp, and 3Qp. In addition to the three Bragg peaks, the reflectivity curve also contains three very small peaks or “shoulders”, which do not fulfill the Bragg condition for the simple lamellar structure and which were only observed in some of the measurements. Although the origin of these smaller peaks is unclear, it is possible that they arise from a modulation of the
4672
dx.doi.org/10.1021/la1048985 |Langmuir 2011, 27, 4669–4678
Langmuir
Figure 4. (a) Adsorption of 2% wt NaAOT/D2O at 25 °C to sapphire and fit (solid line) including the repeating bilayer structure with fluctuations of the lamellae. Also shown (dashed line) is the fit, RI(Q), for the bilayer close to the sapphire surface that is used when applying the crystallographic model. The inset shows the model fit to RC(Q) = R(Q)/RI(Q) shown on a linear scale including fluctuations in the lamellae. (b) Scattering length density profile derived from model fit.
structure. The intensity of the extra peaks is very weak as compared to the main peaks, and they have not been included in the model fit. The lower concentration boundary for the lamellar phase of NaAOT has been discussed, and values from bulk measurements at 20 °C in the range 817% wt has been presented by various authors.4,41,42 However, the observed adsorption from 2% wt NaAOT is in clear contrast to the adsorption of a single bilayer that gives rise to fringes in the reflectivity but with no sharp peaks. The model fit obtained for 2% wt NaAOT in D2O is shown in Figure 4. Near the surface is a 35 ( 2 Å thick NaAOT bilayer. The tails in this layer are packed at the maximum density that is possible for a liquid-like hydrocarbon and is similar to that seen in previous measurements of 0.1% wt NaAOT.43 The head groups on both sides of this bilayer are hydrated with about 9 water molecules per NaAOT molecule, but the inner layer is not separated from the sapphire substrate by any region of pure water. This bilayer that is adsorbed near the surface is followed by a repeating stack of laterally densely packed fluctuating bilayers of thickness tL 51 ( 2 Å and with an area per molecule of just 40 ( 2 Å2. The lamellar bilayers are separated by a large thickness, 148 ( 2 Å, of solvent. The surfactant layer thickness, tL, is constrained to include both surfactant heads and tails. As the surfactant layer is fluctuating, this region also includes a substantial volume of solvent, about one-third by volume. The layer would be well-defined by the hydrocarbon-rich surfactant tails. However, as the surfactant layer includes fluctuations that spread
ARTICLE
out the scattering length density, it is not possible to determine precisely the details of the bilayer structure. The amount of material in the lamellar phase may vary laterally with the level of hydration, either through a uniform change of density within the lamellar layers, or as dense lamellar layer fragments, separated by solvent. The coherence length of the neutron beam is estimated to be more than 1 μm, and the reflectivity will arise from averages of the scattering length for heterogeneous lateral distributions on a scale smaller than this length. If the surface layer were to consist of dense patches that were larger than the coherence length rather than uniform layers, the reflectivity would be the average of that for the surface areas with different coverage. The model described here includes a near-surface layer that is so densely packed that that patches cannot be envisaged as the AOT tails are at the density of the liquid hydrocarbon. Our previous study of concentrations near the critical micelle concentration43 has also indicated that the coverage was uniform close to the surface. The present work cannot rule out entirely that the multilayers are present as patches. However, the fit of the model to the data is better with a gradient of surfactant density in laterally uniform layers rather than a model with less coverage throughout uniformly packed lamellar layers. A simple approach of just determining the intensity of the series of Bragg peaks as a function of Q and taking the slope of a logarithmic plot does not provide a good means to analyze the fluctuation amplitude data for two reasons. First, a significant component of the peak intensity variation arises from the shape of the bilayer repeat. This effect is included in the reflectivity calculations as the thickness of the bilayer and can also be incorporated when fitting RC(Q) directly with a crystallographic model by multiplying with an appropriate form factor. If this term is ignored, the estimated fluctuation amplitude will be erroneously large. A second concern is that the estimate of the background and the reflectivity of the bilayer that is near sapphire surface can significantly alter the estimated intensity of the weak high order Bragg peaks. For these reasons, the preferred approach has been to include these parameters in the model that is used to fit the data. For the 2% wt NaAOT in D2O shown in Figure 4, the second lamellar layer contains molecules with an area of 40 ( 2 Å2. Moving out in the stack toward the bulk, the area per molecule was found to be 42 ( 2 Å2. This is a rather small change and may be due to the history of the sample. For technical reasons, the sample had been left in the cell for an hour before the measurement was made. This would have allowed the sample a longer time to equilibrate. To adequately model the relative intensities of the Bragg peaks, a simple model containing only a gradient in lamellar composition was not sufficient. To provide a reasonable fit, the introduction of an extra term to allow for uncorrelated fluctuations of the interlayer separation had to be included in the model as described in the Interpretation section. Best fits were obtained with a value of fluctuation amplitudes of about 23 Å. When determining the exact number of repeating layers, when the number of repeating layers is large, the difference in scattering is not significant. The depth in sample that can be probed with the neutron beam is limited to about 1 μm. For a low number of lamellar repeats, the width of the Bragg peaks would be strongly correlated to the number of repeating lamellar layers, and the peak intensity depends on both contrast and number of repeats. The number of aligned layers in this study was deduced to be about 20 ( 10 layers. 4673
dx.doi.org/10.1021/la1048985 |Langmuir 2011, 27, 4669–4678
Langmuir
Figure 5. Scattering pattern from a GiSANS measurement with the D22 instrument for 2% wt NaAOT in D2O at 25 °C. Specular reflectivity is obtained through integration of the region around the reflected beam for the separate measurements at each angle. For SURF data, the reflected beam is broader as compared to D22 in the vertical direction shown. Off-specular scattering can be seen outside the area of the reflected specular beam.
An inset in Figure 4 shows the data from 2% wt NaAOT in D2O converted to RC(Q), fitted with the crystallographic model. RI(Q) used to obtain RC(Q) includes the bilayer that is adsorbed directly at the sapphire surface and consists of a region of surfactant with thickness 35 Å that is the same as that observed at 0.1 and 0.3% wt in our previous study.43 The fit shown in the inset in Figure 4 corresponds to fluctuating lamellae with an increasing level of hydration further out in the stack or, alternatively, lamellar islands separated by solvent and represents the same physical structure as that obtained as a fit in the optical calculations. The introduction of a fluctuation amplitude, ξ, in the bilayers of the order 25 ( 5 Å accommodates the observed intensity drop of the Bragg peaks. These plots and the fitted parameters justify the approximation made in the optical calculation that the Bragg peaks are dominant in the reflectivity. The relative amplitude of the fluctuations also suggests that the division by RI(Q) rather than RF(Q) is an adequate approach. This can be justified by scattering theory if the fluctuations of the lamellar crystal are large as compared to the thickness of the bilayer. In practice, the small difference between RC(Q) and RS(Q) is at the level of the background (inset, Figure 4). The analysis of reflectivity data for adsorption of lamellar solutions of NaAOT presents several challenges: the surface layers may not be laterally uniform because of the dynamic fluctuations of the layers or patches in coverage. This will affect the scattering from the material adsorbed at the surface. Further, the interfacial layers are present in equilibrium with a solution that in itself scatters strongly. Measurements in grazing incidence geometry provide a picture of the significant off-specular scattering contribution to the background as shown in Figure 5. The data shown in Figure 5 are part of a series of measurements at many different angles of incidence (measurements with point collimation and a monochromatic beam) and indicate that for the NaAOT/sapphire system, the scattering from the dispersed NaAOT in the bulk is significant. The reflectivity data measured on SURF as shown in Figure 4 were recorded with a range of wavelengths and a fixed single detector. It has not had any background subtracted and will therefore include contributions from surface scattering that are integrated across the lateral resolution as described by Fukuto et al.44 as well as some small-angle scattering background. For this reason, the interpretation of data in this study has been restricted to the major features of the reflectivity curves such as the Bragg peaks and their integrated intensity.
ARTICLE
Figure 6. Measurements for an adsorbed layer of 2% wt NaAOT/D2O at 25 °C with different contrasts of solvent obtained by changing the ratio of D2O to H2O fit the same model. The 100%, 75%, and 50% D2O curves with fit shifted by one for clarity.
At Q greater than 0.1 Å1, the other features of the surface layer and background small-angle scattering are dominant, and so the model fitting is constrained to Q less than 0.1 Å1. To be able to verify the validity of the model and to exclude an isotope effect, additional measurements in varying hydrogen and deuterium composition were made as shown in Figure 6. All four reflectivity curves fit the same model. Closest to the surface is the 35 ( 2 Å NaAOT bilayer followed by a repeating stack of lamellae with a layer thickness, tL, of 40 ( 2 Å, separated by 140 ( 2 Å solvent, tsol. The area per molecule in the repeating lamellae closest to the sapphire surface was found to be 53 ( 2 Å2, increasing to 75 ( 2 Å2 closer to the bulk. The fluctuations were found to be of order 30 Å. The values for the models to the data shown in Figure 6 vary somewhat as compared to the model that fits the data in Figure 4. This is due to the time difference mentioned above and/or the slight variation in the temperature control for the data presented in Figure 6. In this study, an adsorbed lamellar phase was sometimes observed for 1% wt NaAOT with the same d-spacing as observed for 2 wt %. These reflectivity data are shown as Figure S1 in the Supporting Information. In comparison, previous NaAOT studies from reflectivity measurements on silica13 have reported the d-spacing between lamellae to increase on increasing the concentration from 2 to 5 wt % NaAOT. The surface d-spacing was seen to be systematically smaller than in the bulk. It was also reported from reflectivity measurements on silica and air/liquid experiments5,24 the d-spacing being constant in changing the concentration from 2 to 5 wt %. It is not easy to draw any clear conclusions from these varying results. However, it should be noted that the d-spacing of the lamellar phase is well-defined by the peak position Qp. Effect of Temperature. NaAOT shows unusual temperature sensitivity in the range 1530 °C, as can be seen from the reflectivity data shown in Figure 7, and the corresponding fit parameters presented in Table 2. The effect of temperature is clearly visible in that the positions of the first two Bragg peaks move to higher Q. This is similar to previous observations at the silica/liquid interface.5,13 From model fits to the data, the lamellar d-spacing was found to decrease significantly with a small temperature increase, and the volume occupied per molecule increases, as shown in Figure 8. The intensity of the Bragg 4674
dx.doi.org/10.1021/la1048985 |Langmuir 2011, 27, 4669–4678
Langmuir
Figure 7. Effect of temperature on adsorbed 2% wt NaAOT in D2O. (a) As temperature is increased from 15 to 30 °C, the Bragg peaks shift to higher Q as the d-spacing decreases, and the intensity of the peaks decreases significantly. The effect is reversible. The 25, 20, and 15 °C data with fit are shifted by one for clarity. (b) Scattering length density profile derived from model fits of 15 and 30 °C temperature data.
peaks decreases with temperature partly because of the lower amount of material in the lamellae. This is clearly visible in the scattering length density profile in Figure 7. Possible changes in the fluctuation amplitude with temperature are not resolved in the experiment because of the small number of Bragg peaks that can be observed. This decrease in Bragg peak intensity with temperature is in contrast to the results of Li et al.5 at the silica/ water interface, where increasing temperature led to increased intensity and definition of the Bragg peaks. NaAOT forms charged bilayers, and so if the amount of material in the lamellar bilayers decreases with temperature, so does the charge density of the layers. As the lamellar separation is large, the 2 wt % NaAOT system should be dominated by longrange electrostatic forces. Hence, a change in the charge density of the layers may be sufficient to cause the change in interlayer separation. Models of temperature data show the area per molecule and hence the level of hydration to increase with temperature (Figure 8). The thickness of the layers also decreases slightly. This results in an overall increase in the volume of the NaAOT molecules. This may be due to the molecules
ARTICLE
packing less densely in the lamellar bilayers, or it may be due to the lamellar bilayers breaking up into islands as discussed above, the resulting space being filled with solvent. One might also consider the observation that, although the surfactant density decreases further away from the surface, the lamellar d-spacing changes uniformly throughout the system with temperature. This would be a supporting argument for the intra lamellar packing being constant, with lamellae breaking up into fragments further out in the stack. A decrease in lamellar d-spacing with temperature has been observed for other systems similar to that of NaAOT. A lamellar system of the cationic dichain surfactant N,N-didodecyl-N,Ndimethylammonium bromide (DDAB, 2% wt) was in a neutron reflection study at the silica/water interface16 found to have a dspacing that decreased with temperature. A small-angle scattering experiment showed that the area per molecule of DDAB changed less than 10% with temperature and was described as temperature independent.45 This may be due to the fact that at 2% wt, DDAB forms a lamellar phase containing the same concentration as the average bulk solution, which is in contrast to 2% wt NaAOT where the concentration of surfactant in the lamellar phase is significantly higher than that of the average concentration of the coexisting bulk phase. A study on more concentrated NaAOT in brine10 (25% wt NaAOT and 1.5% wt NaCl) describes steric repulsions as an important factor governing the system, as the NaAOT layers in this system are closely packed together. The momentum transfer Qp for the reported sponge phase shows a small increase in going from 27 to 30 °C, indicating a decrease in d-spacing, comparable to the present study, followed by a decrease in Qp with temperatures up to 55 °C. A peak from pure lamellar phase NaAOT is first detected at 33 °C. Screening may mean that NaAOT in the more concentrated system is dominated by electrostatic interactions. Also, as the d-spacing in this system is one-half of that in the present study, the fluctuations of the NaAOT layers become an important factor. In this study, fluctuations did not change significantly with increasing temperature, as shown in Table 2. One might argue that fluctuations should increase with temperature through increased thermal motion of molecules. However, molecules may also thermally increase their motion laterally, restrained by the hydrophobic interaction to protect the hydrocarbon chains. This would then have the effect of a slight overall decrease in layer thickness as the molecules would accommodate for the increased motion of the tails to be within the lamellar layer as shown schematically in Figure 9. At 15 °C, the molecules pack very densely with the more rigid hydrophobic tails within the layer. As the temperature is increased, the tails pick up thermal energy, and, for the tails to still be within the layer, the molecules pack further apart laterally. This enables them to form a thinner layer as is indeed observed. Another plausible explanation as to why the fluctuation amplitude was not observed to change with temperature is that if the temperature increase were to result in lamellar layers breaking up into islands, then the fluctuation wavelength would be expected to decrease and consequently change the amplitude.1719 For the 2% wt NaAOT system, the temperature effect in the range 1530 °C is completely reversible. This is an indication of the change of temperature inducing physical change, with no change in the chemistry of the system. In contrast, the study on DDAB mentioned above showed some changes upon heating/ cooling cycles.16 The DDAB bulk and surface material displayed 4675
dx.doi.org/10.1021/la1048985 |Langmuir 2011, 27, 4669–4678
Langmuir
ARTICLE
Table 2. Parameters for 2% wt NaAOT with Change of Temperature temperature/°C
d-spacing/Å
layer thickness
solvent thickness
(tL)/Å
(tsol)/Å
fluctuations/Å
area per molecule,
area per molecule,
inner layer (A)/Å2
outer layer (A)/Å2
15
213 ( 2
58 ( 2
155 ( 2
30 ( 5
31 ( 2
43 ( 2
20
203 ( 2
52 ( 2
151 ( 2
29 ( 5
31 ( 2
48 ( 2
25
195 ( 2
50 ( 2
145 ( 2
21 ( 5
53 ( 2
78 ( 2
30
185 ( 2
45 ( 2
140 ( 2
24 ( 5
64 ( 2
119 ( 2
Figure 8. Parameters extracted from temperature change data of the lamellar phase with 2% wt NaAOT in D2O. (a) Lamellar spacing decrease with temperature. (b) Molecular volume (for the NaAOT immediately following the tightly bound, near-surface layer) increase with temperature. The error bars are too small to be visible.
Figure 10. Reflectivity versus Q for neutron reflection and I versus Q for small-angle scattering for 2% wt NaAOT/D2O at 25 °C. The Q-value of the peaks and hence the d-spacing are the same both at the surface where the lamellar phase is aligned and in the isotropic bulk.
interpretation of small-angle scattering from the bulk of a solution,47 the absolute intensity, I(Q), can be expressed as IðQ Þ ¼ APðQ ÞSðQ Þ
Figure 9. Schematic representation of the temperature effect on 2% wt NaAOT/D2O. The separation between layers and the density of the NaAOT layer decrease with temperature. When the surfactant packing is less than its maximum density, it is not possible to distinguish between an evenly spread out reduction of density of the lamellar layers, or the formation of densely packed islands of surfactants separated by solvent.
different temperature behavior: the spacing in the bulk was systematically less than that of the layers at the surface, and the results for heating/cooling cycles were less reproducible for the bulk material. Comparison with Structure in Bulk Solution. To get a more complete picture of the aligned lamellar system and its coexisting bulk phase, further measurements were made in grazing incidence (Figure 5) and small-angle scattering geometry (twodimensional scattering patterns can be found in the Supporting Information, Figure S2). A scan was made for many different angles of incidence with a fixed wavelength. Reflectivity was evaluated for each angle by integrating the region of the specular peak. A background was estimated by integration of adjacent regions on the detector. The benefits of these simultaneous measurements are discussed in more detail elsewhere.46 For
where P(Q) is known as the form factor and is determined by the size and shape of the dispersed objects, S(Q) is the structure factor that describes the correlations between dispersed particles, and A depends on the contrast. Models for P(Q) and S(Q) for lamellar structures that include thermal fluctuations of elastic membranes are described in the literature.20 In this study, a concentration range of 0.82 wt % NaAOT was studied with small-angle neutron scattering. A lamellar phase was formed at concentrations above about 1.2% wt NaAOT in coexistence with an isotropic bulk phase. It was found that the interlayer separation for bulk and surface structures were approximately the same, as the Bragg peaks appear at about the same Qp as shown in Figure 10, and do not change proportionately with increases in concentration. The bulk interlayer separation is in agreement with previous X-ray studies on bulk NaAOT at concentrations of roughly 10% wt.14,48 The bulk and the surface structures have similar interlayer separation; however, the material in the bulk lacks the alignment induced by the surface. One can assume the bulk isotropic phase to resemble the outer regions seen in reflection experiments. This would mean that the bulk phase layers would contain a lesser amount of surfactant than on average seen at a surface, where the exact boundary between surface structures and bulk should be considered. In a previous study of dilute NaAOT at a sapphire surface, it was found that each headgroup in the bilayer is 4676
dx.doi.org/10.1021/la1048985 |Langmuir 2011, 27, 4669–4678
Langmuir associated with about nine solvent molecules.43 For the present study, this would correspond to a concentration of about 10% wt NaAOT at the surface, which is significantly higher than the average bulk solution with a concentration of 2% wt, in equilibrium with the surface adsorbed layers. The measurements in grazing incidence geometry as seen in Figure 5 reveal information about the relative intensities of the scattering from the interface and the bulk. It also gives a clear indication of the level of off-specular and background scattering. The 2D scattering pattern shows the reflection peak, the first- and second-order Bragg peaks, as well as a scattering peak arising from roughness or fluctuations near the interface often referred to as Yoneda scattering.49 This scattering is not present in measurements from pure solvent and hence can be attributed to the NaAOT layers at the interface. The Qp values for the data in Figure 10 are similar to those obtained for the 30 °C data shown in Figure 6. There is a small difference of about 3% in the position of the peaks in the reflection and scattering curves. This may be due to the unavoidable exposure of the sample to a higher ambient temperature near the instrument prior to injection in the sample cell. The difference in spacing is of similar magnitude to that seen on changing the temperature by about 2 °C, and so we would not wish to interpret this as a specific surface effect. Analysis of off-specular scattering has been discussed by Malaquin et al.23 specifically for the case of lipid bilayers that resemble in many ways the structures formed by NaAOT molecules. Their calculations explicitly use the model for bilayer fluctuations developed by Helfrich and analyze the diffuse scattering in terms of the diffusing wave Born approximation. However, their study could be made in the absence of any significant concentration of molecules in the bulk solution, and so the scattering that is observed is dominated by the interfacial structure. The data shown in Figure 5 and in the Supporting Information indicate that the scattering in the present study arises substantially from the bulk of the solution. This cannot be simply subtracted, and so the quantitative analysis of the diffuse off-specular scattering is not feasible with the methods that have been described to date.
’ CONCLUSIONS NaAOT forms a lamellar phase even at concentrations of about 1% wt. Near a planar interface, the lamellar structure is seen to be strongly aligned parallel to the surface and extend over the entire interface to a depth of 10 or more bilayers. The material in the bulk lacks the orientational alignment induced by the surface. The NaAOT multilayer structure on sapphire is formed on a dense bilayer of adsorbed NaAOT at the surface. Adsorption at the sapphire surface is similar to the adsorption at hydrophilic and hydrophobic silica surfaces, and this suggests that interface charge is not significant in determining the adsorption. The present study has explored the effects of concentration and temperature. The sharp Bragg peaks observed in the neutron reflection profiles provide a good determination of the interlayer spacing in the lamellar structure. For NaAOT, the interlayer spacing is 199 Å at a concentration of 2% wt at 25 °C. The NaAOT molecules are constrained into bilayers with an average overall thickness of 51 ( 2 Å packed closely laterally with an area per molecule of just 40 ( 2 Å2 near the interface, separated by 148 ( 2 Å solvent. This corresponds to a concentration of about
ARTICLE
10% wt at the surface, which is significantly higher than the average concentration of the coexisting bulk solution. Location of the ordered phase near the interface allows reflection and scattering from the interface to be used to determine details of the lamellar structure. Modeling of the structure requires both some disorder in the lamellar crystal density and fluctuations that reduce the intensity of Bragg peaks and can be described by a DebyeWaller term and have amplitudes of approximately 25 Å. This could correspond to Helfrich fluctuations. The changes observed with temperature in the range 1530 °C are very pronounced with regards to both the amount of adsorbed material and the structure. The lamellar spacing decreases as the temperature increases. The material at the surface is in equilibrium with a large excess in the bulk, and so one cannot exclude that this change in dimension would correspond to a different amount of material. The area per molecule in the surface layers found from models of the reflectivity increases with increasing temperature. This is an indication of a reduction in surfactant density within the lamellae, or possibly the breaking up of lamellae, the space being filled with solvent. The layer thickness decreases slightly with temperature, resulting in an overall increase in volume per molecule. Small-angle neutron scattering experiments were performed to determine the structure in the bulk. Bragg peaks from the bulk solution appear at the same Q-value as for the surface adsorbed lamellar phase. This implies that the d-spacing is the same in the bulk and at the surface. However, the bulk solution lacks the alignment observed at the surface. Grazing incidence smallangle scattering helps determine the relative contributions of the reflected signal from the surface and the bulk, as well as indicates a reasonable background level. Scattering from roughness or fluctuations near the interface, so-called Yoneda scattering, was also detected in these measurements.
’ ASSOCIATED CONTENT
bS
Supporting Information. Figures showing further data measured for grazing incidence and small-angle scattering as well as reflectivity data for a 1% wt solution. This material is available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION Corresponding Author
*E-mail:
[email protected].
’ ACKNOWLEDGMENT We are grateful to the ISIS Facility, Chilton, Oxfordshire, U.K., and the Institut Laue Langevin, Grenoble, France, for allocations of beam time for these measurements. We thank L. Porcar for excellent support at D22 and C.-J. Englund for his help with the design of the sample holder. ’ REFERENCES (1) Laughlin, R. G. The Aqueous Phase Behavior of Surfactants; Academic Press: London, U.K., 1996. (2) Israelachvili, J. Intermolecular and Surface Forces, 2nd ed.; Academic Press: London, U.K., 1991. (3) Shinoda, K.; Lindman, B. Langmuir 1987, 3, 135–149. (4) Rogers, J.; Winsor, P. A. J. Colloid Interface Sci. 1969, 30 247–256. 4677
dx.doi.org/10.1021/la1048985 |Langmuir 2011, 27, 4669–4678
Langmuir (5) Li, Z. X.; Lu, J. R.; Thomas, R. K.; Weller, A.; Penfold, J.; Webster, J. R. P.; Sivia, D. S.; Rennie, A. R. Langmuir 2001, 17, 5858–5864. (6) Li, Z. X.; Lu, J. R.; Fragneto, G.; Thomas, R. K.; Binks, B. P.; Fletcher, P. D. I.; Penfold, J. Colloids Surf., A 1998, 135, 277–281. (7) Kershner, R. J.; Bullard, J. W.; Cima, M. J. Langmuir 2004, 20, 4101–4108. (8) Franks, G. V.; Meagher, L. Colloids Surf., A 2003, 214, 99–110. (9) Zhang, L.; Tian, C.; Waychunas, G. A.; Shen, Y. R. J. Am. Chem. Soc. 2008, 130, 7686–7694. (10) Hamilton, W. A.; Porcar, L. J. Phys.: Conf. Ser. 2010, 251 012034. (11) Hamilton, W. A.; Porcar, L.; Butler, P. D.; Warr, G. G. J. Chem. Phys. 2002, 116, 8533–8546. (12) Li, Z. X.; Lu, J. R.; Thomas, R. K.; Penfold, J. J. Phys. Chem. B 1997, 101, 1615–1620. (13) Li, Z. X.; Weller, A.; Thomas, R. K.; Rennie, A. R.; Webster, J. R. P.; Penfold, J.; Heenan, R. K.; Cubitt, R. J. Phys. Chem. B 1999, 103, 10800–10806. (14) Zhuang, W.-C.; Chen, X.; Yang, C.-J.; Wang, L.-Y.; Chai, Y.-C. Acta Phys.-Chim. Sin. 2005, 21, 1055–1058. (15) Park, D.; Rogers, J.; Toft, R. W.; Windsor, P. A. J. Colloid Interface Sci. 1970, 32, 81–89. (16) McGillivray, D. J.; Thomas, R. K.; Rennie, A. R.; Penfold, J.; Sivia, D. S. Langmuir 2003, 19, 7719–7726. (17) Safran, S. A. Surf. Sci. 2002, 500, 127–146. (18) Helfrich, W. Z. Naturforsch. 1973, 28C, 693–703. (19) Helfrich, W. Z. Naturforsch. 1978, 33a, 305–315. (20) Nallet, F.; Laversanne, R.; Roux, D. J. Phys. II 1993, 3, 487–502. (21) Roux, D.; Safinya, C. R. J. Phys. (Paris) 1988, 49, 307–318. (22) Salditt, T.; M€unster, C.; Mennicke, U.; Ollinger, C.; Fragneto, G. Langmuir 2003, 19, 7703–7711. (23) Malaquin, L.; Charitat, T.; Daillant, J. Eur. Phys. J. E 2010, 31, 285–301. (24) Li, Z. X.; Lu, J. R.; Thomas, R. K.; Penfold, J. Faraday Discuss. 1996, 104, 127–138. (25) Penfold, J.; Thomas, R. K. J. Phys.: Condens. Matter 1990, 2 1369–1412. (26) Sinha, S. K.; Sirota, E. B.; Garoff, S.; Stanley, H. B. Phys. Rev. B 1988, 38, 2297–2311. (27) Muller-Buschbaum, P. Anal. Bioanal. Chem. 2003, 376, 3–10. (28) Abeles, F. Ann. Phys. 1950, 11, 307–309 & 310314. (29) Rennie, A. R., http://material.fysik.uu.se/Group_members/ adrian/bike.htm. (30) Rennie, A. R., http://material.fysik.uu.se/Group_members/ adrian/tram.htm. (31) Nevot, L.; Croce, P. Rev. Phys. Appl. 1980, 15, 761–779. (32) Tidswell, I. M.; Ocko, B. M.; Pershan, P. S.; Wasserman, S. R.; Whitesides, G. M.; Axe, J. D. Phys. Rev. B 1990, 41, 1111–1128. (33) Penfold, J.; Tucker, I.; Thomas, R. K.; Taylor, D. J. F.; Zhang, J.; Zhang, X. L. Langmuir 2007, 23, 3690–3698. (34) Squires, G. L. Introduction to Thermal Neutron Scattering; Dover: Mineola, NY, 1996. (35) Rennie, A. R., http://material.fysik.uu.se/Group_members/ adrian/byban.htm. (36) Penfold, J.; Richardson, R. M.; Zarbakhsh, A.; Webster, J. R. P.; Bucknall, D. G.; Rennie, A. R.; Jones, R. A. L.; Cosgrove, T.; Thomas, R. K.; Higgins, J. S.; Fletcher, P. D. I.; Dickinson, E.; Roser, S.; McLure, I. A.; Hillman, A. R.; Richards, R. W.; Burgess, A. N.; Simister, E. A.; White, J. W. J. Chem. Soc., Faraday Trans. 1997, 93, 3899–3917. (37) www.ill.fr/YellowBook/D22/. (38) Li, Z. X.; Lu, J. R.; Thomas, R. K. Langmuir 1997, 13, 3681–3685. (39) Turner, S. F. Ph.D. Thesis, Cambridge University, United Kingdom, 1998. (40) Caution: Strong oxidizing acids such as dilute piranha solution must be handled and disposed of with caution. Contact of this solution with oxidisable materials such as organic solvents must be avoided as there is an explosion hazard. All work must be performed in a fume hood, and protective clothing must be worn.
ARTICLE
(41) Fontell, K. J. Colloid Interface Sci. 1973, 44, 318–329. (42) Ekwall, P.; Mandell, L.; Fontell, K. J. Colloid Interface Sci. 1970, 33, 215–235. (43) Hellsing, M. S.; Rennie, A. R.; Hughes, A. V. Langmuir 2010, 26, 14567–14573. (44) Fukuto, M.; Gang, O.; Alvine, K. J.; Pershan, P. S. Phys. Rev. E 2006, 74, 031607. (45) Zemb, Th.; Gazeau, D.; Dubois, M.; Gulik-Krzywicki, T. Europhys. Lett. 1993, 21, 759–766. (46) Hellsing, M. S.; Rennie, A. R.; Porcar, L.; Englund, C.-J. Prog. Colloid Polym. Sci. 2011, 138; DOI: 10.1007/978-3-642-19038-4_24. (47) Brumberger, H. NATO ASI Series C: Mathematical and Physical Sciences; Kluwer Academic: Norwell, MA, 1995; Vol. 451. (48) Boissiere, C.; Brubach, J. B.; Mermet, A.; de Marzi, G.; Bourgaux, C.; Prouzet, E.; Roy, P. J. Phys. Chem. B 2002, 106, 1032–1035. (49) Yoneda, Y. Phys. Rev. 1963, 131, 2010–2013.
4678
dx.doi.org/10.1021/la1048985 |Langmuir 2011, 27, 4669–4678
