Neutron Reflection from the Liquid−Liquid Interface: Adsorption of

11704

Langmuir 2005, 21, 11704-11709

Neutron Reflection from the Liquid-Liquid Interface: Adsorption of Hexadecylphosphorylcholine to the Hexadecane-Aqueous Solution Interface Ali Zarbakhsh* and Ara´nzazu Querol Centre for Materials Research (Chemistry Department), Queen Mary, University of London, Mile End Road, London, E1 4NS U.K.

James Bowers Department of Chemistry, University of Exeter, Stoker Road, Exeter, EX4 4QD U.K.

M. Yaseen and Jian R. Lu Biological Physics Group, School of Physics and Astronomy, The University of Manchester, Sackville Street Building, Sackville Street, Manchester M60 1QD, U.K.

John R. P. Webster CCLRC, ISIS Facility, Rutherford Appleton Laboratory, Chilton, Didcot, OX11 0QX U.K. Received July 6, 2005. In Final Form: September 23, 2005 Adsorption of water-soluble, zwitterionic n-hexadecylphosphorylcholine (C16PC) amphiphiles has been examined at the hexadecane-aqueous solution interface using neutron reflectivity (NR) and interfacial tension measurements. The results of both methods indicate that the limiting area per surfactant molecule at the interface at the critical micelle concentration (cmc) is 40 ( 5 Å2. In the NR measurements, two isotopic contrasts have been employed to determine the adsorption isotherm and to explore the structure of the interfacial region. Single-layer model fitting to both isotopic contrasts was only possible for the single sub-cmc concentration studied, where a film thickness of 60 ( 5 Å was obtained; consistent single-layer model fits to both contrasts for concentrations greater than the cmc were not possible, leading to the requirement of a two-layer model with an overall film thickness close to 60 ( 2 Å. This film thickness is appreciably greater than the fully extended C16PC molecular length and cannot be explained purely in terms of thermal broadening. A further result is that the reflectivity data indicate that, as the C16PC concentration increases, the amount of water on the hexadecane side of the interfacial region increases, in contrast to intuitive expectation. These findings are interpreted by conjecturing a structural model in which a trilayer of C16PC molecules is formed at the interface with the water concentrated in the region occupied by the headgroups.

1. Introduction The study of the conformation and organization of biological molecules at interfaces is an important area of biophysical chemistry. Phospholipid monolayers are widely used model systems for the exploration of the properties and behavior of biomembranes. Because natural phospholipids often possess two long carboxyl chains and a large polar head,1 they are sparingly soluble in water and oil but have a considerable preference for the oil-water interface. Thus, structural studies of phospholipids adsorbed at the oil-water interface are an important step in understanding their functions.2,3 Due to many practical difficulties, however, few existing techniques are capable of revealing reliable structural information directly from the oil-water interface. Consequently, many studies of interfacial behavior of phospholipids have been made at * To whom correspondence should be addressed. Email: [email protected]. (1) Cantor, C.; Schimmel, P. R. Biophysical Chemistry; W. H. Freeman and Company: San Francisco, 1979. (2) Mo¨hwald, H. Annu. Rev. Phys. Chem. 1990, 41, 441. (3) Naumann, C.; Dietrich, C.; Lu, J. R.; Thomas, R. K.; Rennie, A. R.; Penfold, J.; Bayerl, T. M. Langmuir 1994, 10, 1919.

the air-water interface as in the case of water-soluble surfactants.4 Neutron reflectometry has become an increasingly important method for structure determination,5-7 and its application has led to an enhanced understanding of the adsorption and structure of phospholipids and related systems at the air-water interface.11,8,9 Neutron reflectivity (NR) studies of zwitterionic surfactants at the airwater interface with deuterated and protonated species have revealed that the hydrocarbon part of the adsorbed film forms a relatively thin layer with the alkyl chain (4) Lu, J. R.; Thomas, R. K.; Penfold, J. Adv. Colloid Interface Sci. 2000, 84, 143. (5) Penfold, J.; Thomas, R. K. J. Phys. Condens. Matter 1990, 2, 1369. (6) Lu, J. R.; Thomas, R. K. J. Chem. Soc., Faraday Trans. 1998, 94, 995. (7) Thomas, R. K.; Penfold, J. Curr. Opin. Colloid Interface Sci. 1995, 1, 23. (8) Yaseen, M.; Wang, Y.; Su, T. J.; Lu, J. R. J. Colloid Interface Sci. 2005, 288, 361. (9) Yaseen, M.; Lu, J. R.; Webster, J. R. P.; Penfold, J. Biophys. Chem. 2005, in press. (10) Lu, J. R.; Li, Z. X.; Thomas, R. K.; Penfold, J. J. Chem. Soc., Faraday Trans. 1996, 92, 403. (11) Hines, J. D.; Garrett, P. R.; Rennie, A. R.; Thomas, R. K.; Penfold, J. J. Phys. Chem. B 1997, 101, 7121.

10.1021/la0518086 CCC: $30.25 © 2005 American Chemical Society Published on Web 11/10/2005

Neutron Reflection from the Liquid-Liquid Interface

being significantly tilted relative to the surface normal.10,11 In recent years, advances have been made in the application of the X-ray reflectivity to the study of surfactants at liquid-liquid interfaces.12,13 Taking advantage of the contrast variation provided by neutron scattering, here we apply our recently developed methodology for the examination of the structure of liquid-liquid interface using neutron reflection14-16 to determine the adsorption isotherm and probe the structure of the adsorbed film for a model phospholipid system. The structure adopted by water-soluble hexadecylphosphorylcholine (C16PC) amphiphilessC16H33-PO4-C2H4-N(CH3)3sat the hexadecane-aqueous solution interface has been investigated. The surfactants studied are zwitterionic and are available with the hexadecyl chain protonated or deuterated, yielding surfactants hC16hPC and dC16hPC, respectively. The principal objectives were to measure the adsorbed amount and to examine how the related interfacial structure varies with surfactant concentration. A parallel interfacial tension study of this system was conducted from which the limiting adsorbed amount at the critical micelle concentration (cmc) is determined. The results presented represent the first adsorption isotherm measured using NR at a liquid-liquid interface and highlights the structure-determination capability by demonstrating the resolution of unexpected features at the interface that differ significantly with the behavior at the air-solution interface.8,9 2. Experimental Section Materials. The synthesis of the fully hydrogenated hC16hPC and the chain-deuterated dC16hPC was carried out using the two-step procedures as shown previously for the shorter-chain homologues hC12hPC and dC12hPC.8 The hydrogenated or deuterated n-hexadecanol was reacted with 2-chloro-2-oxo-1,3,2dioxaphospholane, to afford the product 2-hexadecyl-2-oxo-1,3,2dioxaphospholane. The intermediate product was then purified through a silica flash column (40-60 mesh, Fluka) with a 1:1 mixture of petroleum ether and ethyl acetate, followed by the evaporation of the solvent to obtain the intermediate as light yellow oil. The intermediate product was transferred into a sealed, thick-walled pressure tube and connected to a vacuum line to remove any residual moisture. Trimethylamine was subsequently introduced, and the reaction tube was then placed in an oil bath at 75 °C for 24 h to afford the n-alkyl phosphocholine product. Freshly obtained 2-chloro-2-oxo-1,3,2-dioxaphospholane was used in ca. 0.25 molar excess to the alcohol in the synthesis so that the yield with respect to the deuterated alcohol was improved. The trimethylamine was also used in excess, twice the molar ratio of the intermediate 2-alkyl-2-oxo-1,3,2-dioxaphospholane. The final product was purified by flash silica column chromatography using initially chloroform then distilled methanol. The product was obtained as a white powder with a yield of 75-80%. D2O was obtained from Fluorochem (>99 at. D%), and ultrapure H2O was produced using an Elgastat water purification unit. Hexadecane-h34 and hexadecane-d34 were purchased from Aldrich (99%) and Cambridge Isotope Laboratories (>98 at. D%), respectively. Mixtures of protonated and deuterated solvents were prepared as appropriate for the required neutron refractive index. The aqueous solutions were prepared by appropriate dilution of stock solutions. All solutions and isotopic mixtures were prepared by mass. Interfacial tension, γ, measurements were conducted on a Kru¨ss K9 tensiometer using the ring method (12) Pingali, S. V.; Takiue, T.; Luo, G. M.; Tikhonov, A. M.; Ikeda, N.; Aratono, M.; Schlossman, M. L. J. Phys. Chem. Ser. B 2005, 109, 1210. (13) Schlossman, M. L. Physica B 2005, 357, 98. (14) Zarbakhsh, A.; Querol, A.; Bowers, J.; Webster, J. R. P. Faraday Discuss. 2005, 129, 155. (15) Bowers, J.; Zarbakhsh, A.; Webster, J R P.; Hutchings, L. R.; Richards, R. W. Langmuir 2001, 17, 140. (16) Zarbakhsh, A.; Bowers, J.; Webster, J. R. P. Meas. Sci. Technol. 1999, 10, 738.

Langmuir, Vol. 21, No. 25, 2005 11705 with appropriate density correction. For the interfacial tension measurements, the hexadecane was purified according to the method of Lunkenheimer and Goebel by passing the hexadecane as received through a column packed with alumina seven times,17 and ultrapure H2O was used. Owing to the cost of the hexadecaned34, this procedure was not followed for the oil used for the reflectivity measurements. Neutron Reflection. Neutron reflectometry is a technique sensitive to the average neutron refractive index profile normal to an interface, n(z).10 The dispersive refractive index can be written as

n(λ) ≈ 1 -

λ2 λ Nb + i Nσ 2π 4π

(1)

where λ is the neutron wavelength, Nb ) ∑iNibi and Nσ ) ∑iNiσi with Ni the number density, bi the coherent scattering length, and σi the absorption and incoherent cross-section of nucleus i. The multiple Nb is the scattering length density, which is approximately linearly related to the volume fraction composition, viz. Nb ≈ ∑jφj(Nb)j, where φj is the volume fraction and (Nb)j the scattering length density of component j, respectively. According to eq 1, the large difference in the scattering lengths of 1H (b ) -3.7406 fm) and 2H (b ) 6.671 fm) can be exploited in hydrogenous systems to aid determination of adsorbed amounts and reduce ambiguity in structure determination by tuning the refractive index by isotopic substitution. Here two different isotopic contrasts have been employed. In the first (contrast A), the hexadecane and the aqueous solution are contrast-matched to silicon, i.e., Nb ) 2.07 × 10-6 Å-2, and the surfactant used is dC16hPC, and this contrast enables the determination of the adsorbed amount and the estimate of the film thickness. The second contrast (contrast B) uses hC16hPC with Nboil ) 4.00 × 10-6 Å-2 and Nbsolution ) 6.35 × 10-6 Å-2. Consistency is then sought in the modeling of the reflectivity data from both contrasts. Information regarding the interfacial structure is obtained once this consistency is obtained and the reflectivities described using a common model. In the experiments reported here, a thin hexadecane film is created by spin-coating hexadecane on to a silicon block which has been rendered hydrophobic by coupling of chlorotrimethylsilane. The spun hexadecane film is frozen in place, and the sample cell assembled. With the oil still frozen, the aqueous surfactant solution is admitted until the hexadecane phase is trapped between the silicon substrate and the aqueous subphase. Once the sample chamber is ensured to be bubble-free, the hexadecane film is allowed to melt. The sample cell was thermostated at 298 ( 1 K. Full details of the sample environment and film preparation are given elsewhere.6-8 For a sufficiently thick oil film, the reflectivity, R, is given by

R ) R1 +

AR2(1 - R1)2 1 - AR1R2

(2)

R1, the reflectivity from the silicon-hexadecane interface, and R2, the reflectivity from the hexadecane-aqueous solution interface, are calculated using the optical-matrix method.18 The attenuation factor, A, for the loss of intensity upon the beam crossing the oil film twice is

(

A ) exp

-2χdoil sin θoil

)

(3)

where doil is the thickness of the oil film and θoil is the incidence angle of the neutron beam in the oil phase. A comprehensive description of the data analysis is given elsewhere.14 Reflectivity measurements were conducted on the reflectometer SURF at the ISIS Spallation Neutron Source, Rutherford Appleton Laboratory, Didcot, UK. A polychromatic neutron beam with wavelengths 0.53 < λ < 6.9 Å is used, and by employing (17) Goebel, A.; Lunkenheimer, K. Langmuir 1997, 13, 369. (18) Born, M.; Wolf, E. Principles of Optics: Electromagnetic Theory of Propagation, Interference and Diffraction of Light, 6th ed.; Pergamon Press: Oxford, 1980.

11706

Langmuir, Vol. 21, No. 25, 2005

Zarbakhsh et al.

cmc is located at 1.2 × 10-5 mol dm-3. The LangmuirSzyskowski (eq 4a) and the Gibbs adsorption (eq 4b) isotherms have been employed to determine the average area per molecule at the cmc.

γ ) γ* Γ3(2,1) )

Figure 1. Interfacial tension measured at the hexadecanesolution interface at 298 K plotted as a function of hC16hPC concentration. The dashed (solid) line corresponds to the area per molecule determined using eq 4a (eq 4b). nominal grazing incidence angles of 0.37, 0.62, and 1.30°, a utilizable momentum transfer range 0.011 < Q < 0.537 Å-1, where Q ) (4π sin θ)/λ, is obtained. A horizontal sample geometry is maintained throughout, and the incidence angle is varied by use of a supermirror. The actual angles of incidence are determined by performing detector angle scans in reflection geometry once the sample had been aligned using a series of height and detector angle scans. The collimating slit settings are varied with incidence angle in order to measure all reflectivities with near-constant angular resolution, δθ/θ ≈ 8%, and to project an illuminated length of ∼45 mm on to the sample to ensure under-illumination. The measured reflectivity profiles are normalized relative to the incidence beam monitor spectrum and corrected for detector efficiency as per standard reflectivity measurement on a time-of-flight instrument,19 and the data are subsequently corrected for the wavelength-dependent transmission through the silicon substrate. Normalization by the monitor spectrum also accounts for the supermirror efficiency. The data for each incidence angle are truncated at an appropriate wavelength cutoff, which is determined by the angle of incidence of the neutron beam at the supermirror. The reflectivity data are placed on an absolute scale by obtaining scale factors for each measured angle by calibration against the silicon-D2O reflectivity, which in turn is normalized using total reflection. In the presented data, only the information-rich reflectivity data at 1.3° are shown. However, model fitting has been performed for all angles of incidence. For our present purpose, it is convenient to replace σ in eq 1 with the linear absorption coefficient χ ) χ(λ), which embraces all loss processes and is determined from transmission measurements. Transmission measurements through samples of hexadecane with the different H/D composition were measured in 2 mm Hellma spectrophotometric cuvettes, and these measurements allowed χ to be determined for the oil phase using the Beer-Lambert law T ) exp(-χl), where l is the path length. The oil film thickness is controlled by using identical spinning conditions for all film preparations on the spin-coater. The thickness is calibrated from the double-critical edge measured using contrast B and employment of eqs 2 and 3. R1 is calculated using additional measurements, and the parameters for the silicon oxide layer (d ) 8Å, Nb ) 3.4 × 10-6 Å-2) and the trimethylsilane coupled layer (d ) 5Å, Nb ) 0.5 × 10-6 Å-2) with roughnesses of 2 Å are then fixed in the subsequent model calculations. A flat background has been assumed.

3. Results and Discussion Interfacial Tension Measurements. The interfacial tension isotherm is shown in Figure 1 and has been analyzed using two standard adsorption isotherms to yield an estimate for the area per molecule below the cmc. The (19) Penfold, J. Physica B 1991, 173, 1.

kT c ln 1 + a0 b

(4a)

-1 ∂γ RT ∂ ln c

(4b)

(

(

)

)

T

Here c is the surfactant(3) concentration, γ* the interfacial tension of the bare hexadecane(2)-water(1) interface, and b a parameter that reflects the strength of adsorption and defines the position at which the adsorbed amount is 50% of its maximum value. The resulting fit lines according to eq 4a are shown in Figure 1. According to eq 4a, the limiting area per molecule at the cmc is a0 ) 40.5 ( 3.5 Å2 (obtained using a two-parameter fit with b ) (2.6 ( 0.7) × 10-7 mol dm-3) which is consistent with a0 ) 41.0 ( 3.0 Å2 obtained using eq 4b. Use of eq 4b assumes that the Gibbs dividing surfaces for oil and water distributions coincide at a common location, z ) 0. Neutron Reflection Measurements. The main results of the neutron reflection experiments are summarized in Figures 2-6. The reflectivity data recorded for contrasts A and B are shown in Figures 2 (for five concentrations of dC16hPC) and 5 (for four concentrations of hC16hPC), respectively. For contrast A, it is possible to analyze the data using a single-layer model for all the concentrations studied in addition to a Guinier analysis; the results of both analyses are given in Figure 3. The employed Guinier analysis has been used to determine the adsorbed amount essentially independently from the choice of model and to provide an estimate of the film thickness. In the Guinier analysis of the contrast A data, the reasonable simplifying assumption that the reflectivity from the silicon-hexadecane interface is negligible has been made. The appropriate approximation is then

ln(Q2R) ≈ ln(16π2Γn2) - σ2Q2

(5)

+∞ where Γn ) ∫-∞ Ff(z) dz with Ff(z) the film’s scattering length density and σ2 ) 〈z2〉 - 〈z〉2 is the second moment of the layer about its center of mass, where the nth moment +∞ is defined 〈zn〉 ) Γn-1∫-∞ F(z)zn dz. Thus, for a slab with constant scattering, length density and film thickness, d, and the area per molecule, Apm, are thus determined

d ) x12σ Apm )

∑ibi Γn

(6a) (6b)

The numerator in eq 6b is the molecular scattering length: assuming that the C16D33 chain is 98% perdeuterated, the molecular scattering length for dC16hPC is ∑ibi) 336.1 fm. The results from the Guinier analysis determined using eqs 6a and 6b are given in Figure 3a-c for the five concentrations of dC16hPC studied. The results from a more-detailed analysis using single-layer models, given in Figure 3, show that the main findings remain essentially unchanged by explicit inclusion of all layers, including those at the Si-hexadecane interface. The single-layer model fits to the reflectivity data are shown as solid lines in Figure 2a-e. As one expects, there is a degree of coupling between the thickness and scattering length density of the modeled film. The contour plot shown

Neutron Reflection from the Liquid-Liquid Interface

Langmuir, Vol. 21, No. 25, 2005 11707

Figure 2. Reflectivity data for contrast A (Si-CMSi-hexadecane-CMSi solution) interface for a solution of dC16hPC measured at 1.3°. The reflectivity profiles are shown as a function of the scattering wave vector, Qz. The solid lines in (a-e) correspond to a single layer model with film thickness 60 Å without any film roughness considered. The lines in (f-j) are determined using a two-layer model (22 and 42 Å) with roughness of 4 Å, consistent with the model employed in Figure 5. (1) is the reflectivity data from a bare interface for this contrast.

Figure 3. Analysis of reflectivity data forcontrast A measured at 1.3°: (a) area per molecule, (b) adsorbed amount, and (c) film thickness as a function of dC16hPC concentration. (O) Results of the Guinier analysis using eqs 6a and 6b and (b) results from optical matrix fitting using a single-layer model without roughness. The vertical dotted line indicates the cmc. The horizontal lines indicate the range of acceptable area per molecule determined from eqs 4a and 4b.

in Figure 4 for the 6 × 10-5 mol dm-3 solution reveals that good quality fits to the data are obtained for film thickness 60 ( 2 Å and Nb ) 3.41 ( 0.04 × 10-6 Å-2. The uncertainties in thickness and Nb have been similarly determined for all surfactant concentrations studied. The resulting adsorbed amounts, areas per molecule, and film thickness from this single-layer analysis of the contrast A data are compared with the results of the Guinier analysis in panels a, b, and c of Figure 3, respectively, as a function of concentration. Above the cmc, slightly different results are found from the two methods of analysis. Such a difference could arise if the single-layer model is an oversimplification of the actual scattering length density profile. As we shall see shortly, analysis of the contrast A and B data together strongly indicates that this is the case. Nevertheless, the first main result of these experiments is that the film thickness exceeds the fully extended molecular length (∼35 Å). The contribution to the overall interfacial width, σCW ∼ 10 Å, from thermal capillary waves is insufficient to account for the determined film thickness, and thus, there must be a structural explanation for the broad interface. The area

per molecule above the cmc is found to be 42 ( 2 Å2, which is in good agreement with the interfacial-tension-derived value. We note that for the same surfactant adsorbed at the air-solution interface8,9 the limiting area per molecule at the cmc is 52 ( 3 Å2 both from surface tension and NR measurements and film thickness of 22 Å, suggesting significant tilting of the molecules in the adsorbed monolayer, consistent with the results for previous zwitterionic surfactant systems.10,11 Structure Determination. Figure 5 shows reflectivities measured for contrast B at concentrations corresponding to four of those measured using contrast A. From analysis of the contrast B data, it emerges that except at the lowest (sub-cmc) concentration studied one-layer models do not provide an adequate description of the reflectivity data and recourse to a two-layer model is required. The model can be transferred from one contrast to the other; the relevant model fit lines given in Figures 2 and 5 pertain to a single model for a given concentration. Despite the transferability of the model, the two-layer model that we present is not unique. However, the use of the two contrasts means that the model captures the

11708

Langmuir, Vol. 21, No. 25, 2005

Zarbakhsh et al.

Figure 4. Weighted sum-of-squares deviations as a function of film thickness and scattering length density for the contrast A system at the concentration of 6 × 10-5 mol dm-3 displayed as a contour plot. Similar plots are obtained for all other concentrations.

Figure 5. Reflectivity data for contrast B (Si-CM4-hexadecane-D2O solution) interface for a solution of hC16hPC measured at 1.3°. The reflectivity profiles are shown as a function of the scattering wave vector, Qz. The solid lines in (a-d) correspond to a two-layer model (22 and 42 Å) with a roughness of 4 Å at each interface, consistent with the model employed in Figure 2. (1) is the reflectivity data from a bare interface for this contrast.

essential features necessary to consistently describe the reflectivity data for both contrasts A and B as a function of C16PC concentration. The scattering length density profiles corresponding to the appropriate fit lines in Figures 2 and 5 are given in Figure 6a and b using a common z scale. In the presented model, the interface is divided into two slabs, which each represent inhomogeneities on the hexadecane and solution sides of the interface. The location of the division between the two slabs was chosen after significant preliminary modeling. The layer on the hexadecane side of the interface was assigned a thickness of 22 Å, which was chosen to correspond roughly the C16 alkyl chain length. The thickness of the layer on the solution side then emerges as 42 Å, giving an overall film thickness that is in reasonably good agreement with the average film thickness determined from the earlier analysis. It should be

Figure 6. (a) Representative scattering length density profiles corresponding to the production of the solid line fits in Figure 2f, g, h, and j when modeling the interfacial adsorption of dC16hPC at the hexadecane-solution interface. (b) Representative scattering length density profiles corresponding to the production of the solid line fits in Figure 5a-d when modeling hC16hPC adsorption at the hexadecane-solution interface. The arrows indicate the trend of increasing solution concentration (0.36, 0.6, 2.4, and 6.0 × 10-5 mol dm-3).

stressed here that this division is arbitrary since many different proportions can be modeled; however, the proportions are maintained for all concentrations studied to demonstrate the main manifestations that emerge regardless of precise structural details. Figure 6b reveals the second major result of these measurements: the scattering length density on the hexadecane side of the interface increases with increasing concentration. This result is surprising since it indicates that the amount of D2O in this region must be increasing, which is in conflict

Neutron Reflection from the Liquid-Liquid Interface

Figure 7. Schematic representation of the C16PC trilayer in relation to the hexadecane and aqueous phases. Note, however, that dynamic fluctuations may mean the real interface is very different to the proposed structure illustrated.

with the expectation for the oil side of the interface. On the solution side of the interface, the scattering length density decreases as the solution concentration increases. This finding is consistent with the inclusion of surfactant or oil in this region. The total region of interfacial inhomogeneity extends ∼70 Å, which is in agreement with the film thickness derived above the cmc using the Guinier analysis. Adsorbed Film Structure. The main findings to discuss are the origin of the observed film thickness and apparent inclusion of water in the hexadecane side and C16PC or hexadecane on the aqueous side of the interfacial region. Because C16PC adsorption is pH insensitive, we do not consider pH sensitivity in our conjectured model. The overall thickness of the adsorption region is ∼60-70 Å which, as mentioned earlier, is inconsistent with the estimated molecular length of ∼35 Å. One plausible scenario that is consistent with the observations is a trilayer arrangement of the C16PC at the interface; this is illustrated schematically in Figure 7. Such a structure accommodates the alkyl chains of the first C16PC layer embedded in the hexadecane phase and hydrophilic headgroups of the third C16PC layer immersed in the aqueous phase, as is desirable. The second C16PC layer intervenes, lying headgroup to headgroup with the first layer and with its alkyl chains interdigitated with those of the third layer. Such multilayering would be driven by the charge separation in the surfactant headgroup; the Coulombic interaction energy is minimized if the molecules align with unlike charges adjacent to one another as is naturally incorporated in the trilayer model. Note that the representation given in Figure 7 is of course an oversimplified interpretation of the real structure. Dynamic fluctuations may mean the real interface is very different such that a lot of oil and water is intermixed across the interface. The intermixing can be inferred from Figure 6; the real structure must take into account water in the oil region and vice versa. Analogous behavior might be expected at the airsolution interface. However, at the air-solution interface, the molecules are significantly tilted, and this tilting

Langmuir, Vol. 21, No. 25, 2005 11709

enables close contact between alkyl chains and accommodates charge-ordering within the headgroup region. At the hexadecane-solution interface, the presence of the hexadecane molecules obviously has a profound influence on the nature of the structure. Inclusion of the hexadecane into the alkyl chains in the first C16PC layer permits this layer to form with possibly only a slight tilt relative to the interface normal. The proposed model is consistent with the determined film thickness. It also accommodates the observed variation in layer scattering length densities as a function of concentration. The headgroup region between the first two layers of C16PC is likely to be considerably hydrated. Therefore, as the concentration of surfactant in these layers increases, the amount of water will increase too. This thus offers a reasonable explanation for the observed increase in water content on the hexadecane side of the interface and for the increasing surfactant concentration on the water side of the interface with increasing C16PC concentration. The two-layer model we have used to represent the reflectivity data does not account for an explicit headgroup region, but nonetheless, the model used coveys the salient features. The final issue to address regards whether the areas per molecule obtained from the NR and interfacial tension measurements are consistent with the conjectured trilayer organization at the interface. If we treat the first C16PC layer of the trilayer as being approximately 50 vol% surfactant alkyl chains and 50 vol% hexadecane and the second/third C16PC layers possessing the same alkyl chain coverage as a saturated monolayer, then one expects the areas per molecule at the oil-solution and air-solution interfaces to differ by a factor of ∼1.5. According to the model used here, this difference is a factor of ∼1.3, but the magnitude is certainly consistent with the conjectured model. 4. Summary An adsorption isotherm for surfactant adsorption at an oil-water interface using neutron reflectometry has been measured for the first time. Further to this, by employment of isotopic substitution, a high level of structural detail is determined for the specific case of C16PC adsorption at the hexadecane-aqueous solution interface. These experiments demonstrate a significant development in this area of research and opens up the technique to routine investigations. The neutron reflection and interfacial tension measurements indicate that the limiting area per C16PC molecule at the interface at the cmc is 40 ( 4 Å2. Above the cmc, the large film thickness (∼60-70 Å) and unexpectedly high water content on the hexadecane side of the interface lead to the proposal of a structural model in which a trilayer of C16PC molecules is formed at the interface. This conjectured model is consistent with the reflectivity data recorded using two isotopic contrasts. Further experiments, especially in the sub-cmc region, are proposed to investigate apparent change in film structure occurring in the cmc region. Acknowledgment. The authors wish to thank the CCLRC for allocation of beamtime at ISIS and for provision of consumables and subsistence. LA0518086