Structural Characterization of Surfactant Aggregates Adsorbed in

Apr 8, 2011 - Equation 6 with B = 1 (where I0 is the limiting value of Idiff at q = 0) was ..... Figure 8 presents structural parameters derived from ...
0 downloads 0 Views 4MB Size
ARTICLE pubs.acs.org/Langmuir

Structural Characterization of Surfactant Aggregates Adsorbed in Cylindrical Silica Nanopores Tae Gyu Shin,†,^ Dirk M€uter,‡,|| Jens Meissner,† Oskar Paris,§ and Gerhard H. Findenegg*,† †

Institute of Chemistry, Stranski Laboratory, Technical University Berlin, Strasse des 17. Juni 124, 10623 Berlin, Germany Department of Biomaterials, Max-Planck-Institute of Colloids and Interfaces, Research Campus Golm, 14424 Potsdam, Germany § Institute of Physics, Montanuniversitaet Leoben, Franz-Josef-Strasse 18, 8700 Leoben, Austria ‡

ABSTRACT: The self-assembly of nonionic surfactants in the cylindrical pores of SBA-15 silica with a pore diameter of 8 nm was studied by small-angle neutron scattering (SANS) at different solvent contrasts. The alkyl ethoxylate surfactants C10E5 and C12E5 exhibit strong aggregative adsorption in the pores as indicated by the sigmoidal shape of the adsorption isotherms. The SANS intensity profiles can be represented by a sum of two terms, one accounting for diffuse scattering from surfactant aggregates in the pores and the other for Bragg scattering from the pore lattice of the silica matrix. The Bragg reflections are analyzed with a form factor model in which the radial density profile of the surfactant in the pore is approximated by a two-step function. Diffuse scattering is represented by a TeubnerStrey-type scattering function which indicates a preferred distance between adsorbed surface aggregates in the pores. Our results suggest that adsorption starts with formation of discrete surface aggregates which increase in number and eventually merge to interconnected patches as the plateau value of the adsorption isotherm is approached. A grossly different behavior, viz. formation of micelles as in solution, is found for the maltoside surfactant C10G2, in agreement with the observed weak adsorption of this surfactant in SBA-15.

1. INTRODUCTION The self-assembly of surfactants in confined geometries plays an important role in environmental, chemical, and pharmaceutical technology. Confinement-induced structural changes of surfactant layers adsorbed at the surface of colloidal particles are modifying the forces between the particles and in this way affect the colloidal stability.1 Surfactant adsorption in narrow pores is of importance in surfactant-aided membrane processes2 such as micellar-enhanced ultrafiltration3 or micellar liquid chromatography.4 The ability to control the morphology of surfactant assemblies in nanoscale confinement can also be relevant for applications in nanofluidic devices. For instance, it has been proposed that surfactants can be used to physically open and close nanometer-sized channels and thus tune the permeability of molecular gates.5 Adsorption of nonionic surfactants onto hydrophilic surfaces represents a surface aggregation process reminiscent of micelle formation in bulk solution but occurring below the critical micelle concentration (cmc).6,7 The adsorption of alkyl poly(ethylene oxide) surfactants (CnEm) onto hydrophilic silica was investigated in detail by thermodynamic810 and structuresensitive1115 methods and in theoretical studies.13,16,17 In this case undissociated silanol groups of the silica surface represent the principal sites for adsorption by hydrogen bonding to the ether oxygens of the polyethoxylate head groups. Since water forms strong hydrogen bonds with the silanol as well as the surfactant headgroup, the net free energy of displacement of r 2011 American Chemical Society

water by the ethoxy groups is rather low. Thus, hydrophobic interactions between the surfactant tails become dominant and give rise to aggregative adsorption once a certain surface concentration of surfactant molecules is surpassed. From studies with atomic force microscopy (AFM) on flat surfaces11,12 it was concluded that surface aggregates of CnEm surfactants formed on hydrophilic silica have similar structures as micelles in solution. Globular surface micelles were found for C10E5 and C10E6 but a flat continuous bilayer for C12E5, in agreement with the trends predicted by the packing parameter model. A recent study with small-angle neutron scattering (SANS) indicated, however, that C12E5 is forming discrete surface micelles on silica nanoparticles.18 The difference in the aggregate structure formed on flat silica surfaces and on silica nanoparticles was attributed to the high surface curvature of the nanoparticles, which prevents an effective packing of the surfactant molecules in a bilayer film. Confinement effects due to the proximity of two adsorbing surfaces become significant when the distance between the surfaces approaches the characteristic size of the aggregates. Confinement in one dimension of space (1d) corresponds to the geometry of a liquid between parallel surfaces. Experimental studies of this situation were based mostly on measurements of the normal and lateral forces acting between two approaching Received: January 25, 2011 Revised: March 4, 2011 Published: April 08, 2011 5252

dx.doi.org/10.1021/la200333q | Langmuir 2011, 27, 5252–5263

Langmuir surfaces in the respective liquid medium.19,20 Recently, X-ray reflectivity21 and holographic X-ray diffraction22 have been adopted to reveal confinement-induced ordering in complex liquids between planar parallel surfaces. Confinement in two dimensions (2d) will occur in cylindrical channels or pores. Huinink et al.23 treated the adsorption of nonionic surfactants in hydrophilic cylindrical nanopores as a phase transition between a dilute phase and a bilayer phase and derived a Kelvin-like relation for the influence of the pore curvature on the adsorption behavior. According to this relation the curvature-induced shift of the phase transition relative to a flat surface is proportional to the curvature energy of the surfactant bilayer. Only a few experimental studies have so far addressed the effect of confinement on the adsorption of surfactants in nanopores.10,24,25 Some conclusions about pore size and curvature effects could be drawn, but the ill-defined pore geometries of the materials used in these studies prevented a deeper understanding. In recent years ordered mesoporous silica materials like MCM-41 and SBA-15 became available for studies of confinement effects. SBA-15 is made up of micrometer-sized particles which constitute arrays of cylindrical nanopores arranged side by side in a 2d hexagonal lattice (space group P6mm).26 Due to the high mesoscale order of these materials it has become possible to characterize adsorbed fluid layers in the pores by small-angle scattering of X-rays27,28 and neutrons29,30 (SAXS and SANS). In a preceding paper31 the self-assembly of the surfactant C12E5 in the cylindrical pores of SBA-15 silica was studied by SANS. The structural model adopted in the analysis of the scattering data accounts for the formation of isolated surface micelles at low loadings and the gradual transformation to a patchy bilayer as the loading is increased. Following up this approach, we now present comprehensive scattering data at different scattering contrasts for nonionic surfactants with different head groups and different tail lengths. It is shown that the aggregate structures of the surfactant in the silica matrix are strongly dependent on the type of its headgroup.

2. EXPERIMENTAL SECTION 2.1. Materials. SBA-15 was the same material as in the previous study.31 XRD images exhibit five well-resolved Bragg reflections from which we derive the lattice parameter a0 = 10.2 nm and pore diameter D = 8.1 nm using the procedure described in ref 27. The pore size agrees with the value DKJS = 8.14 nm determined by nitrogen adsorption with the KrukJaroniecSayari (KJS) prescription.32 The BET specific surface area of the sample was as = 840 m2/g, and the ratio of the outer to inner surface area was estimated as 1: 10, based on the typical size of the SBA-15 particles (diameter 0.2 μm, length 2 μm, determined by scanning electron microscopy). The surfactants penta(ethylene glycol)monodecylether, C10E5 (Bachem, purity g 97%), penta(ethylene glycol)monododecylether, C12E5 (Fluka, purity g 98%), decyl-β-D-maltoside, β-C10G2 (Glycon, purity > 99.5%), and D2O (Euriso-top, 99.9% isotope purity) were used without further purification. Reagent-grade water (H2O) was obtained from a Milli-Q 50 pure-water system (Millipore, Billerica, USA). 2.2. Adsorption Measurements. Adsorption isotherms of the surfactants in SBA-15 were determined in H2O solutions from the difference in surfactant concentration before and after equilibration with a known mass of dry silica. Equilibrium concentrations c in the supernatant were obtained from the surface tension σ, based on a calibration curve σ = σ(c). Adsorption data for equilibrium concentrations greater than the cmc, where the surface tension reaches a constant value, were obtained by adding known aliquots of water until σ was greater than

ARTICLE

σ(cmc). Surface tension measurements were made by the Du No€uy ring method using a Kr€uss K11 tensiometer. 2.3. Sample Preparation. For the SANS measurements, samples of SBA-15 containing known amounts of surfactant were prepared in pure D2O and in a H2O/D2O mixture of scattering length density equal to that of SBA-15. Appropriate amounts of surfactant were dissolved in a given volume of water (V = 10 mL) and equilibrated with a known mass of SBA-15 powder (ms ≈ 0.1 g) by sonication (2 min) and rotation of the glass container in a water bath (2 h). After separation from the supernatant, the silica slurry was transferred into specially designed SANS sample cells made from aluminum with quartz glass windows (sample width 1 mm, sample volume Vcell = 0.1 cm3).33 Typically, the sample cells contained 30 mg of SBA-15 with an associated mesopore volume of ca. 30 μL (pore liquid) and 55 μL of extra-particle liquid (bulk solution). For the surfactants C10E5 and C12E5, for which the specific adsorbed amount na reaches a high plateau value nam above the cmc, the amount of surfactant needed to reach a relative adsorption f = na/nam was estimated as n = f 3 nam 3 ms þ cmc 3 V, where the second term represents the amount remaining in the supernatant. In the samples with C10E5 and C12E5 more than 99% of the surfactant was adsorbed and less than 1% remained in the aqueous phase (in monomeric form). For the weakly adsorbing surfactant C10G2 the amount of surfactant was chosen as n = p 3 cmc 3 V, where p is a numerical factor (10 < p < 30). In this case a large part of the surfactant was not adsorbed. 2.4. SANS Measurements. SANS measurements were performed at the Institut Laue Langevin (ILL Grenoble) using the small-momentum transfer diffractometer D16. We used two detector angles (0 and 12) at a wavelength λ = 0.473 nm, covering a range of wave vector transfer of 0.4 < q < 4.5 nm1, where q = (4π/λ)sin θ, with 2θ being the scattering angle. Data reduction and normalization was performed in the same way as for liquid samples. The composition of a H2O/D2O mixture having the same scattering length density as SBA-15 silica was determined by scattering measurements with SBA-15 in a series of H2O/D2O mixtures of different D2O mass fraction μD. From the intensity of the (10) Bragg reflection I(q10) as a function of μD the contrast-match point was found at μD = 0.638, corresponding to F = 3.7  106 Å2. This H2O/D2O mixture was used in all measurements in the contrast-match scenario.

3. THEORETICAL MODEL AND DATA ANALYSIS To account for the SANS profiles in a quantitative way, the total scattering intensity I(q) is taken as a sum of three independent contributions: Bragg scattering from the pore lattice (IBragg), diffuse scattering from surfactant aggregates in the pores (Idiff), and a background (Ib) due to incoherent scattering from protons of H2O and surfactant, i.e. IðqÞ ¼ IBragg ðqÞ þ Idif f ðqÞ þ Ib

ð1Þ

3.1. Bragg Scattering. The spherical average of the total scattering intensity for an assembly of long cylindrical objects is given by

IBragg ðqÞ ¼ KSðqÞjFðqÞj2

ð2Þ

with the structure factor S(q) and form factor |F(q)|2, where F(q) is the scattering amplitude of the cylindrical cross section, and K is a constant. For a perfect 2d-hexagonal lattice, the spherically averaged structure factor is given by S(q) = (∑hkmhkShk(q))/q2, where mhk is the multiplicity factor of a diffraction peak with Miller indices hk (mhk= 6 for the 10, 11, and 20 peaks) and Shk represents delta functions at positions qhk = (4π/a0(3)1/2)(h2 þ 5253

dx.doi.org/10.1021/la200333q |Langmuir 2011, 27, 5252–5263

Langmuir

ARTICLE

scattering function corresponding to γ(r) of eq 5 is Idif f ðqÞ ¼ BðqÞ

I0 2 ð1  I0 =Im Þðq =q2m

 1Þ2 þ I0 =Im

ð6Þ

where Im and qm represent the coordinates of the correlation peak. Equation 6 with B = 1 (where I0 is the limiting value of Idiff at q = 0) was proposed for microemulsion systems by Teubner and Strey,35 and similar expressions have been used to describe transient structures of phase-separating systems.36,38 We adopt this model with the previously proposed36 expression B(q) = q4/ (k þ q4), where k is a constant, to attain Idiff at q = 0. Although eq 6 describes the behavior of an infinite system rather than a fluid confined to a pore, we believe that this expression captures the essence of the microphase separation of the liquid in the pores. The characteristic quantities d and ξ are related to the parameters I0, Im, and qm of eq 6 by Figure 1. Sketch of surfactant aggregate in the cylindrical pores of SBA15 and radial scattering length density profiles F(r) adopted in data analysis. The broken line sketches a probable real density profile and the boxes indicate the equivalent step profile used in the data evaluation: (a) full contrast (in D2O) and (b) contrast match of matrix and solvent (H2O/D2O).

k2 þ hk)1/2. In reality the Bragg reflections are broadened due to lattice imperfections and, in particular in the case of neutrons, because of limited instrumental resolution. They were therefore modeled by Gaussian functions (see section 3.3). The form factor of the cylindrical pores with an adsorbed surfactant layer was modeled by a two-step density profile F(r) as shown in Figure 1. Here, F0, Ff, and Fc represent the scattering length densities of silica matrix, surfactant film, and water in the core space of the pore, respectively, R0 is the pore radius, and Rf is the inner radius of the surfactant film. The scattering amplitude corresponding to this model is FðqÞ ¼

ðFf  F0 ÞR02 ZðqR0 Þ þ ðFc  Ff ÞRf2 ZðqRf Þ ðFf  F0 ÞR02 þ ðFc  Ff ÞRf2

ð3Þ

where Z(qR) = 2J1(qR)/(qR) and J1 is the Bessel function of the first kind and first order.34 Equation 3 applies to the full-contrast scenario, when Fc is the scattering length density of D2O. In the contrast-match scenario, when Fc = F0, F(q) can be written in compact form as FðqÞ ¼

2 ½R0 J1 ðqR0 Þ  Rf J1 ðqRf Þ qðR02  Rf2 Þ

ð4Þ

where the contrast factors have canceled. 3.2. Diffuse Scattering. Diffuse scattering is attributed to the inhomogeneity of the complex liquid in the cylindrical pores. The concentrated aqueous surfactant system is thought of as a disperse two-phase system with domains of characteristic size, reminiscent of a microemulsion. This suggests a damped-periodic form of the two-point correlation function γ(r)35,36 γðrÞ ¼

sinð2πr=dÞ r=ξ e 2πr=d

ð5Þ

with two characteristic length parameters: a wavelength d representing a quasiperiodic distance between domains and a decay length ξ of this quasiperiodic order. The quantity ξ /d is a measure of the polydispersity of the domain size.37 The

#1=2 #1=2 pffiffiffi" pffiffiffi" 1 1 2 2 d ¼ 2π þ1 ξ¼ 1 : qm ð1  I0 =Im Þ1=2 qm ð1  I0 =Im Þ1=2

ð7Þ 3.3. Data Analysis. In the data analysis the uniform background Ib of eq 1 was first subtracted from the individual scattering curves, with the criterion that I(q)  Ib shows Porod behavior at large q (q > 3 nm1). To separate the terms in the sum Idiff(q) þ IBragg(q) of eq 1 the following procedure was adopted: In the first step the function S(q)|F(q)|2 of eq 2 was represented by the sum * (q) divided by q2. The functions of three Gaussian functions Shk * (q) are centered at the respective positions of the Bragg Shk reflections qhk, i.e., S*hk(q) = ahk exp[b(q  qhk)2], where ahk is the amplitude of peak hk. The parameter b of the Gauss peaks was taken to be identical for the peaks hk = 10, 11, and 20 and for all surfactant loadings. A value b = 175 nm2 was obtained by fitting one single data set and then used for all data sets obtained under specified conditions of the scattering experiments. With this representation of the Bragg scattering contribution a fit of the complete set of scattering data can be performed using eq 1, with the diffuse scattering part given by eq 6. It yields best-fit values of the parameters I0, Im, qm, and k of eq 6 and the peak heights a10, a11, and a20 of the three leading Bragg peaks, which in turn are proportional to S(q)|F(q)|2, since the width was constant for all data sets. In the second step of the data analysis the parameters of the form factor model (eq 3) are determined from the intensities of the Bragg peaks. This analysis is similar to that used previously in a small-angle X-ray scattering study of fluids adsorbed in SBA-15. 27,39 Since IBragg (q) is proportional to the product S(q)|F(q)|2, the intensities of the first three Bragg reflections ~I (qhk) are directly proportional to |F(q)|2 at the respective q = qhk. Accordingly, the parameters of the form factor model can be derived from the set of integral Bragg intensities ~I (qhk) by minimizing the variance39  " # ! h i 2 X  ~I ðqhk Þ  2 2 χ ¼ ð8Þ  KjFðqÞj     mhk hk

where the sum is taken over the three leading Bragg reflections. In the expression for the form factor (eq 3) only the parameters Rf (defining the thickness of the surfactant layer) and Ff (scattering length density of the layer) are expected to vary with the surfactant loading. The other parameters were treated as 5254

dx.doi.org/10.1021/la200333q |Langmuir 2011, 27, 5252–5263

Langmuir

ARTICLE

Table 1. Density, Molecular Volume, and Scattering Length Density G of the Materials Used in This Study species

density, g cm3

volume, Å3

Table 2. Critical Micelle Concentration (cmc) and Adsorption Parameters (eq 9) for the Surfactants in SBA-15 (20 C)a

F, 106 Å2

surfactant

cmc, mol L1 5

nam, mmol g1

K

N

c0 /cmc

C12E5

0.963

701

0.13

C12E5

6.5  10

1.25 ( 0.08

49

4.1

0.21 ( 0.04

C10E5 C10G2

0.973 1.0

646 702

0.17 0.61

C10E5 C10G2

8.0  104 1.85  103

1.15 ( 0.08 0.05 ( 0.005

16 10

4.1 2.4

0.27 ( 0.04 0.11 ( 0.04

C12H25

0.80

351

0.39

(EO)5

1.14

321

0.66

G2

1.61

351

1.75

H2O

0.997

30.3

0.56

D2O

1.108

30.3

6.35

SiO2

2.16

3.7

a a n m:

maximum specific adsorption. K: adsorption constant. N: nominal surface aggregation number. c0/cmc: onset concentration of aggregative adsorption relative to critical micelle concentration in the bulk (see eq 9)

to that of the core liquid, Fc (see Figure 1), i.e., relative film density Ff/Fc = FF and relative matrix density F0/Fc = FM, assuming Fc = FD2O and F0 = FSiO2 (see Table 1). Examples for the fit are shown in Figure 2. From measurements in the contrast-matching scenario only the film thickness t = R0  Rf can be extracted, and the information about the scattering length density of the film is lost.

4. RESULTS 4.1. Surfactant Adsorption Isotherms. Figure 3 shows the adsorption isotherms (20 C) of the surfactants C12E5, C10E5, and C10G2 in the present SBA-15 sample. For all surfactants the specific adsorbed amount na reaches a plateau value nam shortly above the cmc. The isotherms of C12E5 and C10E5 exhibit a pronounced S shape. They can be represented by the one-step model equation for aggregative adsorption by Gu and Zhu,40 which we express in terms of relative concentrations c/cmc as

na ¼ nam Figure 2. (Top) Experimental integral intensities ~I hk (b) of the three leading Bragg reflections and fit by the form factor model function according to eq 3 (line). (Bottom) Step density profiles F(r) resulting from the fits in the upper part. The examples represent C12E5 in SBA-15 at full contrast (in D2O) at f = 0.3 (left) and 1 (right).

Figure 3. Adsorption isotherms of the surfactants C12E5 (9), C10E5 (b), and C10G2 (() in SBA-15 (20 C) plotted as specific adsorbed amount na versus relative concentration c/cmc, and fits of the data by eq 9 (curves). (Inset) Isotherm of C10G2 plotted on an enlarged scale.

Kðc=cmcÞN 1 þ Kðc=cmcÞN

ð9Þ

where K and N are constants. The onset concentration of aggregative adsorption, c0, is related to these parameters as c0/ cmc = [(N  1)/(N þ 1)](Nþ1)/NK1/N. A fit of the adsorption data by eq 9 is shown in Figure 3. Values of the fit parameters nam, K, and N and onset concentration c0 are given in Table 2. For C12E5 and C10E5 we find c0/cmc = 0.24 ( 0.06, which implies that surface aggregation in the narrow pores occurs at a significantly lower concentration than at flat silica surfaces, where c0/cmc = 0.7 ( 0.2.10,16 The plateau value of the adsorption corresponds to a surface concentration Γm = nam/as of 1.4 (C12E5) and 1.3 μmol m2 (C10E5), which is lower by a factor 4 than on flat silica surfaces.41 The low values of c0/cmc and Γm are indicators of strong confinement of the surfactant aggregates in the pores of SBA-15. The values of Γm for C10E5 and C12E5 are consistent with adsorption isotherms of the surfactant C8E4 in a series of controlled-pore silica glass (CPG-10) materials reported by Dietsch et al.10 These authors found a decrease in Γm from 6.2 to 2.4 μmol m2 as the mean pore width decreased from 50 to 10 nm. The even lower values of Γm for C10E5 and C12E5 in the present SBA-15 material can be attributed to the larger chain length of these surfactants compared to C8E4 and the smaller pore diameter of our SBA-15 sample as compared to the CPG-10 material of smallest mean pore size studied by these authors. The volume fraction of surfactant in the pore at a given relative adsorption f = na/nam is estimated as φ¼f

constants: The pore radius R0 was taken as DKJS/2 = 4.07 nm, and the scattering length densities in eq 3 were expressed relative 5255

0:9nam MT ~T vp F

ð10Þ

dx.doi.org/10.1021/la200333q |Langmuir 2011, 27, 5252–5263

Langmuir

ARTICLE

Table 3. C10G2 in Slurry Samples of SBA-15 in ContrastMatching H2O/D2O: Fit with Form Factor of Prolate Ellipsoidsa c/cmc

ε

R, nm

V, nm3

N

30

1.60 ( 0.05

2.0 ( 0.1

34 ( 2

116 ( 6

20

1.64 ( 0.05

1.9 ( 0.1

35 ( 2

118 ( 6

10

1.70 ( 0.05

2.1 ( 0.1

43 ( 2

146 ( 6

Parameters of prolate ellipsoids (R, minor axis; ε, ellipticity; V, volume; N, aggregation number) for three different surfactant concentrations c/cmc in the aqueous phase.

a

Figure 4. (a) Scattering profiles I(q) for SBA-15 in contrast-matching H2O/D2O: (O) without surfactant, ()) with C10G2, and (b) with C10E5. Error bars of scattering data do not exceed the size of the data points. The total amount of surfactant in the cell was 2.6 mg for C10G2 and 2.8 mg (f = 0.15) for C10E5. (b) Scattering profiles for C10G2 (relative concentrations c/cmc = 30, 20, 10) in the presence of SBA-15 in contrast-matching H2O/D2O, and fits of the data by the form factor of prolate ellipsoidal micelles; fit parameters are given in Table 3.

~T the mass density of the with MT being the molar mass and F surfactant, vp is the specific pore volume of SBA-15 (vp= 1.052 cm3 g1), and the factor 0.9 takes into account that ca. 10% of the surfactant is adsorbed at the outer surface of the silica particles (see section 2.1). For the volume fraction at maximum adsorption (f = 1) this yields φm= 0.452 for C12E5 and φm= 0.385 for C10E5. Adsorption of C10G2 in the pores of SBA-15 is much weaker than that for C10E5, in agreement with adsorption studies at flat silica surfaces.9 From the maximum adsorption nam (Table 2) the pore volume fraction is estimated as φm= 0.020 at the cmc of C10G2. 4.2. Scattering Profiles: Influence of Adsorption Strength. Representative scattering intensity profiles I(q) of silica slurry samples in contrast-matching H2O/D2O are shown in Figure 4a. In the absence of surfactant only a uniform scattering background Ib due to incoherent scattering from protons is detected. In the presence of surfactant diffuse scattering appears as a signature of surfactant aggregates embedded in a uniform background. In addition, the leading Bragg reflection from the pore lattice of SBA-15 shows up in the presence of C10E5 but not for C10G2, although the amount of surfactant in the two samples was similar (2.8 mg for C10E5, 2.6 mg for C10G2). This can be attributed to the different adsorption behavior of the surfactants: The strongly adsorbed C10E5 is accumulating in the pores, causing a significant pore volume fraction (φ ≈ 0.07) and thus a significant contrast between pore space and matrix. The weakly adsorbed C10G2 is distributed over the entire liquid volume (pores and interparticle space), and thus, the scattering contrast between pore space and matrix remains low. Slurry samples of SBA-15 with C10G2 in contrast-matching H2O/D2O exhibit SANS profiles similar to those of micellar solutions of the surfactant in water. Results for liquid-phase concentrations c/cmc of 10, 20, and 30 in the slurry samples are shown in Figure 4b. The data can in principle be described with a form factor of spherical micelles, but this would give an unrealistic broad size distribution. Figure 4b shows a fit of the data with the form factor of prolate ellipsoids. The resulting parameters are given in Table 3. Within the margins of experimental error (poor statistics at high q, see Figure 4b) the values for the three concentrations can be considered as identical (R = 1.65 nm; ε = 2.0). The values of the minor and major axes, R and εR, are smaller than what is expected for the

entire micelle,42,43 but compatible with the respective values of the hydrocarbon core of micelles of aggregation number N = 130 ( 15 (using N = V/vtail, with V = (4π/3)εR3 and vtail = 0.30 nm3) (see Table 1). This is plausible in view of the fact that the outer layer of head groups of C10G2 micelles has low contrast against the H2O/ D2O solvent: With an estimated hydration number of 3 water molecules per maltoside headgroup43 one finds Fshell = 2.4  106 Å2 (at the chosen H2O/D2O solvent composition) and thus ΔFshell = |Fshell  Fsolvent| = 1.3  106 Å2, as compared to the core Fcore = 0.4  106 Å2 and ΔFcore = |Fcore  Fsolvent| = 4.1  106 Å2. Accordingly, the SANS profiles reflect mainly the size and shape of the micelle core. Hence, we may conclude that the aggregates of C10G2 formed in the slurry samples are similar to the micelles formed in the absence of the silica, in agreement with the finding that this surfactant is only marginally adsorbed at the pore walls. Scattering intensity profiles for the strongly adsorbed surfactants C10E5 and C12E5 in SBA-15 were obtained for relative adsorption f = na/nam in the range 0.151. Results for C10E5 at contrast-match conditions in H2O/D2O are displayed in Figure 5, and results for C12E5 at full-contrast conditions in D2O are shown in Figure 6. In both cases one finds strong diffuse scattering superimposed with Bragg reflections from the pore lattice. At low relative adsorption the diffuse scattering intensity Idiff(q) is a monotonic decreasing function of q. At higher f a maximum in Idiff(q) appears in the experimental q range, and this maximum moves to higher q as f increases. The intensity of Bragg reflections depends on the solvent contrast and the surfactant filling factor: Whereas in H2O/D2O at low surfactant loadings the Bragg reflections are weak (Figure 5), at full contrast in D2O and low f the (10) Bragg reflection is the prominent feature (Figure 6). At high adsorption of surfactant in the pores the differences between the two contrast scenarios become less pronounced and in both cases the intensities of the (11) and (20) Bragg reflections become higher than that of the (10) reflection, which is a signature of film formation at the pore wall. 4.3. Results for C12E5 at Different Solvent Contrasts. SANS intensity profiles for C12E5 in SBA-15 measured at full contrast are presented in Figure 6. Also shown are the constituting contributions from diffuse scattering and Bragg scattering as obtained on the basis of the theoretical model outlined in section 3. Most data sets could be fitted very well with this model. Scattering profiles for this system measured in contrast-matching H2O/D2O were already reported in ref 31. R Figure 7 shows the integral scattering intensity ~I = I(q)q2 dq and its contributions from diffuse scattering and Bragg scattering, ~I diff and ~I Bragg, as a function of the relative adsorption f for the two contrast scenarios. Remembering that scattering in the 5256

dx.doi.org/10.1021/la200333q |Langmuir 2011, 27, 5252–5263

Langmuir

ARTICLE

Figure 5. Scattering intensity profiles I(q) in contrast-matching water for C10E5 in SBA-15 at different surfactant loadings: (a) 15%, (b) 30%, (c) 45%, (d) 60%, (e) 80%, and (f) 100% (20 C): scattering data (black circles) and fit by the present model (full line) with the contribution by diffuse scattering (dashed-dotted red lines) and of the three leading Bragg peaks (blue areas). Error bars of I(q) data do not exceed the size of the data points. Positions of the Bragg peaks (qhk) are marked by vertical lines.

contrast-match scenario is equivalent to a two-phase system of surfactant and matrix, we expect a dependence of ~I on the overall volume fraction of surfactant Φ as ~I  Φ(1  Φ). Here, Φ is related to the pore volume fraction φ (eq 10) by Φ = (msvp/ Vcell)φ, where ms is the mass of SBA-15 in the sample cell (volume Vcell). For C12E5 at maximum adsorption (f = 1) this yields Φm = 0.14, i.e., a value well below Φ = 0.5 at which ~I reaches a maximum. Thus, the dependence of ~I on f in Figure 7b is consistent with the behavior expected on the basis of the twophase model. In the full-contrast scenario (Figure 7a) ~I exhibits a more complex behavior at high relative adsorption, which may be attributed to a minimum in the overall contrast in this regime (see below). In both contrast scenarios ~I diff represents the dominating contribution over the entire range of surfactant fillings. ~I Bragg is small at low surfactant concentration in the pores but becomes significant at high f in the contrast-match scenario. Figure 8 presents structural parameters derived from the fit of Idiff(q) with eq 6 for C12E5 in SBA-15. Results from the measurements at full contrast (D2O) are compared with those from the earlier measurements at contrast match (H2O/D2O).31

From the primary fit parameters qm and I0/Im (Figure 8a and b) the physical parameters d (quasiperiodic separation) and ξ (correlation length) are calculated using eq 7 (Figure 8c and d). The estimated error bars on d and ξ are large at low surfactant fillings but much smaller at high f. In most cases the values of d and ξ derived from the full-contrast series are consistent with those from the contrast-match series. Results from the Bragg scattering contribution IBragg(q) for C12E5 in SBA-15 are displayed in Figure 9. The integral Bragg intensities ~I 10, ~I 11, and ~I 20 obtained in the full-contrast series (Figure 9a) and the respective results from the contrast-match series (Figure 9b)31 exhibit different dependencies on the surfactant filling factor f. These differences are most pronounced at low f as a consequence of the different scattering length density profiles F(r) in the pore (see below). Values of the average thickness t and scattering length density Ff of the surfactant layer at the pore wall as derived from the integral Bragg intensities on the basis of the form factor model are shown in Figure 10. The values of t as a function of the filling factor f are compared with the respective values resulting from the measurements at contrast-match conditions (Figure 10a). The two data sets are in 5257

dx.doi.org/10.1021/la200333q |Langmuir 2011, 27, 5252–5263

Langmuir

ARTICLE

Figure 6. Scattering intensity profiles I(q) at full contrast (D2O) for C12E5 in SBA-15 at different surfactant loadings: (a) 15%, (b) 30%, (c) 45%, (d) 60%, (e) 80%, and (f) 100% (20 C): symbols and curves as in Figure 5.

Figure 7. C12E5 in SBA-15: Integrated scattering intensity, ~I , and constituent contributions from diffuse and Bragg scattering, ~I diff and ~I Bragg, as a function of the surfactant filling factor f: (a) full contrast in D2O and (b) contrast matched in H2O/D2O.

qualitative agreement and indicate a steady increase of t with increasing f, up to a maximum layer thickness of 2.53 nm at maximum adsorption. Values derived from the full-contrast series are somewhat higher (typically by 0.5 nm) than those

from the contrast-match series. This may be a consequence of approximating the radial scattering length density profile F(r) by a two-step model. If the interface from the surfactant layer to core liquid has a nonsymmetric density profile, lower values of the mean radius Rf (and thus higher values of the layer thickness t) will be obtained the higher the scattering length density of the core liquid (cf. Figure 1). The mean scattering length density of the film Ff (here expressed relative to the density of D2O) is shown in Figure 10b. The results, although subject to large uncertainty at low filling fractions, indicate that Ff decreases with increasing filling fraction, being lower than the matrix density F0 at high filling fractions but somewhat higher than F0 at low fillings. We return to this point in the Discussion. 4.4. Comparison of C10E5 and C12E5. The scattering intensity functions for C10E5 in SBA-15 measured in the contrast-match scenario (Figure 5) were analyzed in the same way as the data for C12E5. A comparison of the parameters resulting for the two surfactants is shown in Figure 11. The fit parameters qm and Io/ Im(Figure 11a and 11b) and the derived physical parameters d 5258

dx.doi.org/10.1021/la200333q |Langmuir 2011, 27, 5252–5263

Langmuir

Figure 8. C12E5 in SBA-15: Parameters derived from diffuse scattering Idiff(q) in the two contrast scenarios as a function of surfactant filling factor f: (a) position of correlation peak qm, (b) intensity ratio I0/Im, (c) quasiperiodic distance d, and (d) correlation length ξ. Results from measurements at full contrast in D2O (9) are compared with those in contrast-matching H2O/D2O (b).

ARTICLE

Figure 10. C12E5 in SBA-15: Parameters of the scattering length density profile derived from the Bragg intensities: (a) effective film thickness t, comparison of results from measurements at full contrast in D2O (9) and in contrast-matching H2O/D2O (b); (b) mean scattering length density of the surfactant layer Ff/FD2O, results from two-parameter fit of data measured at full contrast (9) and calculated from the layer thickness and adsorbed amount via eqs 11 and 12 (O). The matrix density F0/FD2O is indicated by the dotted line.

not accessible from the scattering data obtained at contrastmatch conditions.

5. DISCUSSION 5.1. Surfactant Density Profile. The surfactant density profile in the pores F(r), which we approximate by a step function with a layer of uniform density, gives an interesting result. From measurements at full contrast (in D2O) the thickness t and density Ff of the layer are both accessible (Figure 10a and 10b). The adsorption layer is made up of surfactant and water (see Figure 1), and its mean scattering length density is related to the volume fraction of surfactant in the layer, θ, by

Ff ¼ θFT þ ð1  θÞFD2O

Figure 9. C12E5 in SBA-15: Intensities ~I hk of the three leading Bragg reflections hk (b, 10; 9, 11; 2, 20) as a function of surfactant filling factor f: (a) measurements at full contrast (in D2O) and (b) measurements at contrast-matching conditions (in H2O/D2O).

and ξ for the surfactant aggregates in the pores (Figure 11c and 11d) show similar trends as a function of the relative adsorption f. We note that the values of d for C10E5 are lower by about 1 nm than for C12E5 at given f, while the values of the correlation length are equal within experimental uncertainty. A comparison of the mean layer thickness t of the two surfactants is shown in Figure 11e. A nearly linear increase of t with f is found for both surfactants up to a relative adsorption f ≈ 0.6 and a further (weaker) increase at higher f. The maximum value of the mean film thickness is 2.0 nm for C10E5 and 2.6 nm for C12E5. The mean density of the surfactant layer is

ð11Þ

where FT and FD2O are the scattering length densities of pure surfactant and D2O. Since FT < FD2O (Table 1), an increase in θ causes a decrease of Ff. Hence, from Figure 10a and 10b we conclude that both the layer thickness and the surfactant volume fraction in the layer increase with the surfactant filling factor. This leads to a crossover from a monotonic decreasing profile (Fc > Ff > F0) at low surfactant loadings to a situation where Ff is lower than Fc and Fo at high loadings, as shown in Figure 2. A shortcoming of the determination of the surfactant density profile from the intensities of the Bragg reflections is that only three Bragg reflections of the SBA-15 matrix can be resolved quantitatively in neutron scattering experiments. In measurements at contrast-match conditions, one parameter (layer thickness t) is derived from these three data points. In measurements at full contrast in D2O, on the other hand, two parameters (layer thickness t and mean layer density Ff) have to be extracted from just three data points (see Figure 2). We can check the consistency of the resulting layer densities by estimating θ from 5259

dx.doi.org/10.1021/la200333q |Langmuir 2011, 27, 5252–5263

Langmuir

ARTICLE

Table 4. Layer Thickness t and Surfactant Volume Fraction θ in the Adsorption Layera C10E5

C12E5

f

t , nm

θ

θhydr

t , nm

θ

θhydr

0.30

0.7

0.35

0.5

0.9

0.35

0.5

0.45

1.6

0.27

0.39

1.6

0.32

0.47

0.60 0.80

1.9 2.0

0.32 0.41

0.47 0.60

2.0 2.25

0.37 0.45

0.52 0.65

1.0

2.0

0.52

0.75

2.65

0.52

0.73

Layer thickness t from one-parameter fit of Bragg intensities measured in contrast-matching H2O/D2O; volume fractions θ and θhydr of plain surfactant and hydrated surfactant in the adsorption layer estimated by eq 12. a

Figure 11. Comparison of results for C10E5 (2) and C12E5 (b) in SBA15 in contrast-matching H2O/D2O as a function of surfactant filling factor f: (a) position of correlation peak qm, (b) relative intensity Im/I0, (c) quasiperiodic distance d, (d) correlation length ξ, and (e) layer thickness t.

the known relative adsorption f and the layer thickness t as derived from measurements in contrast-matching H2O/D2O using the relation θðf , tÞ ¼

0:9fnam MT =~ F T vp φðf Þ ¼ ψðtÞ ðt=R0 Þð2  t=R0 Þ

ð12Þ

In eq 12, ψ(t) = (1  R2f /R02) with Rf = R0  t represents the fraction of the pore space that constitutes the surfactant layer. Results for θ as a function of the relative adsorption f and layer thickness t are given in Table 4, and values for Ff at full scattering contrast obtained on the basis of eqs 11 and 12 are included in Figure 10b. It can be seen that the layer densities Ff derived in this way show the same trends as those found in the two-parameter fit: With increasing f the layer density Ff decreases from values greater than F0 to values lower than F0. Accordingly, there is a crossover from a monotonic decreasing density profile to a situation where the scattering length density of the film is lower than that of the silica matrix and the liquid in the core, as shown in Figure 2. Closer inspection of Table 4 shows a further interesting result. For both surfactants we find that at loadings up to f = 0.6 the addition of surfactant causes mainly an increase of the layer thickness t while the volume fraction θ remains relatively low and roughly constant. As the loading is further increased the layer thickness is not growing much further; instead, the surfactant volume fraction in the layer is increasing. For both surfactants we find θ ≈ 0.5 at maximum loading. Note that θ represents the volume fraction of dry surfactant. Since the surfactant heads are hydrated, the volume fraction of hydrated surfactant, θhydr, is a

more relevant figure. Values of θhydr (based on two water molecules per oxyethylene group)13 are also given in Table 4. For both surfactants we find θhydr ≈ 0.75 at maximum loading. From the sigmoidal shape of the adsorption isotherms (Figure 3) we infer that the surfactant is not distributed uniformly in the adsorption layer but is forming compact surface aggregates. The above value of θhydr then implies that 75% of the pore walls is covered with surfactant aggregates of a typical thickness 2 nm (for C10E5) or 2.6 nm (for C12E5). This difference in layer thickness is greater than to be expected from the difference in chain length of the two surfactants. It hints at a somewhat different organization of the surfactant in the surface aggregates. However, we refrain from speculation in view of the limited accuracy in the determination of the layer thickness. Obviously the thickness of the surfactant layer at the pore walls remains much smaller than a bilayer on a flat silica surface (4.2 nm for C12E5),41 indicating a highly distorted structure of the surface aggregates in the confined pore space. 5.2. Size of Surface Aggregates. Information about the mean size of surface aggregates and their distribution in the SBA-15 matrix can be obtained from the parameters d and ξ, which we derived from the diffuse scattering profiles using eq 7. At low surfactant loadings we expect isolated surface micelles. In this case d represents the mean distance between the surface micelles and ξ the correlation length (which is related to the aggregate size) of this quasiperiodic arrangement of surface micelles in the porous matrix. For C12E5 at the lowest loading (f = 0.15), d is close to the center-to-center distance a0 of the pore lattice (10 nm). This then corresponds to an arrangement with one surface micelle every 10 nm in each pore. If the size of surface micelles is independent of the loading, the observed decrease of d implies that the number of surface aggregates increases with f. The mean size of the surface aggregates at a given surfactant loading can be estimated from the known adsorbed amount and the mean separation of the surface aggregates in the matrix. Specifically, the average volume of an aggregate Vagg is derived from the specific volume of adsorbed surfactant (va) and the number of surface micelles per unit mass of silica (which we denote by N) as Vagg = va/N. When assuming that the surface micelles in the porous matrix are arranged in a simple cubic lattice of lattice constant d, the number of surface micelles per unit mass of silica is given by N = vtot/d3, where vtot is the volume of unit mass of the silica matrix. For a perfect 2D hexagonal crystallite of SBA-15 this is given by vtot = [~F0(1  2πS2/(3)1/2)]1, where ~0 is the mass density of the silica matrix. For the S = R0/a0 and F present SBA-15 sample S = 0.399. The specific volume of 5260

dx.doi.org/10.1021/la200333q |Langmuir 2011, 27, 5252–5263

Langmuir

ARTICLE

Table 5. Surface Aggregate Size Estimated from the QuasiPeriodic Distance in the Silica Matrixa surfactant

f

d, nm

N, 1018g1

Vagg, nm3

Rsph, nm

Rdisk, nm

C10E5

0.30

7.4

2.7

44

2.2

2.0

0.45 0.60

6.4 5.8

4.2 5.8

43 42

2.2 2.2

2.0 1.9

0.80

5.1

8.1

40

2.1

1.9

1.0

4.7

10.4

39

2.1

1.8

0.15

9.0

1.5

47

2.2

1.6

0.30

8.2

2.0

72

2.6

2.1

0.45

7.4

2.7

80

2.7

2.2

0.60

6.7

3.6

78

2.7

2.2

0.80 1.0

6.1 5.3

4.8 7.4

80 64

2.7 2.5

2.2 1.9

C12E5

a

d: quasi-periodic separation of surfactant aggregates. N: number of surfactant aggregates per unit mass of the silica matrix. Vagg: volume of dry surfactant aggregate. Rsph: radius of spherical aggregate. Rdisk: radius of disk-like surface aggregate (see text).

surfactant in the pores was estimated as va = 0.9 fnamMT/~ FT (see ~0= 2.16 g cm3. eq 10) with the values of nam from Table 2 and F Results for Vagg as a function of the relative adsorption are presented in Table 5. These values of Vagg again correspond to the volume of dry (unhydrated) surfactant aggregate. For C10E5 we find Vagg = 40 nm3, nearly independent of the surfactant loading. This corresponds to spherical micelles of radius Rsph = 2 nm, which is in the size range of micelles in solution. For C12E5, an aggregate volume similar to that of C10E5 is found at low surfactant loadings (f = 0.15) but aggregate volumes up to 80 nm3 are found at higher loadings, corresponding to spherical micelles of radius 2.7 nm. Note, however, that spherical surface micelles of radius Rsph = 2 nm (i.e., a diameter of 4 nm) or greater are not compatible with the finding that the mean thickness of the surfactant layer at the pore wall is not exceeding a value tmax of 2 nm for C10E5 and 2.6 nm for C12E5 (Table 4). This suggests that surface micelles are flattened to a disk-like shape. For disk-like micelles (radius Rdisk, thickness H) with a hemicylindrical perimeter (cylinder radius H/2) the aggregate volume is Vagg = πRdiskH(Rdisk þ πH/4). The disk radius Rdisk was estimated from the aggregate volume on the assumption that H is given by the maximum layer thickness of surfactant in the pore (H = tmax). Values of Rdisk based on this assumption are included in Table 5. They suggest that the overall diameter of the disk-like surface micelles, Ddisk = 2Rdisk þ H, is about 6 nm for C10E5 and 7 nm for C12E5. It is expected that the surface micelles will adapt to the curved pore wall in such a way that the heads of surfactant molecules in the outer sheet experience favorable interactions with surface silanol groups. The model of isolated surface micelles in the pore space is reasonable for low surfactant loadings but becomes questionable when the mean distance between aggregates comes close to the size of the aggregates. Inspection of Table 5 shows that this is the case for loadings f > 0.3. Beyond this limit the surfactant aggregates can merge to larger patches. We have seen that at the highest loading (f = 1) the volume fraction of surfactant in the surface layer is θ ≈ 0.5 (dry surfactant) and θhydr ≈ 0.75 (hydrated surfactant). In this regime, when a substantial part of the pore walls is covered by surfactant patches, the characteristic distance d between surfactant patches is determined mostly by the pore diameter (8 nm) and the wall thickness of the silica matrix (2 nm): For surfactant layers of 2.6 nm thickness (C12E5) the

center-to-center distance between surfactant patches disposed visa-vis in the same pore is about 5.5 nm and the minimum center-tocenter distance for layers disposed at the wall of neighboring pores is about 4.5 nm. Hence, it is plausible to find that d attains a value near 5 nm at the highest loading (Table 5). In line with this interpretation we find that the correlation length ξ of the quasiperiodic arrangement of surfactant aggregates in the matrix reaches high values at the highest surfactant loadings. 5.3. Relation to Published Work. The present work confirms the results of our recent study31 and shows that the surfactants C10E5 and C12E5 assemble to surface aggregates of similar structures in the cylindrical pores. The somewhat smaller values of aggregate size and maximum layer thickness of C10E5 in the pores are consistent with the shorter tail length of this surfactant. The grossly different self-assembly observed for C10G2 confirms earlier reports of the different adsorption behavior of alkyl ethoxylate and alkyl glucoside or maltoside surfactants on silica9 and other oxide surfaces.44 The different behavior is believed to arise from the strong water-bonding capacity of the multiple OH groups of the sugar head, which causes a lower affinity for the silica surface as compared to the ethoxylate group.42,43,45 This conclusion is supported by the significantly smaller differential molar enthalpy of adsorption in the initial low-affinity region of the isotherm of C8G1 in comparison to C8E4.9 The present work also shows that the onset concentration of aggregative adsorption of C10E5 and C12E5 in the pores (c0/cmc in Table 2) is lower than on flat or weakly curved surfaces (where c0/cmc = 0.7 ( 0.2). This may be due to a higher number of silanol anchor points per aggregate on the highly curved pore walls as compared with a flat surface24 or a surface of opposite (convex) curvature, as in the case of silica nanoparticles.18 It is desirable to check this conjecture by a method which directly probes effects of curvature on the surface aggregates. Qiao et al.25 employed 2H NMR using a selective deuterium label in different positions of the surfactant tail to probe the mean curvature of surface aggregates of C12E5 surface micelles confined in controlled-pore glass (CPG-10) materials. From the order parameter Sx of the CD bond as a function of the label position x in the surfactant tail it was concluded that the surface aggregates of C12E5 in CPG-10 materials (mean pore width 13 nm or greater) had an average curvature similar to that of the lamellar phase. Hence, a flat bilayer or disrupted bilayer type structure of the surface aggregates was inferred. However, it is likely that this will not apply for the surfactant aggregates in SBA-15 due to the much stronger confinement in 8 nm cylindrical pores. Finally, we note that the aggregate structures of surfactants observed in the pores of SBA-15 silica bear little resemblance with the cornucopia of polymorphic structures reported in a simulation study based on dissipative particle dynamics (DPD) simulations of a short-chain surfactant/water system confined to a nanotube.46 It seems that the interaction parameters used to describe the interaction of the surfactant head groups with water and a hydrophilic or hydroneutral pore wall was not appropriate to describe the adsorption of the ethoxylate surfactants at silica surfaces in a realistic manner.

6. CONCLUSIONS Self-assembly of nonionic surfactants in the cylindrical pores of SBA-15 silica has been investigated by SANS. C10E5 and C12E5 exhibit strong aggregative adsorption as evidenced by the sigmoidal shape of the adsorption isotherms. The SANS intensity 5261

dx.doi.org/10.1021/la200333q |Langmuir 2011, 27, 5252–5263

Langmuir

’ AUTHOR INFORMATION Corresponding Author

*Phone: þ49 30 3142 4171. Fax: þ49 30 3142 6602. E-mail: fi[email protected].

Present Addresses

)

profiles I(q) exhibit diffuse scattering with a correlation peak resulting from surfactant aggregates in the pores, superimposed with Bragg reflections resulting from the pore lattice of SBA-15. These two contributions to I(q) yield complementary information about the surfactant assembly in the pores: The Bragg intensities are analyzed on the basis of a form factor model of the cylindrical pores, which yields the mean thickness and scattering length density of the adsorbed layer. The diffuse scattering can be represented by a TeubnerStrey-type scattering function which is applicable to systems of weakly ordered objects with a preferred distance between next neighbors. Major conclusions of this study are (i) SANS profiles measured at different solvent contrasts (full contrast with pure D2O or contrast match of the silica matrix with a H2O/D2O mixture) can be analyzed by the same structural model and yield concordant values of the model parameters. (ii) Two parameters of the density profile (layer thickness t and mean scattering length density Ff of the surfactant layer) can be deduced from Bragg scattering contribution at full contrast, while only the layer thickness is accessible from the profiles measured at contrast-match conditions. However, since the values derived from one-parameter fits of the form factor at contrast-match conditions are more reliable, it is preferable to use these values of the layer thickness and extract the volume fraction of the surfactant in that layer indirectly from the adsorption isotherm. (iii) For the surfactants C10E5 and C12E5 we find that at relative adsorption up to f = 0.6 the addition of surfactant causes mainly an increase of the effective layer thickness t while the volume fraction θ remains relatively low. As f is further increased, the mean surfactant volume fraction in the layer increases while the layer thickness remains nearly constant. The layer thickness at maximum loading in the 8.1 nm pores is 2.0 nm for C10E5 and 2.6 nm for C12E5. The volume fraction of hydrated surfactant in the layer reaches a maximum value θhydr = 0.75 in both cases. (iv) The appearance of a correlation peak in the diffuse scattering indicates a preferred distance d between adsorbed surfactant aggregates. The large d found at low loadings suggests that adsorption leads to isolated surface micelles in this regime. The mean size of the surface aggregates in the pores can be deduced from d and the known amount of adsorbed surfactant. The aggregate volume corresponds to spherical micelles of diameter 4.4 nm for C10E5 and up to 5.4 nm for C12E5. However, it is conjectured that surface micelles of lower thickness are formed in the pores to optimize the interactions of the head groups with the strongly curved pore wall. (v) With increasing surfactant loading the number of surface aggregates in the pores increases. Above a certain threshold concentration these aggregates will merge to interconnected patches of a surfactant bilayer. The high values of the correlation length ξ at relative adsorption f g 0.6 indicate that the nature of the adsorbate is predominantly layer-like in this regime.

ARTICLE

Nano-Science Center, Department of Chemistry, University of Copenhagen, Universitetsparken 5, 2100 Copenhagen, Denmark ^ Institute of Nano Science and Technology, Department of Physics, Hanyang University, 17 Haengdang-dong, Seongdong-gu, Seoul, 133-791, Korea

’ ACKNOWLEDGMENT We wish to thank Bruno Deme for advice and assistance with the neutron scattering measurements at Institut Laue Langevin (ILL) and Bhuvnesh Bharti for help with data analysis. We gratefully acknowledge the allocation of neutron beam time at ILL and travel grants provided by ILL to T.S. and G.H.F. Financial support by the German Science Foundation (DFG) in the framework of SFB 448 and IGRTG 1524 is also gratefully acknowledged. ’ REFERENCES (1) Leermakers, F. A. M.; Koopal, L. K.; Lokar, W. J.; Ducker, W. A. Langmuir 2005, 21, 11534. (2) Mudler, M. Basic Principles of Membrane Technology; Kluwer Academic Publishers: Dordrecht, 1992. (3) In Surfactant-Based Separation Processes; Scamehorn, J. F., Harwell, J. H., Eds.; Surfactant Science Series; Marcel Dekker: New York, 1989; Vol. 33. (4) Ruiz-Angel, M. J.; Carda-Broch., S.; Torres-Lapasio, J. R.; Garcia-Alvarez-Coque, M. C. J. Chromatogr. A 2009, 1216, 1798. (5) Schmuhl, R.; van der Berg, A.; Blank, D. H. A.; ten Elshof, J. E. Angew. Chem., Int. Ed. 2006, 45, 3341. (6) Klimenko, N. A.; Koganovskii, A. M. Kolloidn. Zh. 1974, 36, 151. Klimenko, N. A.; Kofanov, V. I.; Sivalov, E. G. Kolloidn. Zh. 1981, 43, 287. (7) Levitz, P.; van Damme, H. J. Phys. Chem. 1986, 90, 1302. (8) Seidel, J.; Wittrock, C.; Kohler, H. H. Langmuir 1996, 12, 5557. (9) Kiraly, Z.; B€ orner, R. H. K.; Findenegg, G. H. Langmuir 1997, 13, 3308. Kiraly, Z.; Findenegg, G. H. Langmuir 2000, 16, 8842. (10) Dietsch, O.; Eltekov, A.; Bock, H.; Gubbins, K. E.; Findenegg, G. H. J. Phys. Chem. C 2007, 111, 16045. (11) Grant, L. M.; Tiberg, F.; Ducker, W. A. J. Phys. Chem. B 1998, 102, 4288. (12) Tiberg, F.; Brinck, J.; Grant, L. M. Curr. Opin. Colloid Interface Sci. 2000, 4, 411. (13) B€ ohmer, M. R.; Koopal, L. K.; Janssen, R.; Lee, E. M.; Thomas, R. K.; Rennie, A. R. Langmuir 1992, 8, 2228. (14) Penfold, J.; Staples, E.; Tucker, I. Langmuir 2002, 18, 2967. (15) Steitz, R.; M€uller-Buschbaum, P.; Schemmel, S.; Cubitt, R.; Findenegg, G. H. Europhys. Lett. 2004, 67, 962. (16) Levitz, P. Langmuir 1991, 7, 1595. (17) Bock, H.; Gubbins, K. E. Phys. Rev. Lett. 2004, 92, 135701. (18) Lugo, D.; Oberdisse, J.; Karg, M.; Schweins, R.; Findenegg, G. H. Soft Matter 2009, 5, 2928. Lugo, D.; Oberdisse, J.; Lapp, A.; Findenegg, G. H. J. Phys. Chem. B 2010, 114, 4183. (19) Mugele, F.; Salmeron, M. Phys. Rev. Lett. 2000, 84, 5796. (20) Israelachvili, J. N.; Gourdon, D. Science 2001, 292, 867.  (21) Perret, E.; Nygard, K.; Satapathy, D. K.; Balmer, T. E.; Bunk, O.; Heuberger, M.; van der Veen, J. F. Europhys. Lett. 2009, 88, 36004.  (22) Nygard, K.; Satapathy, D. K.; Perret, E.; Padeste, C.; Bunk, O.; David, C.; van der Veen, J. F. Soft Matter 2010, 6, 4536. (23) (a) Huinink, H. P.; de Keizer, A.; Leermakers, F. A. M.; Lyklema, J. Langmuir 1997, 13, 6452. (b) Huinink, H. P.; de Keizer, A.; Leermakers, F. A. M.; Lyklema, J. Langmuir 1997, 13, 6618. (24) Giordano, F.; Denoyel, R.; Rouquerol, J. Colloids Surf. A 1993, 71, 293. 5262

dx.doi.org/10.1021/la200333q |Langmuir 2011, 27, 5252–5263

Langmuir

ARTICLE

(25) Qiao, Y.; Sch€onhoff, M.; Findenegg, G. H. Langmuir 2003, 19, 6160. (26) Zhao, D. Y.; Feng, J. L.; Huo, Q. S.; Melosh, N.; Fredrickson, G. H.; Chmelka, B. F.; Stucky, G. D. Science 1998, 279, 548. (27) J€ahnert, S.; M€uter, D.; Prass, J.; Zickler, G. A.; Paris, O.; Findenegg, G. H. J. Phys. Chem. C 2009, 113, 15201. (28) Findenegg, G. H.; J€ahnert, S.; M€uter, D.; Prass, J.; Paris, O. Phys. Chem. Chem. Phys. 2010, 12, 7211. (29) Erko, M.; Wallacher, D.; Brandt, A.; Paris, O. J. Appl. Crystallogr. 2010, 43, 1. (30) Mascotto, S.; Wallacher, D.; Kuschel, A.; Polarz, S.; Zickler, G. A.; Timmann, A.; Smarsly, B. M. Langmuir 2010, 26, 6583. (31) M€uter, D.; Shin, T.; Deme, B.; Fratzl, P.; Paris, O.; Findenegg, G. H. J. Phys. Chem. Lett. 2010, 1, 1442. (32) Jaroniec, M.; Solovyov, L. A. Langmuir 2006, 22, 6757. (33) Shin, T.; Findenegg, G. H.; Brandt, A. Prog. Colloid Polym. Sci. 2006, 133, 116. (34) Glatter, O.; Kratky, O. Small-angle X-ray scattering; Academic Press: New York, 1983. (35) Teubner, M.; Strey, R. J. Chem. Phys. 1987, 87, 3195. (36) Fratzl, P.; Lebowitz, J. L.; Penrose, O.; Amar, J. Phys. Rev. B 1991, 44, 4794. (37) Chen, S.-H.; Chiang, S.-L.; Strey, R. J. Appl. Crystallogr. 1991, 24, 721. (38) Binder, K.; Fratzl, P. In Materials Science and Technology: Phase Transformations in Materials, 2nd ed.; Kostorz, G., Ed.: Wiley-VCH: Weinheim, Germany, 2001; Vol. 5 (39) Zickler, G. A.; J€ahnert, S.; Wagermaier, W.; Funari, S. S.; Findenegg, G. H.; Paris, O. Phys. Rev. B 2006, 73, 184109. (40) Gu, T.; Zhu, B.-Y. J. Chem. Soc., Faraday Trans. I 1989, 85, 3813. (41) Tiberg, F. J. Chem. Soc., Faraday Trans. 1996, 92, 531. (42) Cecutti, C.; Focher, B.; Perly, B.; Zemb, T. Langmuir 1991, 7, 2580. (43) Dupuy, C.; Auvray, X.; Petipas, C.; Rico-Lattes, I.; Lattes, A. Langmuir 1997, 13, 3965. (44) Matsson, M. K.; Kronberg, B.; Claesson, P. M. Langmuir 2004, 20, 4051. (45) Drummond, C. J.; Warr, G. G.; Grieser, F.; Ninham, B. W.; Evans, D. F. J. Phys. Chem. 1985, 89, 2103. (46) Arai, N.; Yasuoka, K.; Zeng, X. C. J. Am. Chem. Soc. 2008, 130, 7916.

5263

dx.doi.org/10.1021/la200333q |Langmuir 2011, 27, 5252–5263