Tuning the Viscoelasticity of Nonionic Wormlike Micelles with β

May 17, 2013 - (1) (2)where ω is the oscillation angular frequency, G′ and G″ the storage and loss modulus, respectively. Gk and τk are the G0 a...
0 downloads 0 Views 4MB Size
Article pubs.acs.org/Langmuir

Tuning the Viscoelasticity of Nonionic Wormlike Micelles with β‑Cyclodextrin Derivatives: A Highly Discriminative Process Marcelo A. da Silva,† Evelyne Weinzaepfel,† Hala Afifi,† Jonny Eriksson,‡ Isabelle Grillo,§ Margarita Valero,∥ and Cécile A. Dreiss†,* †

Institute of Pharmaceutical Science, King’s College London, Franklin-Wilkins Building, 150 Stamford Street, London SE1 9NH, United Kingdom ‡ Department of Chemistry−BMC, Uppsala University, SE-751 23 Uppsala, Sweden § Institut Laue-Langevin (ILL), DS/LSS, 6, rue Jules Horowitz, B.P. 156, 38042 Grenoble Cedex, France ∥ Departamento de Química Física, Universidad de Salamanca, 37008 Salamanca, Spain S Supporting Information *

ABSTRACT: We report the influence of five β-cyclodextrin (β-CD) derivatives, namely: randomly methylated β-cyclodextrin (MBCD), heptakis (2,6-di-O-methyl)-β-cyclodextrin (DIMEB), heptakis (2,3,6-tri-O-methyl)-β-cyclodextrin (TRIMEB), 2-hydroxyethyl-β-cyclodextrin (HEBCD) and 2hydroxypropyl-β-cyclodextrin (HPBCD), on the self-assembly of mixtures of nonionic surfactants: polyoxyethylene cholesteryl ether (ChEO10) and monocaprylin (MCL). Mixtures of ChEO10/MCL in water form highly viscoelastic wormlike micelle solutions (WLM) over a range of concentrations; herein, the composition was fixed at 10 wt % ChEO10/3 wt % MCL. The addition of methylated β-CDs (MBCD, DIMEB, TRIMEB) induced a substantial disruption of the solid-like viscoelastic behavior, as shown from a loss of the Maxwell behavior, a large reduction in G′ and G″ in oscillatory frequency-sweep measurements, and a drop of the viscosity. The disruption increased with the degree of substitution, following: MBCD < DIMEB < TRIMEB. Cryo-TEM images confirmed a loss of the WLM networks, revealing short rods and disc-like aggregates, which were corroborated by small-angle neutron scattering (SANS) measurements. Critical aggregation concentrations (CAC), measured by fluorescence spectroscopy, increased in the presence of DIMEB for both ChEO10 and MCL, suggesting the existence of interactions between methylated β-CDs and both surfactants involved in WLM formation. Instead, hydroxyl-β-CDs had a very different effect on the WLM. HPBCD only slightly reduced the solid-like behavior, without suppressing it. Quite remarkably, the addition of HEBCD reinforced the solid-like characteristics and increased the viscosity 10-fold. Cryo-TEM images confirmed the subsistence of WLM in ChEO10/MCL/HEBCD solutions, while SANS data revealed a slight elongation and thickening of the worms, and an increase of associated water molecules. CAC data showed that HPBCD had little effect on either surfactant, while HEBCD strongly affected the CAC of MCL and only slightly affected the ChEO10. For both DIMEB and HEBCD, time-resolved SANS measurements showed that morphology changes underlying these macroscopic changes occur in less than 100 ms.



INTRODUCTION

the ability to form inclusion complexes (ICs) through noncovalent interactions with a variety of molecular guests that can fit either fully or partially inside their cavity, including drugs,7,8 surfactants,9−12 and polymers.1,6 β-CDs in particular are widely used as solubilizing agents in the pharmaceutical field; they are well tolerated in humans without toxic sideeffects and are present in several injectable drug formulations approved by the FDA.7,8 A substantial effort has been devoted to modifying native cyclodextrins to improve their properties, in particular solubility, by controlling the nature of the

Over recent decades, substantial interest in soft matter has been dedicated to building functional supramolecular structures, based on molecular recognition, host guest chemistry and other types of noncovalent interactions.1,2 Intricate supramolecular structures can be crafted and tuned through noncovalent interactions, leading to the design of “smart” materials, which have the ability to switch their behavior in response to environmental conditions. Cyclodextrins, cyclic oligosaccharides consisting of six, seven, or eight glucopyranose units (α-, β-, and γ-CD),3 have been widely employed as molecular building blocks to create functional and responsive supramolecular structures.1,4−6 Their toroidal shape and hydrophobic cavity imparts them © XXXX American Chemical Society

Received: April 24, 2013 Revised: May 17, 2013

A

dx.doi.org/10.1021/la4015338 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

Scheme 1. Structure of the Surfactants Used in This Study and Repeating Unit of the β-CD Derivatives

also more relevant to commercial and technological products, which usually contain a multitude of components.25 Host− guest chemistry with cyclodextrins offers an additional handle to tune aggregate morphology by selective binding to one of the surfactants, either driven by higher affinity (which may be due to a better geometrical fit) or, in the case of charged systems, to favor an electroneutral stoichiometry ratio, thus leading to the removal of one of the components from the aggregates.15,26 While in single-surfactant systems, the addition of β-CD is expected to weaken or break the aggregates,9 in mixed surfactant systems, the selective removal of one surfactant can be exploited to adjust the geometry of the aggregates.15 It is well documented in the literaturealthough not well understoodthat differently substituted cyclodextrins exhibit distinct behaviors.27−30 For instance, the complexation of bile salts with a range of methylated β-CDs revealed that methylation in position O3 (in heptakis (2,3,6-di-O-methyl)-βcyclodextrin or TRIMEB, Scheme 1) hinders complexation by partially blocking the cavity entrance, while methyl groups in O2 (in heptakis (2,6-di-O-methyl)-β-cyclodextrin or DIMEB, Scheme 1) promotes complexation by extending the hydrophobic cavity.27 Methylated α- and β-CD adsorb on the surface of decanoate micelles, while the corresponding unsubstituted “parent” cyclodextrins form host−guest complexes with the decanoate ions in solution.28 We also previously reported a similarly discriminative behavior with Pluronic micelles,29,30 where doubly methylated β-CD (DIMEB) induced an instantaneous rupture of the polymeric micelles, while othersome closely relatedderivatives (TRIMEB, 2-hydroxyethyl-β-CD, and 2-hydroxypropyl-β-CD) had little to no effect on the micelles. Interestingly in these systems, both NMR and time-resolved SANS30,31 ruled out a host−guest interaction mechanism and suggest instead the establishment of weak, nonspecific, surface interactions between the cyclodextrin and the polymer chains constituting the micelles.

substituents around the rim and the pattern of substitution, in particular for β-CD, which has a very low solubility (0.0163 mol·L−1), because of a particularly stable crystalline state due to hydrophobic interactions and matching hydrogen bonds.3 A number of studies have focused on the interactions of cyclodextrins with surfactants,10−12 but fewer have focused on their micellar aggregates.10,13,14 In interacting with micellar aggregates, cyclodextrins can act as structure modulators, in the vast majority of cases disrupting the organized assemblies by “snatching” one of the components and forming an inclusion complex with it.10 In a limited number of cases, cyclodextrins have been reported to promote aggregate growth,10,15−17 possibly through the selective removal of the less bound surfactant molecules from the aggregates or other mechanisms. The effect of CD/surfactant complexation on wormlike micelles however has been scarcely investigated,16,17 and the use of cyclodextrins as a handle to tune the viscoelasticity of solutions has not been reported. Wormlike micelles are long, flexible cylindrical aggregates which, above a threshold concentration, entangle into transient viscoelastic networks,18 some showing remarkable solid-like properties reminiscent of gels.14,19−22 It is therefore expected that the most likely outcome of CD/surfactant interaction, i.e., a disruption of the micellar aggregates by complexation, will lead to a collapse of the networks and thus a drop in the rheological properties, thus offering a trigger for a “sol/gel” transition. Interestingly here, we demonstrate for the first time that, depending on the type of substituent on the β-cyclodextrin rim, either a destructive or a constructive effect can be achieved, and that CDs can thus be conveniently used as a handle to control the macroscopic properties. Mixed surfactant systems are of interest because of the polymorphism of self-assembly structures accessible through simple tuning of the composition, and the possible synergy resulting from the combination of properties of the various surfactants in the mixture.23,24 Combinations of surfactants are B

dx.doi.org/10.1021/la4015338 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

Scheme 2. Schematic Representation of the Interaction of DIMEB and HEBCD with ChEO10/MCL Wormlike Micelles, As Inferred from Rheology, SANS, and Cryo-TEMa

a

DIMEB breaks-up the micelles into short aggregates, while HEBCD leads to a reinforcement of the solid-like behavior, possibly due to the association of HEBCD with the worms. hydroxyethyl-β-cyclodextrin (HEBCD), and 2-hydroxypropyl-β-cyclodextrin (HPBCD, Mw = 1460 g·mol−1). The molecular structure of the β-CD derivatives studied is given in Scheme 1. It is expected that the degree of substitution (number of methyl groups per molecule of βcyclodextrin) increases in the following order MBCD < DIMEB < TRIMEB.27 Pyrene was obtained from Sigma Aldrich UK. For the preparation of the samples for SANS and Cryo-TEM experiments D2O was used (Euriso-top, 99.85% purity). For all other experiments, ultrapure water (18.2 MΩ·cm, Millipore-filtered) was used. All materials were used as received. Note that in the following, sometimes the notation “β-CD” is used as a generic term to refer to the β-CD derivatives studied, not to the native (unmodified) β-CD, which was not studied here. Methods. Preparation of the Solutions. For the rheology and SANS measurements, the surfactant solutions were prepared by mixing appropriate amounts of stock solutions of ChEO10 (20 wt %) with MCL as a solid and β-cyclodextrin stock solution (25 wt %). MCL is expected to form bilayers in aqueous solution,35,36 while ChEO10 is known to form micelles .32,33 All surfactant solutions contained 10 wt % ChEO10 and 3 wt %, with 3 to 9 wt % β-CD. For the SANS experiments, the samples were made in D2O instead of water to provide contrast and diluted 10-fold to avoid structural peaks caused by intermicellar interactions.33 The exact same samples were used for Cryo-TEM measurements. Rheological Measurements. Small-amplitude oscillatory and steady-shear experiments were performed on a dynamic straincontrolled rheometer ARES (TA Instruments) using cone-and-plate geometry (50 mm, 0.02 rad), with a temperature-controlling Peltier unit and a sample cover to minimize evaporation. Experiments were run at 20.0 °C. All solutions were left to rest at least 7 days at room temperature. They were then vortex-mixed and equilibrated at 20.0 °C (in a water-bath) for 24 h before conducting the rheological measurements. After loading, samples were allowed to rest for a few minutes before the start of the experiments to ensure dissipation of any preshearing due to manipulation and loading. Measurements were carried out in duplicate or triplicate, with very good reproducibility. The results reported in this work are examples of typical data obtained, not averages. Oscillatory frequency sweep tests covering the range of 100 to 0.1 rad·s−1 were performed in the linear viscoelastic regimestrain of 5%as determined by dynamic strain sweep measurements. Steadyshear rate flow curves were acquired covering the range of 0.01 to

In this contribution, we focus on nonionic wormlike micelles formed by adding a small headgroup surfactant (monocaprylin, MCL) to polyoxyethylene cholesteryl ether (ChEO10) (Scheme 1).32−34 To this viscoelastic phase at a fixed composition, we add a range of related β-cyclodextrin derivatives: randomly methylated β-cyclodextrin (MBCD), heptakis (2,6-di-O-methyl)-β-cyclodextrin (DIMEB), heptakis (2,3,6-tri-O-methyl)-βcyclodextrin (TRIMEB), 2-hydroxyethyl-β-cyclodextrin (HEBCD), and 2-hydroxypropyl-β-cyclodextrin (HPBCD). The complexation of β-cyclodextrin with either the main surfactant (ChEO10) or the cosurfactant (MCL) is expected to reduce the solid-like character of the solutions, as the removal of any component induces a shortening of the WLM disrupting the entangled network. Instead, we observe a behavior which is critically dependent on the nature of the β-cyclodextrin substituentsas reported also for a different system29−31 inducing either a slight weakening, a complete loss of the solidlike behavior or a reinforcement, as illustrated in Scheme 2 for two extreme scenarios. We find that the origin of these effects cannot be fully explained by the formation of CD/surfactant inclusion complexes (leading to the extraction of one or both surfactants from the micelles), but that other types of mechanisms have to be considered. We use rheology, SANS, and Cryo-TEM to characterize the macroscopic behavior and the nanoscale morphology of the systems over a range of compositions, and time-resolved SANS to monitor the evolution of the mixtures over time. To further examine the mechanism of interaction, the effect of cyclodextrins on the surfactants aggregation behavior is monitored by fluorescence spectroscopy, using pyrene as a probe.



EXPERIMENTAL SECTION

Materials. Polyoxyethylene cholesteryl ether (ChEO10) was purchased from Ikeda Corporation, Yokohama, Japan. The cosurfactant monocaprylin (MCL) was obtained from Sigma Aldrich UK (Scheme 1). All β-cyclodextrin derivatives were also obtained from Sigma Aldrich UK: randomly methylated β-cyclodextrin (MBCD), heptakis (2,6-di-O-methyl)-β-cyclodextrin (DIMEB, FW = 1331.36 g·mol−1), heptakis (2,3,6-tri-O-methyl)-β-cyclodextrin (TRIMEB), 2C

dx.doi.org/10.1021/la4015338 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

1000 s−1 in two steps (0.01 to 1 and 1 to 1000 s−1) in order to optimize acquisition time. Zero-shear viscosity (η0) was obtained by extrapolation from the flow curves. Values of the relaxation time τR and plateau modulus G0 were obtained by fitting the frequency sweeps with a multi-element Maxwell model where possible.

G′(ω) =

∑ Gk k

G″(ω) =

∑ Gk k

tn = an − 1t1,

TAM1 =

(1)

1+

(2)

where ω is the oscillation angular frequency, G′ and G″ the storage and loss modulus, respectively. Gk and τk are the G0 and τR for the kth element, respectively. A “global” G0 is defined as follows: G0 =

∑ Gk k

(4)

t1 + t DT 2

and

TAMn =

Tn − 1 + Tn + t DT 2

(5)

with tDT being the dead time needed for the filling of the cell estimated at 70 ms for the flow rate used. The stock solutions of wormlike micelles (surfactant mixtures) and various amounts of either DIMEB or HEBCD were prepared by weighting the appropriate amounts of surfactant, β-CD, and deuterated water. Appropriate volumes of stock solutions (total 804 μL) were then mixed in the stopped-flow cell with a flow rate of 3 mL·s−1, to obtain the target concentrations of 1 wt % ChEO10/0.3 wt % MCL and 0.3−0.6 wt % β-CD. SANS Data Fitting. SANS data were fitted using a model for a cylinder with an elliptical cross-section, where L is the length of the cylinder and rminor and rmajor are the minor and major radii, respectively. The following function is calculated:40

ωτk ω2τk2

1 − an t1 1−a

Tn is the accumulated time after mixing, with t1 = 100 ms and a = 1.1. Sixty-eight frames were measured for a total time of 651.7 s (ca. 11 min). The average time after mixing TAMn for the nth run is as follows:

ω2τk2 1 + ω2τk2

Tn =

(3)

The fits were performed using the nonlinear fitting tool from Microcal Origin 6.0 software. Initially, a set of parameters were obtained for the G′ curves, this set was then used as an input for fitting G″. Subsequently, the results were used to generate a new fit for G′ data; this was repeated until a fit of good quality could be obtained for both G′ and G″. The purpose of the fitting with multiple elements is to provide a more quantitative interpretation of the rheological data. The number of elements chosen was based solely on the minimal number needed to obtain a good fit and are not attributed a physical meaning. However, they give a useful measure of the departure from the pure Maxwell behavior.32 Cryogenic Transmission Electron Microscopy (Cryo-TEM). CryoTEM measurements were performed with a Zeiss TEM 902A instrument (Carl Zeiss NTS, Oberkochen, Germany). The instrument was operated at 80 kV in zero-loss bright-field mode. Digital images were recorded under low dose conditions with a BioVision Pro-SM Slow Scan CCD camera system Proscan elektronische Systeme GmbH, Germany and iTEM software (Olympus Soft Imaging Solutions GmbH, Münster, Germany). An underfocus of 1−3 μm was used in order to enhance contrast. The preparation procedure has been described in detail elsewhere.37 Specimens for examination were prepared in a climate chamber with temperature and humidity control (temperature of 25 °C and relative humidity of approximately 98− 99%). Thin films of sample solution were prepared by placing a small drop of the sample on a copper grid supported perforated polymer film, covered with thin carbon layers on both sides. After the drop was blotted with filter paper, thin sample films (10−500 nm) spanned the holes in the polymer film. Immediately after blotting, the samples were vitrified by plunging them into liquid ethane, held just above its freezing point. Samples were kept below −165 °C and protected against atmospheric conditions during both transfer to the TEM and examination. Several images were taken of each sample studied, which were all consistent, and representative examples are presented here. Small-Angle Neutron Scattering. Static SANS and time-resolved SANS (TR-SANS) measurements were carried out mainly on the D22 instrument at the ILL (Grenoble, France) and SANS2D at ISIS (Didcot, UK) for some additional measurements (SI3 of the Supporting Information, SI). The wavelength λ was set at 6 Å, the peak flux of the cold source. The sample-to-detector distance was 4 m with a collimation at 5.6 m and a detector offset of 400 mm to maximize the available q-range (1.2 × 10−2 Å−1 < q < 0.26 Å−1). A 7 × 10 mm2 sample aperture was used. Raw data were corrected for electronic background, empty cell and normalized by water using the new Lamp (Large Array Manipulation) ILL software developed for SANS data treatment.38 Kinetic SANS measurements were carried out using the stoppedflow technique;39 the sample path length in the Biologic SFM-300 stopped-flow apparatus was 1 mm. The scattering measurements were made with an acquisition time for each frame n of tn following a geometric series during 10 min:

I(q) =

scale Vcyl

∫ d Ψ ∫ dφ ∫ p(θ , φ , Ψ)F 2(q , α , Ψ) sin θdθ

+ bkg

(6)

with the functions:

F(q , α , Ψ) = 2

J1(a) a



sin(b) b

1/2 a = q − sin(α)⎡⎣rmajor 2 sin 2 Ψ + rminor 2 cos2 Ψ⎤⎦

b=q

L cos(α) 2

(7) (8) (9)

where the “scale” is a scaling factor containing information such as volume fraction and contrast, the angles θ and Φ define the orientation of the axis of the cylinder and angle ψ is the orientation of the major axis of the ellipse with respect to the vector q. Data modeling used the routines from the NIST SANS group encoded within the SansView program from the DANSE project.40 Fitted curves included numerical smearing to approximate instrumental q resolution and also a flat background to allow for residual incoherent scattering not present in the solvent background cell subtraction. SANS data were also fitted with a rod (or disc) model. For N randomly oriented rods of length L and radius R, the form factor P(q) is given by the following:

P(q) = N

∫o

π /2

F 2(q)sin(γ )dγ

(10)

where F(q) = (Δρ)V((sin(1/2qL cos γ)/1/2qL cos γ)/(2J1(qR sin γ)/ qR sin γ)) in which J1 is the Bessel function of first order, and γ is the angle between the q vector and the axis of the rod. In this model, water molecules associated with the ethylene oxide head of ChEO10 were taken into account by varying the scattering length density of the micelles (i.e., by calculating a weighed scattering length density of the surfactant, DIMEB and D2O). For the rod/disc fittings, the program FISH was used.41 FISH uses standard iterative least-squares fitting in which selected parameters of the chosen model can be refined to optimize the fit. Parameters were refined from several starting points to ensure that a global (rather than a local) minimum had been found. In all fits, the “scale” parameter was left to float, and the value returned by the fit checked against its calculated value to confirm consistency of the fit (an error of ±10% was deemed acceptable). Scattering length densities of 4.19 × 10−7 Å−2 for ChEO10 (bulk density 1.065 g·cm−3), 4.05 × 10−7 Å−2 for MCL (density of 1.044 D

dx.doi.org/10.1021/la4015338 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

g.cm−3), 11.9 × 10−7 Å−2 (density of 1.25 g·cm−3), and 13.6 × 10−7 Å−2 (density of 1.31 g·cm−3) for DIMEB and HEBCD, respectively, were calculated. Fluorescence Measurements. A stock solution of pyrene in acetone (1.71 × 10−2 mol·L−1) was initially prepared. An aliquot of 35 μL of this solution was placed in a 100 mL volumetric flask, and the solvent was evaporated to air. The residue was then dissolved in either pure water or a β-CD (HPBCD, HEBCD, or DIMEB) aqueous stock solution (1.0 w/v%, i.e., ∼7 × 10−3 mol·L−1), resulting in a final concentration of pyrene [pyr] = 5.99 × 10−6 mol·L−1. These solutions were then subsequently used as the solvent for the surfactant solutions. Stocks solutions of ChEO10 (3.0%) and MCL (3.4%) in water were prepared. An appropriate aliquot of these solutions were dissolved in the pyrene/H2O or pyrene/β-CD/H2O solutions. Solutions of different surfactant concentration were obtained by diluting the concentrated surfactant solution with the appropriate solvent. The fluorescence emission spectra were recorded on a Cary Eclipse fluorescence spectrophotometer (Varian, Oxford, UK) with λexc = 340 nm. Fluorescence intensities at 373, 384, 393 nm and, when it appears, also at the excimer band centered at 490 nm, were measured. For each surfactant, the critical aggregation concentration value was determined both in the absence and presence of HEBCD, HPBCD, and DIMEB by using the intensity of the most sensitive peak. At least two repeats were performed for each sample.

Figure 2. Oscillatory frequency sweep measurements showing the storage and loss moduli G′ and G″ as a function of frequency ω for mixtures of 10 wt % ChEO10/ 3 wt % MCL (black ⬟) in the presence of 6 wt % of the following β-CD derivatives: MBCD (pink ■), DIMEB (red ⧫), TRIMEB (purple ⬢), HEBCD (green ▲), and HPBCD (blue ●) (G′, filled symbols, G″, open symbols). Solid lines show fits to the Maxwell model.



drop in viscosity, the disappearance of the Newtonian plateau in the flow curves (Figure 1), the loss of the Maxwell behavior and drop in G′ and G″ values in the dynamic measurements (Figure 2). The addition of hydroxyethyl and hydroxypropyl derivatives has a different effect. With HPBCD, a small drop in all rheological parameters is measured, but the overall solid-like viscoelastic behavior is preserved. The magnitude of the drop is dependent on HPBCD concentration (Table 1), but even at

RESULTS Macroscopic Behavior. Mixtures of ChEO10/MCL in water form highly viscoelastic wormlike micellar solutions over a range of concentrations.33 Steady-state measurements of 10 wt % ChEO10/3 wt % MCL show a typical shear-thinning behavior (Figure 1), with a well-defined Newtonian plateau up

Table 1. Parameters Derived from a Maxwell Fit to the Frequency Sweep Curves Presented in Figure 3 for ChEO10/ MCL Wormlike Micelles (10 wt %/3 wt %) in the Presence of HEBCD and HPBCD τR (s)

G0 (Pa)

Figure 1. Flow curve showing the viscosity η as a function of shear rate γ̇ for 10 wt % mixtures of ChEO10 with 3 wt % MCL (black ⬟) in the presence of 6 wt % of the following β-CD derivatives: MBCD (pink ■), DIMEB (red ⧫), TRIMEB (purple ⬢), HEBCD (green ▲), and HPBCD (blue ●).

samples (wt %)

G01

G02

τR1

τR2

η0 (Pa·s)

pure WLM 3% HEBCD 6% HEBCD 9% HEBCD 6% HPBCD 9% HPBCD

66 51 40 22 33 15

16 15 7.1 20 5.8

0.054 0.18 0.37 0.58 0.10 0.016

0.021 0.029 0.090 0.007 0.10

4.5 13 20 19 3.0 1.3

the highest concentration studied (9 wt % HPBCD, Table 1), the weakening of the properties is not comparable to the one observed with methyl-derivatives, neither in the magnitude, nor in the nature of the change. With the addition of HEBCD, surprisingly, an increase in the viscoelastic properties is observed. A closer analysis of the flow curves (Figure 1) reveals for both HPBCD and HEBCD a reduction of the critical shear rate (γ̇C), the shear rate that defines the limit of the Newtonian plateau. γ̇C drops from 8 to 3 and 1 s−1, from pure WLM to WLM/HPBCD and WLM/HEBCD solutions, respectively (with 6% β-CD added). η0, instead, varies from 4.2 to 3.0 and 20 Pa·s from pure WLM to WLM/HPBCD and WLM/HEBCD solutions, respectively. Instead, with the addition of the methylated β-CDs, the Newtonian plateau is suppressed, and the samples studied are well within the shearthinning range even at the lowest shear rates measured. The weakening of the viscoelastic properties upon adding β-CD increases in the order: MBCD < DIMEB < TRIMEB, both from steady-state and oscillatory measurements, suggesting a

to γ̇ = 8 s−1 and a zero-shear viscosity (η0) of 4.2 Pa·s (Figure 1). Frequency sweeps follow a typical Maxwell behavior with a plateau modulus (G0) of 66 Pa and a single relaxation time (τR) of 0.05 s (Figure 2). This surfactant concentration (10 wt % ChEO10/3 wt % MCL) was kept constant in the remainder of the work. Figures 1 and 2 show the effect of adding 6 wt % of the five types of β-cyclodextrins derivatives studied on the frequency sweep and steady-shear flow curves. From these data, two types of behavior can be distinguished. Methylated β-CDs (MBCD, DIMEB, and TRIMEB) cause the largest changes in the rheological behavior and always in a direction that suggests a disruption of the WLM structure, as inferred from the large E

dx.doi.org/10.1021/la4015338 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

Figure 3. Influence of hydroxyethyl-β-CD (HEBCD) concentration on the rheological response of 10 wt % ChEO10/3 wt % MCL solutions. HEBCD concentrations of 0 (black ■), 3 (light green ●), 6 (green ⧫), and 9 wt % (dark green ▲). (A) Oscillatory frequency sweeps (G′, filled symbols; G″, open symbols). Solid lines represent fits to the Maxwell model. (B) Steady-shear rate flow curves.

The addition of HEBCD to ChEO10/MCL induces a deviation from the expected behavior for a single-element Maxwell model, which can be explained by various factors. It is common to observe WLM relaxation processes to be dominated by a single relaxation time and to follow the Maxwell model. It is generally accepted that WLM relaxation is due mainly to two processes (within the frequency range studied): reptation and reversible chain-scission.42 The reptation process depends strongly on the aggregate length,45,46 and due to the inherent polydispersity of the micellar aggregates, a distribution of relaxation times would be expected. However, due to the transient nature of the micellar aggregates, the continuous breakage and reformation of the micelles adds a second process for the system to achieve relaxed conformations. Cates proposed42 that for WLM systems where the micellar breaking time (τbreak) is very small in comparison to the reptation time (τrep) (i.e., for τbreak ≪ τrep) the whole relaxation process is dominated by a single relaxation time τR = (τbreak + τrep)1/2. Outside this condition, the reptation becomes more relevant and a distribution of relaxation times is expected. Thus, one explanation for the deviation could be a change in the micellar dynamics increasing the micelles lifetime, which would result in a longer τbreak. Another possibility is the appearance of a new relaxation process, other than chainscission and reptation. For mixtures of WLM and telechelic polymers, this effect has been observed, where the addition of a polymer leads to an extra relaxation process involving the polymer dissociation from the WLM aggregates.47 A similar situation could arise due to an inclusion complex formation with β-CD. In order to clarify which one of the suggested possibilities could apply to this particular case, a master curve based on the data presented in Figure 3B was constructed, as described in previous studies,32,48 and is presented in Figure 4. The rationale behind creating a master curve based on an HEBCD concentration shift is that it should be possible to obtain the same value of any viscoelastic parameter by either a change in time or physicochemical state, as long as the change uniformly affects all of the relaxation times, i.e., that there is no addition or removal of relaxation processes.32,49 Thus, the existence of a master curve implies that no new relaxation processes have been added to the system with the addition of HEBCD, and therefore the changes are only attributable to changes in micellar breakage/reformation kinetics. Microstructure. Cryo-TEM. Cryo-TEM images of ChEO10/ MCL wormlike micelles have been published elsewhere33 and confirmed the presence of long wormlike micelles. Figure 5

correlation between the number of methyl-substituted groups on the β-CD rim, and their disruptive effect. Therefore, two types of rheological behavior can be distinguished: a Maxwell-like behavior, for the pure WLM solutions and the solutions with HPBCD and HEBCD, and a non-Maxwell-like behavior for solutions with methylated derivatives, where both G′ and G″ values show an inverse dependence on the number of methyl substituents (Scheme 2). For solutions containing β-CD hydroxyl derivatives, however, it was not possible to fit the frequency sweeps with a singleelement Maxwell model. The addition of a second element improved the quality of the fits, in particular for the HEBCD systems; additional elements (>2) did not further improve the fits. Table 1 shows the results of the fits for pure WLM and the systems where HPBCD and HEBCD were added. Comparing the longest relaxation time for these systems, we observe that the relaxation time increases progressively from pure WLM to HPBCD to HEBCD solutions. The global G0 is slightly lower for the systems with β-CD: 53 and 55 Pa for HEBCD and HPBCD, respectively, against 66 Pa for the pure WLM, and is not very sensitive to the type of hydroxyl-β-CD. In Figure 3, the effect of HEBCD concentration (3 to 9 wt %) on the rheological behavior of ChEO10/MCL solutions is presented. The effect observed is stoichiometric: by adding more β-CD, the effect gradually increases, for the longest relaxation time almost linearly, and up to 6 wt % for the zeroshear viscosity (Tables 1 and SI1 of the SI), suggesting an increase of the micellar length.18,42,43 However, interestingly, the value of the plateau modulus G0 gradually decreases (SI1 of the SI, Table 1), pointing to a decrease in the entanglement density, as follows:44 G0 = vkT

(11)

where v is the density of entanglements points, k the Boltzmann constant, and T the temperature. For HPBCD (Table 1), all of the rheological parameters decrease continuously with increasing HPBCD concentration, suggesting a shortening of the micellar aggregates.18,42,43 For the methylated β-CDs, the effect is much more drastic, and it is no longer possible to measure G0, τR, or η0. However, the viscosity curves are shifted downward, and G″ decreases with a gradual increase in DIMEB from 0 to 9 wt % over the whole range of frequencies measured (SI2 of the SI). This suggests major changes of the micellar morphology, possibly a reduction in aggregate size leading to a loss of the entangled network, as suggested by the drastic reduction of both G′ and τR. F

dx.doi.org/10.1021/la4015338 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

scattering from ChEO10 originates from finite slightly elongated objects,32,33 as observed from the presence of a near-plateau at low q, and more elongated structures in the presence of MCL, as visible from the q−1 behavior at low q.33 With the addition of DIMEB, a clear departure from the rod-like form factor is observed, with a substantial decrease of the intensity (Figure 6A): I(0) drops by over 5-fold with 6 wt % DIMEB, to ∼0.7 cm−1, and shows a plateau at low q while the main oscillation of the form factor is shifted to higher q-values; both changes suggest a transition to finite, substantially smaller structures. Interestingly, at the intermediate concentration of 3 wt % DIMEB, the pattern substantially differs both from the initial micellar form factor and the final one. The scattering curve follows a q−2 dependence at low q, thus departing from the elongated rod shape (with q−1 dependence) and hinting to a flat bilayer; in addition, the samples are slightly turbid, thus suggesting phase separation or the presence of lamellar structures. Additional measurements on a wider range of compositions (SI3 of the SI) indeed show that mixtures of ChEO10 and DIMEB induce the formation of lamellar structures at high surfactant/CD ratios, as visible from the presence of a clear Bragg peak in the scattering pattern (SI3 of the SI shows data for 5 wt % ChEO10 and a ∼5-fold ChEO10/ DIMEB ratio compared to the present study). The pure ChEO10 micelles were fitted by a cylinder model with an elliptical cross-section, as previously reported,33 with a radius of 37 Å, axial ratio of 2.1, and length of ca. 160 Å. With the addition of 6 wt % DIMEB (Figure 6A), the scattering curve could be fitted equally well with a sphere or rod model, yielding a small radius of ca. 14 Å and a length of ∼20 Å (for the rods), thus confirming that the addition of DIMEB induces a break-up of the large micellar structures into smaller scattering objects. The ChEO10/MCL wormlike micelles alone (in the absence of DIMEB) can also be fitted to an elliptical cylinder model with a radius of 29 Å, axial ratio of 1.5 and a much more extended length of ca. 400 Å (Figure 6C). With the addition of DIMEB, a similar behavior to the one observed with pure ChEO10 micelles is obtained: at 3 wt % DIMEB, a slight decrease in the scattering intensity with a q−2 dependence at low q occurs. Without additional information on the sample, we did not attempt to fit the curves, as various models could equally be used; it is quite possible that a mixture of structures is present at this intermediate concentration. Increasing DIMEB concentration to 6 wt % results in a substantial decrease in the scattering intensity and a near plateau at low q, reflecting a drop in size. The plot was fitted using a model of small rods or spheres, showing again the presence of small aggregates of similar dimensions to those obtained with ChEO10 micelles and 6 wt % DIMEB (rods with a radius of ca. 13.7 Å and length of 17.0 Å). The presence of small discshaped aggregates in WLM + 6 wt % DIMEB solution corroborates the morphologies detected in the Cryo-TEM images. In contrast, the addition of HEBCD leads to limited changes in the scattering patterns: a slight decrease of the intensity for the pure ChEO10 micelles, and slight changes at intermediate q for ChEO10/MCL wormlike micelles. The scattering curves of pure ChEO10 micelles in the presence of 3 and 6 wt % HEBCD could be fitted using the ellipsoidal cylinder model, revealing a slight decrease in micellar radii (31 Å instead of 37 Å for pure micelles), but no change in the rods length (Figure 6B). For the WLM, the addition of HEBCD leads to an increase in the

Figure 4. Master curve obtained from shifting the frequency sweep curves from Figure 3a. The master curve was obtained by dividing all frequencies by the frequency value at the crossover point (G′ = G″), and equally dividing all the moduli by the respective value of the moduli at the crossover point. HEBCD concentrations of the following: (light green ●) 3, (green ⧫) 6, and (dark green ▲) 9 wt %.

Figure 5. Cryo-TEM images of 10 wt % ChEO10/3 wt % MCL wormlike micelles with 6 wt % HEBCD (left) and 6 wt % DIMEB (right). Arrows in the right image indicate disc-like structures in different views: face-on-angle (black) and edge-on-angle (white). The scale bars represent 100 nm.

shows the effect of adding HEBCD (left) and DIMEB (right). In the presence of HEBCD, elongated structures are observed and no major structural change is detected, compared to the original wormlike solutions.33 The addition of DIMEB instead induces a remarkable shortening of the worms and also suggests the presence of other morphologies, including discs. Therefore, the drop in viscoelastic behavior induced by adding DIMEB can clearly be ascribed to a change in the packing of the surfactant in the aggregates, leading to a transition to shorter worms and possibly other types of smaller self-assembled structures, clearly reducing or suppressing entanglements. No obvious change in WLM structure can be deduced from the images of HEBCD/ WLM to explain the reinforcement measured by rheology; the structure of the worms seems to overall be preserved, but this does not exclude either a lengthening or thickening of the worms, which would be difficult to detect in these images, or the appearance of other types of relaxation mechanisms (which, however, was ruled out by the construction of the master curve). Small-Angle Neutron Scattering Measurements. Smallangle neutron scattering measurements were performed both on pure ChEO10 micelles and their mixtures with MCL in D2O. In order to avoid the effect of intermicellar interactions, the solutions were diluted 10 times (to 1 wt % ChEO10), but for simplicity and to link with the rheology, the text refers to the original concentrations. Figure 6 shows the scattering patterns from the pure micelles and in the presence of 3 and 6 wt % DIMEB and HEBCD. In the absence of cyclodextrin, the G

dx.doi.org/10.1021/la4015338 | Langmuir XXXX, XXX, XXX−XXX

Langmuir

Article

Figure 6. SANS experimental curves from 10-fold dilutions of the following: (A) 10 wt % ChEO10 micellar solutions (black □) with 3 wt % (red ○) and 6 wt % DIMEB (red ◊). (B) 10 wt% ChEO10 micellar solutions (black □) with 3 wt % (light green ○) and 6 wt % HEBCD (green ◊). (C) Wormlike micelles (10 wt % ChEO10/3 wt % MCL) (black □) with 3 wt % (red ○) and 6 wt % DIMEB (red ◊). (D) Wormlike micelles (10 wt % ChEO10/3 wt % MCL) (black □) with 3 wt % (light green ○) and 6 wt % HEBCD (green ◊). Fits to the elliptical rod and rod models are represented by solid lines.

Figure 7. Time-resolved small-angle neutron scattering data from 10% ChEO10/3% MCL wormlike micelles (■), and in the presence of A: 3 wt % DIMEB at different time points after mixing (0.1 s, pink ▲; 1.5 s, pink □; 622.1 s, red ∗); B: 6 wt % DIMEB (0.1 s, light pink ▲; 1.5 s, light pink □; 622.1 s, light pink ∗) and C: 6 wt % HEBCD (0.1 s, light green ▲; 1.5 s, green □; 622.1 s, dark green ∗). All solutions are diluted 10-fold.

origin of the reinforcement observed in the rheology measurements (Scheme 2). Further Clues on the Mechanism of Interaction. Kinetics of Structural Rearrangements. In order to bring further insight into the nature of the morphological changes, and in particular the rate at which they occur, time-resolved SANS were performed by adding either DIMEB or HEBCD to the worms and measuring the subsequent changes in scattering patterns over time, using a stopped-flow equipment (Figure 7). Quite remarkably, with both β-CD derivatives, the first frame (