Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

View Article Online / Journal Homepage / Table of Contents for this issue

Soft Matter Dynamic Article Links < C

Cite this: Soft Matter, 2012, 8, 11478


www.rsc.org/softmatter PAPER
Nanoscale structure of surfactant-induced nanoparticle monolayers at the
oil–water interface†
Davide C. E. Calzolari,‡a Diego Pontoni,*a Moshe Deutsch,b Harald Reichertac and Jean Daillantd
Published on 20 September 2012. Downloaded by Open University on 18/06/2013 04:09:37.

Received 30th June 2012, Accepted 20th August 2012


DOI: 10.1039/c2sm26520f

Water-dispersed silica nanoparticles (NPs) do not adsorb to the interface between immiscible bulks of
water and hexane. Adding, however, a surfactant (cetyltrimethylammonium bromide, CTAB) to water
induces the formation of a NP monolayer (ML) at this model liquid–liquid interface. We determined
the ML’s structure in situ at deeply buried planar water–hexane interfaces with sub-nanometer
resolution by high energy X-ray reflectivity. Detailed modeling of the data yields the NPs’ interfacial
concentration and water immersion depth, allowing calculation of the NP average contact angle Q. At a
CTAB concentration fc ¼ 0.75 mM, comparable to the critical micelle concentration, a dilute NP
monolayer is found with Q  128 . At lower surfactant concentration (fc ¼ 0.05 mM) we find a dense
ML of close-packed NPs with an unexpectedly high Q  146 . The structure and adsorption scenarios
of the NPs at the interface are discussed, highlighting the relevance of the present method for
quantitative studies of a broad range of systems and applications.

1 Introduction neglected in the standard DE calculation.14 The latter yields


(24  10)kT and (5  2)kT for the high and low fc systems,
Self-assembly of nano-particles (NPs) at liquid–liquid interfaces respectively, thus highlighting the need for more advanced
is of fundamental scientific interest1–3 and has a great techno- interfacial binding theories, particularly in the case of dense MLs
logical impact.4,5 It is, therefore, intensely studied both theoret- of interacting nano-sized particles.
ically6–10 and experimentally.11–13 Accurate measurements of the
NP’s contact angle Q (Fig. 1(a)) and binding energy DE14 are very
challenging, particularly for NPs residing at deeply buried
interfaces between immiscible liquids. We report Q measure-
ments and classical DE calculations for silica NPs adsorbed at a
flat interface between hexane and aqueous solutions of cetyl-
trimethylammonium bromide (CTAB) surfactant. The interfa-
cial structure is determined with Angstr€  om resolution by
high-energy X-ray reflectivity (XRR) for two CTAB concentra-
tions, fc, well below and near the critical micelle concentration,
fcmc. At a low fc (0.05 mM) we find a dense ML of close-packed
NPs with Q ¼ (146  4) , while a higher fc (0.75 mM) yields a
less dense ML with lower Q ¼ (128  6) . The NP close-packing
scenario suggests a solid-like character for the low-fc ML, which
is likely stabilized by strong inter-NP interactions that are

a
European Synchrotron Radiation Facility, 6 rue Jules Horowitz, BP220,
38043 Grenoble, France. E-mail: pontoni@esrf.fr
b
Physics Department and Institute of Nanotechnology and Advanced
Materials, Bar-Ilan University, IL-52900 Ramat Gan, Israel Fig. 1 System description and X-ray scattering geometry. (a) Schematic
c
Max-Planck-Institut f€
ur Metallforschung, Heisenbergstr. 3, D-70569 representation of two silica nanoparticles at the buried water–hexane
Stuttgart, Germany
d
interface. The particle’s contact angle Q is obtained from modeling of the
CEA, IRAMIS, LIONS, bat. 125 CEA Saclay, F-91191 Gif-sur-Yvette
X-ray reflectivity data. The other parameters are described in the text. (b)
Cedex, France
† Electronic supplementary information (ESI) available. See DOI: Schematic cross-section of the sample environment (not to scale).
10.1039/c2sm26520f The concentric 2 mm thick glass beakers (dashed black) are centered by a
‡ Present address: Physics Department, Fribourg University, Ch. du 2 mm thick Teflon ring (dashed white). The inner diameter of the external
Musee 3, CH-1700 Fribourg, Switzerland. beaker is 58 mm.

11478 | Soft Matter, 2012, 8, 11478–11483 This journal is ª The Royal Society of Chemistry 2012
View Article Online

The NPs are rendered surface-active by adsorbing surfactant ID02 beamline of the ESRF25 using 12.4 keV X-rays and an
molecules on their surface, thus transforming them from image-intensified CCD detector.26 CTAB–NP adsorption was
hydrophilic to partially hydrophobic. This partial wettability pre-characterized (Fig. 2(c) and (d)) by interfacial tension
modification drives the NPs to the oil–water interface, forming measurements using a Wilhelmy-type tensiometer with a
and stabilizing oil-in-water (o/w) or water-in-oil (w/o) emul- roughened glass plate that ensured complete wetting. The XRR
sions, depending on the degree of particle hydrophobicity. measurements (Fig. 3 and 4) were performed using a High
Extensive studies15–18 of these systems uncovered numerous Energy Micro-Diffractometer27,28 at the ID15A beamline of the
fascinating effects including the spectacular microemulsion ESRF using a 69.8 keV X-ray beam delivering 5  1011
double phase inversion.19,20 We study a typical water-dispersed photons per second in a focal spot of 5  20 mm2 (V  H). The
silica NPs–CTAB–hexane model system to allow a direct sample environment consisted of two concentric cylindrical
Published on 20 September 2012. Downloaded by Open University on 18/06/2013 04:09:37.

comparison with rheology and optical microscopy measure- PYREX beakers separated by a thin Teflon ring (Fig. 1(b)).
ments of the same system,21,22 which suggested that surfactant– The inner beaker (5 cm diameter) was over-filled with the water
NP reorganization occurs at the interface before equilibrium is
reached. Computer simulations23 support these conclusions,
indicating post-adsorption surfactant detachment from the
NPs, causing their expulsion from the interface. Experimental
corroboration of these novel insights requires direct determi-
nation of the layer’s nano-scale structure, and, in particular, the
NPs’ immersion depth, hav (Fig. 1(a)), Q and DE. For non-
interacting particles of known size, DE is obtainable from the
macroscopic interface tension, gow,24 and yields hav and Q. The
resultant hav values were verified by direct optical microscopy24
for micro-particles, which is not possible for nano-particles.
Conversely, our method determines directly hav and Q for NPs
by X-ray reflectivity, without requiring binding energy
measurements or assuming the absence of lateral interactions
between the interfacially absorbed NPs.

2 Materials and method


We used charge stabilized water dispersions of bare silica NPs
(concentration  15 wt%) having a nominal specific area of Fig. 3 Conventional box-model fitting. X-ray reflectivity from the
500 m2 g1 (Bindzil 15/500 from Akzo-Nobel). Both hexane interface of hexane with a water solution of 1 wt% silica NPs, but no
and CTAB (Sigma-Aldrich) had a purity $99%. CTAB was surfactant (,,  104), only CTAB molecules at 0.05 mM concentration
used as received while hexane was further purified by double (O,  102), and both 1 wt% NPs and 0.05 mM CTAB (B). The curves
filtering through alumina powder. High purity ($99.999%) have been vertically shifted from each other for clarity. The nanoparticle
monolayer formation at the interface requires the presence of both the
NaCl (Sigma-Aldrich) was roasted at about 600  C for several
surfactant and nanoparticles. The continuous lines are box-model
hours before adding it to the water phase at 1 mM concen-
reflectivity fits corresponding to the electron density profiles presented in
tration in order to promote CTAB adsorption onto the silica the inset (a). The orange lines indicate the box-model density profile with
particles.21 The solutions were prepared using ultrapure (thick line) and without (thin line) interfacial roughness. Dashed red lines
deionized water (resistivity 18.2 MU cm1) produced by an indicate the best fit of the physical model discussed in Fig. 4. The circle
ELGA purification system. Particle sizing (Fig. 2(a)) was per- representing the particles in the inset is drawn to scale based on the
formed by small-angle X-ray scattering (SAXS) at the average particle size determined by SAXS (5.34 nm).

Fig. 2 System pre-characterization by complementary techniques. (a) Particle size characterization by small-angle X-ray scattering. (b and c)
Time dependence of the water–hexane interface tension at low (0.05 mM) and high (0.5 mM) CTAB concentrations in the absence (b) and presence (c) of
1 wt% silica nanoparticles.

This journal is ª The Royal Society of Chemistry 2012 Soft Matter, 2012, 8, 11478–11483 | 11479
View Article Online
Published on 20 September 2012. Downloaded by Open University on 18/06/2013 04:09:37.

Fig. 4 Physical model fitting. Measured reflectivity curves R (symbols) normalized by the theoretical Fresnel reflectivity Rf. Continuous lines indicate
physical model fits for the 1 wt% NP system with 0.05 mM CTAB (B) and 0.75 mM CTAB (O). The dashed lines are model calculations in which only
the particle average immersion hav (Fig. 1(a)) is varied by 0.2 nm. The best-fit density profiles are plotted in the inset (a), where the circles representing
the particles are drawn to scale and positioned at the fitted hav values of 0.45 nm (B, red profile) and 1.04 nm (O, green profile). For the high density
monolayer (B), the XRR can be fitted only by close packed NPs, with a lateral separation (Fig. 1(a)) s ¼ 0.2  0.3 nm, as depicted in the inset (b). For
the low density monolayer (O), the XRR can be fitted either by a uniform interface covered by well-separated NPs (s ¼ 1.9 nm, inset (c)) or close packed
NPs covering only 60% of the interface (s ¼ 0 nm, inset (d)).

phase such that its surface was 1 mm above the beaker edge. this period CTAB-coated NPs slowly adsorb to the interface
The upper edge of the beaker was precisely cut and mechan- where the NPs and CTAB molecules rearrange until a ther-
ically roughened in order to favor a stable pinning of the water modynamically stable structure is reached. Our XRR
triple-phase line thus avoiding water overflowing into the measurements address this structure, several hours after inter-
external beaker. Hexane was gently poured into the external face formation.
beaker until it completely covered the water surface thus The second issue is the determination of the NPs size distri-
forming a buried water–oil interface. All parts in contact with bution, needed to model the measured XRR, as discussed below.
the liquid phases were cleaned by immersing in Piranha solu- This distribution was measured by bulk small-angle X-ray scat-
tion followed by extensive rinsing in ultrapure water. XRR tering (SAXS), as shown in Fig. 2(a). Here q ¼ (4p/l)sin(qs/2),
profiles were measured repeatedly at various lateral sample l ¼ 0.1 nm, and qs are the wavevector transfer, X-ray wavelength,
positions between 1 and 12 hours after the formation of the and scattering angle, respectively.
interface to ensure that the final stable state was being The maximum at qm x 0.38 nm1, due to the system’s struc-
probed and to exclude any irradiation-induced structural ture factor S(q), indicates an average inter-NP distance of 2p/qm
modifications. x 16.5 nm. The local minimum at qp x 0.15 A 1, due to the
particle’s form factor P(q), yields an approximate average
3 Results and discussion particle radius of rav x 4.5/qp x 3 nm. A 100-fold dilution (black
diamonds, Fig. 2(a)) eliminates the peak, which further confirms
3.1 Interface tension and SAXS measurements that the original stock solution contains mostly individual, un-
Before discussing the XRR results, two issues must be aggregated NPs. A fit (line) to a Schultz distribution P(q) model31
addressed. First, the attainment of structural equilibrium was yields an average NP radius rs ¼ 2.7 nm and a distribution width
ensured by monitoring the time evolution of gow, as shown in ss ¼ 0.9 nm.
Fig. 2(b). At fc ¼ 0.05 mM, well below fcmc ¼ 0.92 mM of the
NaCl-free solution,29 gow x 28 mN m1, significantly below the
3.2 XRR measurements
pure water–hexane gow ¼ 50.8 mN m1.30 A 10-fold higher fc ¼
0.5 mM ( fcmc reduces gow to (5 mN m1. For both fc a We now return to our main results, the XRR measurements,
stable gow is reached minutes after interface formation, con- shown in Fig. 3(b) (symbols), where R(qz), qz ¼ (4p/l)sin(a), a,
firming a fast CTAB adsorption and thermodynamic equilib- and l ¼ 0.0178 nm are the reflected fraction of the incident
rium attainment. Adding fnp ¼ 1 wt% of NPs constrains gow at X-ray beam’s intensity, the interface-normal wavevector trans-
30–35 mN m1 at both fc (Fig. 2(c)), suggesting no increase in fer, the X-ray’s incidence angle (Fig. 1(b)) and wavelength,
the interfacial CTAB with increasing bulk fc, but rather an respectively. In the absence of either CTAB or NPs (lower
increased CTAB adsorption onto the water-suspended NPs’ curves), R is monotonic, implying an interfacial electron density
surfaces. Furthermore, the NPs cause gow to decrease slowly profile (r) varying monotonically between r of water and
over 1 to 2 hours. Rheological studies22 suggest that during hexane. The top curve, however, exhibits a pronounced dip, the

11480 | Soft Matter, 2012, 8, 11478–11483 This journal is ª The Royal Society of Chemistry 2012
View Article Online

 

clear signature of an interfacial layer. The dip’s position, qz z rhe  rw z


ri ðzÞ ¼ ri ðrhe ; rw ; si ; zÞ ¼ 1 þ erf pffiffiffi : (2)
1.9 nm1, yields a thickness of d ¼ (2p/qz) x 3.5 nm, showing 2 2si
the layer to be a monolayer as sketched in Fig. 3(a). Thus,
Fig. 3 demonstrates that an interfacial monolayer forms only r0 describes the NPs’ contribution to the interfacial electron
when both CTAB and the NPs are present. Moreover, Fig. 4 density:
demonstrates that the monolayer’s structure varies with fc. The  
r0 ðzÞ ¼ r0 s; rp ; z
lower-qz, and slightly shallower, dip at high fc in Fig. 4 implies
2prp  
a thicker monolayer of a lower average r. However, deriving ¼ pffiffiffi  z2 þ 2hðh þ rÞz  hðh þ 2rÞ (3)
2
quantitative conclusions from the XRR curves requires a 3ð2r þ sÞ
physically motivated model for r. This we discuss next. for h # z # h + 2r, and r0(z) ¼ 0 otherwise. Eqn (3) is the density
Published on 20 September 2012. Downloaded by Open University on 18/06/2013 04:09:37.

The two monotonic Rs in Fig. 3 are well fitted (solid lines) by profile of a monolayer of hexagonally ordered spheres of radius
an error-function model r (eqn (2) below) increasing mono- r, immersion depth h (h > 0 denotes water immersion), lateral
tonically from rhe ¼ 228 e nm3 of hexane to rw ¼ 334 e nm3 separation s (Fig. 1(a)), and electron density rp, fixed at 630 e
of water over a width si, due to the interfacial roughness nm3, as calculated from the pNP’s
caused by thermally excited capillary waves (CWs).32 The fit  density, (2.1  0.1)
ffiffiffiffiffiffi mass  g
cm3.34 Gðxav ; sx ; xÞ ¼ ð1=sx 2pÞexp  ðx  xav Þ2 =2sx 2 are
yields si ¼ 0.43 nm for NP-only (,) and si ¼ 0.54 nm for the normalized Gaussian distributions of r and h, with averages
CTAB-only (O) not too far from sCW x 0.35 nm calculated32 rav and hav and widths sr and sh. G(hav,sh;h) describes the NP’s
for a typical g ¼ 32 mN m1 and an instrumental resolution vertical disorder mostly due to CWs, but also to thermal fluc-
dqz ¼ 0.15 nm 1. Small remaining fit discrepancies (Fig. 3) tuations in the NP’s immersion depth around the instantaneous
may arise from interface adsorption of surfactants (CTAB- interface position. The subtracted term in eqn (1) accounts for
only),33 or marginal interfacial NP layering (NP-only). Both the liquid displaced by the NPs. The r(z) model above was fitted
effects can also induce a small increase of the interface to the Fresnel-normalized reflectivity profile R/Rf (Fig. 4) using
roughness as indeed found experimentally. Parratt’s recursive algorithm,35 keeping rw, rhe and rp fixed, and
Modeling the top R in Fig. 3 is more complicated. A simple varying the roughness and particle size parameters only within
box model, approximating the interfacial layer by a uniform narrow ranges straddling the values derived from the SAXS and
high-density slab, yields a good fit (orange line), showing the XRR control measurements discussed above. Since the NPs’
layer to be single-NP thick and to consist of densely packed vertical disorder is correlated with, and dominated by, the CWs’
NPs, mostly immersed in hexane, thus confirming the NPs’ si, we limited si < sh < 2si. It is important to note that only hav
hydrophobicity when CTAB is present in the solution. and s were completely free fit parameters, with hav controlling the
However, this averaged density-level model provides no infor- position and s the height of r(z)’s interfacial peak (Fig. 4(a)).
mation on the NPs’ packing within the monolayer. In addition, Thus the number of completely free parameters in our physical
this box model allows only a generic assessment of the hydro- model is not larger than that of a simple box model. Limiting or
phobic versus hydrophilic character of the particles based on the fixing the other parameters of our model by independent
degree of asymmetry of the interfacial r(z) peak and on a measurements or theoretical calculations is essential for avoiding
qualitative evaluation of its position with respect to the low inter-parameter correlations in the fit, which may lead to
density oil phase and higher density water phase. unphysical parameter values. The fits based on eqn (1) yield the
solid lines in Fig. 4, the corresponding r(z) profiles in Fig. 4(a),
3.3 Physical model for XRR and the parameter values in Table 1.

A more physically motivated model is needed to extract quan-


titative conclusions from the measured data. Such a model
Table 1 Physical model parameters for fits reported in Fig. 4 and dis-
should contain, in addition to the conventional error function cussed in the text. fc indicates the surfactant concentration, with an
describing the liquid–liquid interface (eqn (2)), a term uncertainty of 0.02 mM. rav (Fig. 1(a)) and sr are the average NP radius
describing the exact electron density profile of solid spherical and standard deviation of the Gaussian radius distribution, respectively.
particles bound to the interface. Instead of the generic box hav (Fig. 1(a)) and sh are the average immersion of the NPs in the water
subphase and the corresponding Gaussian standard deviation, respec-
thicknesses and roughnesses, the parameters of a physical tively. si is the CW-induced roughness of the hexane–water interface. The
model should be directly related to physical properties of the error bar associated with the NP and interface-normal structural
system, such as the particles’ average radius and polydispersity, parameters is (0.1 nm. s is the lateral separation between the surfaces
of neighboring silica NPs (Fig. 1(a)), with an associated uncertainty of
and the average vertical separation between the particles’
(0.3 nm. Q is the NP’s contact angle (Fig. 1(a)) calculated using the
centers and the position of the liquid–liquid interface. An fitted values for rav and hav. DE is the NP’s binding energy calculated
implementation of such a model, including also the NPs’ in- using the fitted hav and the measured interface tension gow  28 mN m1.
plane packing details and Gaussian (G) size and water immer- Only the two parameters in bold were left completely free for fitting the
XRR profiles. The remaining parameters were fixed (rav and sr) or
sion distributions, is given by: strongly constrained (si and sh) by complementary measurements and
ðð
    theoretical calculations
rðzÞ ¼ ri ðrhe ; rw ; si ; zÞ þ r0 s; rp ; z  r0 ðs; ri ðzÞ; zÞ
fc/mM rav/nm sr/nm si/nm sh/nm hav/nm s/nm Q/ DE/kT

Gðrav ; sr ; rÞGðhav ; sh ; hÞdrdh: (1) 0.05 2.67 0.44 0.60 0.60 0.45 0.2 146  4 5  2
0.75 2.67 0.44 0.58 0.59 1.04 1.9 128  6 24  10
ri describes the si-wide pure water–hexane interface:

This journal is ª The Royal Society of Chemistry 2012 Soft Matter, 2012, 8, 11478–11483 | 11481
View Article Online

Before discussing the parameter values reported in Table 1, it interfacial tensions of both high- and low-fc cases (Fig. 2(c)),
is important to note that the two experimental profiles presented yields, respectively, DEh ¼ 23.6 kT and DEl ¼ 4.5 kT. Error
in Fig. 4 appear to be markedly different already upon simple estimations focusing in particular on the role of interfacial
visual inspection: surfactant addition causes an overall decrease CTAB molecules (ESI, Fig. 3 and 4, and Table 1†) yield the cor-
of the measured R/Rf profile, while the local dip at qz z 1.9 nm1 responding maximum error estimates DEH ¼ 24  10 kT and
shifts to lower qz and becomes shallower. The observed changes DEL ¼ 5  2 kT.
of measured reflectivity are connected to variations in the NP The classical DE calculations would suggest that at low fc,
ML’s surface-normal electron density profile r(z). Several DE is only slightly above kT and the NPs’ binding is very weak,
attempts to fit both curves in Fig. 4 by changing only the liquid– while at high fc the binding is five-fold stronger. The low DE
liquid roughness (si) and the vertical NP disorder parameter (sh) obtained at low fc, however, contrasts starkly with the highly
were unsuccessful. Conversely, similar si and sh values but stable, close-packed (s z 0 nm) XRR-determined monolayer
Published on 20 September 2012. Downloaded by Open University on 18/06/2013 04:09:37.

different water immersion hav and in-plane NP spacing s values structure. The two quantities used to calculate DE are accu-
(Table 1) yield good fits to both experimental datasets (Fig. 4). rately determined by either direct measurements (gow, Fig. 2) or
The existence of multiple r(z) profiles agreeing well with a single detailed XRR modeling (hav), which we believe to be physically
reflectivity curve cannot be excluded a priori. A systematic search robust and comprehensive. We must conclude, therefore, that
for alternative solutions was performed using model-indepen- the classical formula itself14,16 is inaccurate,37 particularly at low
dent stochastic approaches in which the only input parameters fc. This is not too surprising, since that formula assumes non-
are the materials’ bulk electron densities and the expected interacting particles, while the observed close packing implies
thickness of the interfacial layer. As shown in the ESI, Fig. 5,† strongly interacting NPs, possibly leading to irreversible
such stochastic searches converge to solutions that are in rather aggregation into a solid-like interfacial NP monolayer, as
good agreement with the physical model results. found by light microscopy22 for the same system studied here. A
The parameter values summarized in Table 1 lead to the solid-like character would make single-particle detachment
following conclusions: at low-fc the NPs are mostly immersed in impossible, thus stabilizing the monolayer even at the high
hexane, hav ( 0.5 nm, and basically in lateral contact with each measured Q  146 .
other, s  0.2  0.3 nm (Fig. 4(b)). At high-fc, the particles are The s z 0 spacing also implies a non-simple NP self-assembly
less oil-immersed (hav T 1 nm) thus suggesting lower NP process for the monolayer at low-fc. In the bulk, there is no
hydrophobicity. The NP monolayer density is lower, as reason to assume non-uniform CTAB adsorption on the NP’s
demonstrated by a lower r(z) peak (Fig. 4(a), green line) and a surface. However, s z 0 implies no CTAB intrusion between the
larger s  1.9 nm (Fig. 4(c)). However, since XRR is sensitive NPs. Thus, the CTAB coating of the monolayer’s NPs must be at
only to the average density, not to the actual in-plane NP order, least anisotropic, if not completely absent, indicating that a
an identical fit is obtained assuming a 60% interface coverage by surfactant/NP structural rearrangement indeed occurs during
close packed (s ¼ 0 nm) NPs, and a 40% coverage by CTAB interfacial NP adsorption, as suggested by computer simulation23
molecules (Fig. 4(d)). Grazing-incidence small angle X-ray and surface tension and rheology measurements.21 The rear-
scattering (GISAXS), which could, in principle, resolve the in- rangement is possible because CTAB is only electrostatically
plane NP packing,36 was found to be unfeasible since the stronger physisorbed, not chemically bound, to the NP surface. Aniso-
background scattering from the bulk liquid masked the weaker tropic CTAB coverage imparts anisotropic hydrophobicity,
scattering from the monolayer. By contrast, the fit’s high sensi- which may drive the NPs’ interfacial self-assembly. The drive is
tivity to the surface-normal order is demonstrated by the model- evidently strong enough to overcome, at the surface, the inter-
calculated R/Rf (Fig. 4, dashed lines) obtained when hav is particle repulsion provided, in the bulk, by the colloidal charge
changed by 0.2 nm from the best fit values, keeping all other stabilization mechanism.
parameters unchanged. The pronounced deviations demonstrate For low fc the bulk NPs are known22 to become hydrophobic,
that hav is determinable to better than 0.2 nm (see also the ESI, albeit insufficiently so to aggregate. Thus, the bulk NPs’ surface
Fig. 1 and 2†). activity is maximal. At high fc, the NPs can undergo partial
flocculation in the bulk, yielding a lower interfacial activity. This
may explain the monolayer’s lower r at high fc. An alternative
3.4 Discussion
explanation may be found by invoking the formation, at high fc,
A well-defined hav value allows determining accurately two of a more compact and dense CTAB shell covering the surface of
physically important NP properties: the contact angle and the the water-dispersed NPs. Such a dense surfactant shell would
binding energy. The former, given by Q ¼ arccos(1 + hav/rav), prevent the NPs from coming into close contact once they reach
yields Ql ¼ 146.1 and Qh ¼ 127.6 for low and high fc. Simula- the interface.
tions, using eqn (1), provide estimates for the systematic errors XRR cannot discriminate between a phase-separated NP–
arising from neglecting immersion depth variations due to NP size surfactant film (Fig. 4(d)) and a uniformly expanded NP layer
polydispersity, non-CW induced disorder in the NPs vertical (Fig. 4(c)). However, the local structure of aggregated NP
positions, and the CTAB contribution to the monolayer’s r. patches (Fig. 4(d)) at high fc and of a uniform close-packed
The resulting maximum error bars yield QL ¼ 146  4 and QH ¼ monolayer at low fc are identical, and thus expected to yield the
128  6 . Clearly, even with these relatively large error estimates Q same Q  146 , in spite of the different average monolayer
remains significantly different for the two fc values. Using the densities. Finding a smaller contact angle (128 ) at higher fc
classical formula for non-interacting particles DE ¼ pgowhav2,14,16 supports, therefore, a uniformly expanded NP layer (Fig. 4(c))
with an average gow ¼ 28  4 mN m1 for the varying, yet close, over a phase-separated layer (Fig. 4(d)).

11482 | Soft Matter, 2012, 8, 11478–11483 This journal is ª The Royal Society of Chemistry 2012
View Article Online

4 Conclusions 6 S. Levine, B. D. Bowen and S. J. Partridge, Colloids Surf., 1989, 38,


345–364.
In conclusion, combining conventional gow measurements with 7 K. Danov, P. Kralchevsky, K. Ananthapadmanabhan and A. Lips,
Langmuir, 2006, 22, 106–115.
cutting-edge synchrotron techniques we have determined with 8 F. Bresme and J. Faraudo, J. Phys.: Condens. Matter, 2007, 19,
high precision the interface-normal nano-scale structure of 413101–413134.
CTAB/nano-silica monolayers formed at buried water–hexane 9 H. Ma, M. Luo and L. L. Dai, Phys. Chem. Chem. Phys., 2008, 10,
interfaces. The NP–water contact angle was observed to vary 2207–2213.
10 Y. Song, M. Luo and L. L. Dai, Langmuir, 2010, 26, 5–9.
with fc within the high hydrophobicity range (Q T 130 ). 11 B. P. Binks, Curr. Opin. Colloid Interface Sci., 2002, 7, 21–41.
Classical calculations14,16 yield weak NP adsorption, DE ¼ 12 R. Aveyard, B. Binks and J. Clint, Adv. Colloid Interface Sci., 2003,
5–25 kT. The contrast between the low DE values and the 100, 503–546.
13 S. Kubowicz, M. A. Hartmann, J. Daillant, M. K. Sanyal,
observed high stability of the NP monolayers demonstrates the
Published on 20 September 2012. Downloaded by Open University on 18/06/2013 04:09:37.

V. V. Agrawal, C. Blot, O. Konovalov and H. Moehwald,


inaccuracy of the binding energy calculations, in particular at Langmuir, 2009, 25, 952–958.
low fc. The microscopic structural scenarios revealed by our 14 S. Levine, B. D. Bowen and S. J. Partridge, Colloids Surf., 1989, 38,
measurements are supported by computer simulations23 and 325–343.
15 K. Esumi, J. Colloid Interface Sci., 2001, 241, 1–17.
macroscopic measurements.21,22 The present method, which 16 A. B€oker, J. He, T. Emrick and T. P. Russell, Soft Matter, 2007, 3,
allows studying quantitatively a broad spectrum of interfacial 1231–1248.
NP binding phenomena, should prove useful for basic sciences 17 N. G. Eskandar, S. Simovic and C. A. Prestidge, Phys. Chem. Chem.
and applications alike. We hope that this study will motivate also Phys., 2007, 9, 6426–6434.
18 C. Vashisth, C. P. Whitby, D. Fornasiero and J. Ralston, J. Colloid
theoretical and simulation work leading to a deeper under- Interface Sci., 2010, 349, 537–543.
standing of the energetics, formation, and hydrophobicity- 19 Z. G. Cui, L. L. Yang, Y. Z. Cui and B. P. Binks, Langmuir, 2010, 26,
related properties of these technologically important nano-scale 4717–4724.
20 B. P. Binks and J. A. Rodrigues, Colloids Surf., A, 2009, 345, 195–201.
multi-component interfaces. 21 F. Ravera, E. Santini, G. Loglio, M. Ferrari and L. Liggieri, J. Phys.
Chem. B, 2006, 110, 19543–19551.
22 F. Ravera, M. Ferrari, L. Liggieri, G. Loglio, E. Santini and
Acknowledgements A. Zanobini, Colloids Surf., A, 2008, 323, 99–108.
23 M. Luo and L. L. Dai, J. Phys.: Condens. Matter, 2007, 19, 375109–
We thank Theyencheri Narayanan (ESRF) for access to the 375123.
ID02 beamline and for helpful suggestions concerning SAXS 24 K. Du, E. Glogowski, T. Emrick, T. P. Russell and A. D. Dinsmore,
data modeling, Ran Teller (Bar-Ilan) for assistance with interface Langmuir, 2010, 26, 12518–12522.
25 T. Narayanan, Curr. Opin. Colloid Interface Sci., 2009, 14, 409–415.
tension measurements, Peter Greenwood and Bo Larsson (Akzo 26 D. Pontoni, T. Narayanan and A. R. Rennie, J. Appl. Crystallogr.,
Nobel) for providing colloids and detailed sample information, 2002, 35, 207–211.
Ben Ocko, Dominique Langevin and Luc Belloni for helpful 27 H. Reichert, V. Honkim€aki, A. Snigirev, S. Engemann and H. Dosch,
discussion, and the ESRF for support and provision of beam- Phys. B, 2003, 336, 46–55.
28 V. Honkim€aki, H. Reichert, J. S. Okasinski and H. Dosch,
time. Support to M.D. by the US-Israel Binational Science J. Synchrotron Radiat., 2006, 13, 426–431.
Foundation, Jerusalem, is gratefully acknowledged. 29 J. Mata, D. Varade and P. Bahadur, Thermochim. Acta, 2005, 428,
147–155.
30 S. Zeppieri, J. Rodriguez and A. L. L. de Ramos, J. Chem. Eng. Data,
References 2001, 46, 1086–1088.
31 D. Pontoni, S. Finet, T. Narayanan and A. Rennie, J. Chem. Phys.,
1 A. K. Boal, F. Ilhan, J. E. DeRouchey, T. Thurn-Albrecht, 2003, 119, 6157–6165.
T. P. Russell and V. M. Rotello, Nature, 2000, 404, 746–748. 32 A. Braslau, P. S. Pershan, G. Swislow, B. M. Ocko and J. Alsnielsen,
2 A. Bausch, M. Bowick, A. Cacciuto, A. Dinsmore, M. Hsu, Phys. Rev. A, 1988, 38, 2457–2470.
D. Nelson, M. Nikolaides, A. Travesset and D. Weitz, Science, 33 L. Tamam, D. Pontoni, Z. Sapir, S. Yefet, E. Sloutskin, B. M. Ocko,
2003, 299, 1716–1718. H. Reichert and M. Deutsch, Proc. Natl. Acad. Sci. U. S. A., 2011,
3 H. W. Duan, D. Y. Wang, D. G. Kurth and H. M€ ohwald, Angew. 108, 5522–5525.
Chem., Int. Ed., 2004, 43, 5639–5642. 34 B. Larsson, Akzo–Nobel Corporation, personal communication.
4 Y. Lin, H. Skaff, T. Emrick, A. D. Dinsmore and T. P. Russell, 35 L. G. Parratt, Phys. Rev., 1954, 95, 359–369.
Science, 2003, 299, 226–229. 36 D. Pontoni, K. J. Alvine, A. Checco, O. Gang, B. M. Ocko and
5 M. G. Nikolaides, A. R. Bausch, M. F. Hsu, A. D. Dinsmore, P. S. Pershan, Phys. Rev. Lett., 2009, 102, 016101.
M. P. Brenner, D. A. Weitz and C. Gay, Nature, 2002, 420, 299–301. 37 D. L. Cheung and S. A. F. Bon, Phys. Rev. Lett., 2009, 102, 066103.

This journal is ª The Royal Society of Chemistry 2012 Soft Matter, 2012, 8, 11478–11483 | 11483

You might also like