INVITED ARTICLE
www.rsc.org/pccp | Physical Chemistry Chemical Physics
Modifying the liquid/liquid interface: pores, particles and deposition
Robert A. W. Dryfe*
Received 20th December 2005, Accepted 7th March 2006
First published as an Advance Article on the web 20th March 2006
DOI: 10.1039/b518018j
The modification of the liquid/liquid interface with solid phases is discussed in this article.
Modified interfaces can be formed with molecular assemblies, but here attention is focussed on
solid materials such as mesoscopic particles, or microporous and mesoporous membranes. Charge
transfer across the modified liquid/liquid interface is considered in particular. The most obvious
consequence of the introduction of such modifying components is their effect on the transport to,
and the transfer of material across, the liquid/liquid interface, as measured voltammetrically for
example. One particularly interesting reaction is interfacial metal deposition, which can also be
studied under electrochemical control: the initial formation of metal nuclei at the interface
transforms it from the bare, pristine state to a modified state with very different reactivity.
Deposition at interfaces between liquids is compared and contrasted with the cases of metal
deposition in bulk solution and conventional heterogeneous deposition on conducting solid
surfaces. Comparison is also made with work on the assembly of pre-formed micron and
nanometre scale solids at the liquid/liquid interface.
Introduction
The study of interfaces, with increasing spatial and temporal
resolution, has been a central theme in physical chemistry over
the past century. Many of the most detailed studies have been
performed on the surfaces of solids, in contact with either
vapour or liquid phases. By contrast, the interfaces between
liquids and gases and those between immiscible liquids have
been studied intensively via macroscopic techniques, such as
interfacial tension measurements.1 The nature of the liquid/
liquid (L/L) and liquid/gas interfaces (their fluidity, the volatility of many liquids at ambient conditions and their electrically insulating character) makes it difficult or impossible to
transpose techniques such as high resolution electron or probe
microscopy, for example, to these interfaces. The question of
interfacial structure is altogether different in meaning at fluid
interfaces: beyond the structure of the individual molecular
component(s) of the interface, and the macroscopic structure
(shape) of the interface, it is difficult to talk of structure on any
intermediate (mesoscopic) length scale. However when one
considers a heterogeneous or ‘‘modified’’ interface between
liquids (see below) details of the size, density and distribution
of the heterogeneities become important. Overall the relationship between structure and reactivity—on a microscopic level—is much less clear at liquid surfaces, as compared to solid
surfaces. Surface-sensitive techniques, such as sum-frequency
spectroscopy and X-ray reflection,2,3 have recently been applied to the surfaces of liquids and are finally beginning to
yield accurate structural parameters, such as the thickness of
the interfacial region, or ‘‘mixed layer’’.4
School of Chemistry, University of Manchester, Oxford Road,
Manchester, UK M13 9PL. E-mail: [email protected];
Fax: +44 161 275 4734
This journal is
c
the Owner Societies 2006
In many ways, the L/L interface is a unique environment
because of the discontinuity in physical properties engendered
by the presence of the interface. By analogy with solid
surfaces, it can be contended that the current challenge is to
extend the insight gained from recent ‘‘high-resolution’’ experimental techniques and advances in computation,5 from
pristine to ‘‘complex’’ L/L interfaces. Emulsions and cell
membranes can be thought of as ‘‘modified’’ L/L interfaces,
where the modifying agent is a molecule or a supramolecular
assembly (i.e. a surfactant or phospholipid/protein assembly,
respectively). L/L interfaces can also be modified by the
adsorption of mesoscopic or microscopic colloidal particles
(e.g. of organic polymers, or oxide materials) at the interface.
The concept of using particles as surfactants has a long
pedigree, however it has recently been given a new impetus
by recent developments in the synthesis and characterisation
of nanoparticulate materials, which have been shown to display unusual segregation behaviour when adsorbed to L/L
interfaces.6 The surface energies of the water/organic, water/
solid and organic/solid interfaces (as well as the line tension at
the three-phase boundary) dictate the stability of the assembly,
as will be discussed below.
A given L/L interface can also be modified by transformation into an ensemble of smaller interfaces. The distinction
between this, and the previous case, where some kind of
particle assembly is formed at the interface is illustrated in
Fig. 1. The ensemble can be formed by supporting the interface within a porous material. Ideally the material should have
a well-defined pore size and, for the formation of a stable
interface at a known position, it should ideally be either
hydrophilic or hydrophobic. If the modifying material has a
regular (periodic) structure then, by analogy with the formation of microelectrodes,7 an array is formed. In the case of a
microelectrode ensemble/array, the material used must be
electronically insulating, to restrict any charge transfer
Phys. Chem. Chem. Phys., 2006, 8, 1869–1883 | 1869
Fig. 1 Schematic illustrating the definitions employed throughout
this article of (a) an ‘‘ensemble’’ versus (b) an ‘‘array’’. Note that the
individual elements of the array/ensemble could be individual particles
present at a L/L interface, or pores present in a solid material
separating two immiscible liquid phases.
processes to the conductive substrate. In the L/L case, ideally
both electronic and ionically insulating behaviour is required.
This approach has two general advantages: (i) the formation
of a micron-scale (or smaller) interfaces means that masstransport to each ensemble element from the solution phase(s)
is increased, and (ii) the presence of the material serves to
stabilise the L/L interface against the shearing effects of flow,
which in turn means that forced convection can be used to
enhance mass-transport to the L/L interface. In terms of
interfacial processes, these experimental configurations can
be used to investigate either ion transfer (IT) or electron
transfer (ET) at the L/L interface. Prior work from my
laboratory has focussed on the IT process, largely because
there are still a number of outstanding issues relating to this
deceptively simple event.8 The mechanism of IT and, in
particular the existence—or otherwise—of a kinetic barrier
to the ion de-solvation/re-solvation process is still a point of
debate. The interfacial IT process can also be used to control
the composition of porous materials located at, or near, the L/
L interface, as will be discussed later in this article.
Finally, instead of placing pre-formed particles at the L/L
interface, materials such as metallic nanoparticles can be
synthesised (i.e. grown and deposited) at the L/L interface.
Metal deposition, by reduction of metal ions using a molecular
reductant, allows the growth and assembly of particles to be
performed in situ. This process is particularly interesting
because metallic nanoparticles are typically formed via solution phase reduction processes.9 There is an enormous and still
burgeoning literature on the formation of such particles, with
the ultimate aim being control over the size and shape of the
1870 | Phys. Chem. Chem. Phys., 2006, 8, 1869–1883
materials. Control can be achieved by ‘‘capping’’ the growing
particle using appropriate ligands and/or other ‘‘restriction
agents’’, e.g. microemulsions.10 The synthetic work in this area
is well ahead of the physical chemistry: the factors controlling
particle nucleation and growth are poorly understood on a
quantitative level. A recent review on solution phase ‘‘nanowire’’ preparation has noted that the mechanism of the growth
process has not been ‘‘rigorously worked out, even for the
simpler case of nanospheres’’.11 The solution phase synthetic
strategies are, in the main, empirical with a lack of quantitative
data on how such particles are assembled—which hampers
understanding of how their growth may be controlled. My
research group–and others—have been developing an approach to this problem, which involves localising the deposition process to interfaces,12–14 specifically to L/L interfaces
(see Fig. 2). In this case, solution phase metal deposition
occurs at the L/L interface, with the metal precursor located
in phase 1 and the reducing agent in phase 2, meaning that the
reduction process is interfacial. This approach simplifies the
problem by localising the reaction to a single plane (the
interface), rather than allowing a random distribution of
nucleation events throughout the volume of the solution. It
also means that surface-sensitive and/or surface-specific
Fig. 2 Schematics representing (a) homogeneous ET in the bulk of a
solution, case (i), and the case where a metallic (nano)-particle is
formed in solution, case (ii); (b) heterogeneous ET at the solution/
electrode interface, case (i), and the formation of a metallic phase on a
conducting substrate, via heterogeneous ET, case (ii); and (c) ET at the
L/L interface, case (i), and the formation of a metallic phase at a L/L
interface, also via heterogeneous ET, case (ii).
This journal is
c
the Owner Societies 2006
experimental techniques can be applied to probe the deposition and/or assembly of particles at this interface.15 The
question, given the interfacial locus of the process, then
becomes: how faithfully does nucleation and growth at the
L/L interface correspond to nucleation and growth in bulk
solution? Are there any factors specific to the interface, which
cause the bulk and interfacial processes to differ? It has been
noted that ET in general at the L/L interface (as opposed to
the specific case of ET leading to deposition) can be treated as
intermediate between homogeneous ET and conventional
heterogeneous (electrode/electrolyte) ET.16 Accordingly here,
we will contrast some of the specific features associated with
deposition at the L/L interface with the conventional electrodeposition process on solid surfaces and metal deposition in
bulk solution phases. The unique ability of electrodeposition
to control the interfacial supersaturation means a ‘‘golden
triangle’’ relating the driving force for deposition, the rate of
deposition and the morphology of the resultant deposit can be
constructed (see Fig. 3). Much of the fundamental electrochemical surface science work of the past 15 years has been
concerned with elaborating the relationships between these
three parameters for well-defined systems (generally noble
metal deposition on single crystal surfaces).17
However, given that nucleation on solid surfaces proceeds
mainly from defect sites, a meaningful comparison of nucleation rates from surface to surface can only be made with some
knowledge of the density (and nature) of the defect sites, or by
use of well-defined single crystals as substrates for electrodeposition. By contrast, deposition at the L/L interface—at least
for the initial stages of deposition—involves a ‘‘defect-free’’
surface, thus it is all the more surprising that so little is known
about how the vertices of the triangle in Fig. 3 are inter-related
at this interface.
The final section of the article will discuss potential overlap
between the electrochemically-oriented work and recent intriguing results presented, in the main, by colloid scientists. This
work has described the assembly of micron and sub-micron
scale particles at L/L interfaces.6,18,19 Interestingly, size dependent assembly and long-range order have been demon-
Fig. 3 The idealised triangle linking the driving force for the formation of a new phase, to the rate of the formation of that phase, and its
resultant morphology. Note that potentiostatic deposition offers control over the ‘‘driving force’’ vertex, whereas galvanostatic deposition
acts on the ‘‘rate’’ vertex.
This journal is
c
the Owner Societies 2006
strated in a number of systems. The possibility of introducing
the applied potential difference as an additional degree of
freedom is presented.
Ion transfer at the modified L/L interface: mesoporous materials
Electrochemical techniques are powerful probes of solution
phase thermodynamic, kinetic and transport parameters. Voltammetric measurements give direct measures of interfacial
fluxes, which are otherwise difficult to de-convolute from
processes within the bulk of the phase. Following the instrumental developments pioneered in Montpellier and Prague in
the 1970s,20,21 such techniques have been applicable to charge
transfer processes at the L/L interface: much is made of this
experimental approach in this article.
Returning to conventional electrochemistry, the benefits of
using electrode materials of micron-scale dimensions are well
documented. The resultant convergent diffusive flux leads to a
higher rate of mass transport to the electrode surface and,
frequently, the flux of electroactive material (the current) tends
to a steady-state, which facilitates measurement and analysis.22 Microelectrode arrays/ensembles (see preceding section)
can also be formed, which combine some of the advantages of
microelectrodes (higher current densities) with those of macroscale electrodes (higher absolute currents). Broadly the behaviour of the array is dictated by two related factors which
determine whether the array acts as N individual microelectrodes, or whether it takes on a new ensemble character.23–25
The first factor is the thickness of the steady-state diffusion
field (d), compared to the inter-element separation (2R): generally if d > 2R, then ‘‘ensemble’’ behaviour due to a single
overlapping diffusion field dominates, whereas the converse
condition leads to the retention of individual behaviour.
Experimentally, the simplest way to generate the microelectrode array is to modify the conducting substrate with an
insulating, porous overlayer. This gives rise to a second factor,
namely the degree to which the individual elements are
recessed with respect to the insulating surround: using the
overlayer approach means the recess depth (L) can be significant. Experimentally the important factor is the size of L
compared to d, although the former will influence the latter.
Conducting surfaces can be modified with insulating membrane materials: commercially available g-alumina and tracketched polymer membranes have been employed to form
microelectrode ensembles.26 The deposition of metallic, and
other, materials within the pores of such materials continues to
be an area of intense activity. Electrodeposition has been
widely employed to introduce such conducting materials into
‘‘templated’’ structures, because of the degree of control the
electrochemical approach permits (see preceding section, deposition at the L/L interface is discussed subsequently). Following the approach used for metallic electrodes, we have used
both g-alumina and track-etched polyester membranes to
modify the L/L interface.27,28 These materials are commercially available with pore radii as low as 10 nm. Alternatively,
single small-scale L/L interfaces (analogous to microelectrodes) can be formed using a laser-ablated hole in a polymer
film, or a micropipette, to define the contact area between the
two liquid phases.29–31 The former approach is restricted, to a
Phys. Chem. Chem. Phys., 2006, 8, 1869–1883 | 1871
first approximation, by the beam diameter of the ablating light
source, whereas the pipette approach has been extended to
orifice radii (r) as low as 10 nm.32 For an inlaid disc geometry,
the mass-transfer coefficient (kmt) is given by:33
kmt ¼
4D
pr
ð1Þ
where D is the diffusion coefficient of the molecule under
study. Quantification of rapid interfacial processes, such as IT
or ET, requires that kmt exceeds the rate constant of the
interfacial process. Thus, in principle, for r approaching 10
nm, interfacial rate constants exceeding 1 cm s1 should be
measurable.
Again, formation of nanometre scale L/L interfaces has
parallels with analogous work on the formation of ‘‘nanoelectrodes’’ using nanometre scale objects to define disc-34,35 or
band-shaped electrodes.36 On the nanometre scale, a key issue
becomes the fidelity of the interface: over this length scale, how
reproducibly can we construct a conducting surface? Defects
due to the (generally) poly-crystalline nature of electrode
surfaces, such as grain boundaries, are inevitably present.
Investigations of nucleation on electrode surfaces have confirmed that nucleation proceeds from higher energy sites, such
as step edges between crystalline planes,37 thus the density and
nature of these sites will affect the dynamic process. Recent
electrochemical studies of pyrolytic graphite surfaces, for
example, have suggested that there is a considerable difference
between the ET rates measured at the basal plane compared to
the edge plane of the graphite.38 For macro, or even micro,
electrodes the effects of such disparities will be averaged over
the entire surface, giving a reasonable degree of consistency
between different samples of the same nominal dimension,
assuming that the surfaces have been treated identically. It
becomes more difficult to guarantee such consistency on the
nanometre scale, as the defect scale approaches that of the
entire interface, without resorting to detailed structural characterisation of each interface under study. We therefore
suggest that the L/L interface may be an inherently more
reproducible way to study charge transfer at nanometre scale
interfaces.
The g-alumina and polyester materials referred to above
have been used at the static and hydrodynamic (flowing and
rotating configurations) L/L interfaces to study simple and
facilitated IT processes. Flow based systems have featured
prominently in studies of interfacial kinetics at L/L interfaces,39–41 although the earliest studies often resorted to ex
situ detection methods. An improvement was the development
of dropping liquid systems, which included a microelectrode
probe at defined distances from the expanding drop, permitting the more direct extraction of kinetic information from the
L/L interface.42 During the 1990s, a number of studies
emerged that combined flowing solutions with voltammetric
(in situ) measurements of the interfacial flux.43–45 In the
absence of a membrane, only rather limited hydrodynamic
studies can be performed, due to instability of the interface.46
However, the initial membrane-supported work was not fully
interpreted in terms of known hydrodynamic models.43 Uncertainties about the role of the membrane were partly responsible for the qualitative nature of the earliest studies. In
1872 | Phys. Chem. Chem. Phys., 2006, 8, 1869–1883
particular, the cellulose membranes employed in these studies
swell on contact with aqueous solutions, and their irregular
pore size and pore distribution made interpretation of the data
in terms of established mass-transport models difficult.43,44
Gellifaction of one of the liquid phases was used to stabilise
the interface, though again, initial hydrodynamic studies were
largely qualitative.47 Girault and co-workers pursued an alternative approach, fabricating an array of micro-interfaces by
laser ablation of polyethylene terephthalate (PET) films, consisting of ca. 100 array elements, for both quiescent and
flowing solutions.48 A complication was the finite number of
array elements, meaning that a significant number of the
elements lay on the array’s perimeter, with a different masstransport regime applying to the outlying elements. Sawada
and co-workers used a thin-layer flow cell with a stationary
organic phase stabilised below a hydrophilic dialysis membrane, to detect ions using pulsed amperometry.49 Girault’s
group used a micro-machined PET film to separate an organic
gel from a flowing aqueous phase.48,50 In one of the above
cases, it was shown that the aqueous stream flow rate affected
the IT current,48 however full interpretation in terms of
established models of hydrodynamic voltammetry was not
performed.
The key to quantitative exploitation of this approach is
knowledge of the flow profile, enabling mass-transport to/
from the interface to be calculated. We have adapted models
developed for the corresponding microelectrode ensembles,
due to the random pore distribution of the membranes (see the
AFM image of a PET membrane, Fig. 4) to give approximate
descriptions of the mass-transport to the L/L interface under
convective and static conditions. Under static conditions, an
overlapping diffusion field results for both types of membrane
under experimentally accessible conditions, hence the flux
(current) to the overlapping ensemble is identical to that for
macroelectrodes of the same geometric area, even though the
fraction of exposed surface (1y) may tend to zero in the
ensemble case.27,28,51 By contrast, these models have not been
extended to hydrodynamic systems, specifically where laminar
Fig. 4 An atomic force microscope image of a polyester membrane,
of the type used to modify the L/L interface for hydrodynamic IT
experiments. The horizontal scale bar is shown (microns), the vertical
contrast (nm) is given in the inset.
This journal is
c
the Owner Societies 2006
flow in a channel configuration or a rotating disc system is
used to augment mass-transport.52 Instead, the behaviour of
microelectrode arrays under hydrodynamic conditions has
been interpreted in terms of ‘‘limiting cases’’ of the behaviour
in static solution.53 We have found, however, for L/L arrays
formed in both the channel flow and rotating configurations,
the hydrodynamic voltammograms for IT across the modified
interface show identical dependence on mass-transport rate to
the corresponding macroelectrodes, see Fig. 5 and 6. Furthermore, the overlapping diffusion layer model appears to hold
under these conditions, at least for the polymer membranes,
since the limiting currents can be quantitatively predicted
using both the flow and rotating conditions, using the entire
membrane area as the ‘‘active’’ area.54,55
The prime difference is the increase in mass-transport in the
array case, which means that more rapid heterogeneous
processes (i.e. those with a higher k0) can be measured. The
apparent rate constant, kapp
0 , and the true value are related by
the following expression:51
kapp
= k0(1 y)
0
(2)
We have thus employed the above L/L ensemble approach to
measure IT under static and agitated conditions,54,55 and
hence determine kinetics of the IT process. Standard rate
constants of the order of 1 cm s1 have been determined for
ions such as tetraethylammonium using this type of approach.54
One potential complication with the membrane-modified
interface is the stability of the interfacial position, within the
interior of the modifying material. Sensitivity to interfacial
position has been noted for studies with both ensembles and
individual pores.27,56,57 This factor is related to the recess
length, L, discussed above and stems from the contact angle
of the L/L interface with respect to the solid used to modify
the interface. Contact angle effects are, in fact, extremely
important in determining the geometry of particles adsorbed
and deposited at L/L interfaces as will be discussed below.
The bulk of the above section has dealt with the importance
of using systems with known hydrodynamics: predictable
mass-transport leads to calculable chemical parameters, such
as interfacial transfer rates. On the microscopic scale, however, it may be argued that less is known about interfacial
boundary layers, particularly for immiscible streams in microfluidic systems.58 One possible development would be the
converse experiment, i.e. using the flow rate dependence of a
kinetically well-characterised interfacial process to report on
the hydrodynamics at or near the L/L interface. Fluorescent,
as well as electrochemical, detection could be envisaged as a
feasible tool to extract such information, particularly in microfluidic systems.59 In fact, such an experiment has been proposed (although not realised to date) as a method to determine
the mesoscopic roughness of the L/L interface.60
Ion transfer at the modified L/L interface: zeolite materials
Microporous materials, specifically zeolites, can also be used
to modify IT processes at the L/L interface. Aluminous
zeolites, like L/L interfaces, are ionic conductors. Two possible applications follow immediately from the modification of
This journal is
c
the Owner Societies 2006
Fig. 5 (a) Typical hydrodynamic voltammograms obtained when the
membranes of Fig. 4 are used to stabilise the L/L interface in a rotating
configuration; (b) the mass-transport (rotation frequency, o) dependence of the resultant limiting current. Tetraethylammonium IT was
employed in these experiments, see ref. 54 for details. Reprinted with
permission from ref. 54, r (2002) American Chemical Society.
Phys. Chem. Chem. Phys., 2006, 8, 1869–1883 | 1873
tive) materials in situ at the L/L interface, conceivably using
electrochemical methods to control the deposition process (see
below). We note that Maraček and co-workers have reported
the formation of silicate films at the electrified L/L interface,
using transfer of the templating ion to initiate the deposition
process.68 The key question with this type of study is the extent
to which the variable potential difference can be used to
control the morphology of any resultant film. For the moment, this question remains unanswered.
Electron transfer at the L/L interface: metal deposition at the
bare interface
Fig. 6 (a) A typical hydrodynamic voltammogram obtained when the
membranes of Fig. 4 are used to stabilise the L/L interface in a channel
flow configuration; (b) the mass-transport (volume flow rate, Vf)
dependence of the resultant limiting current. Again tetraethylammonium IT was employed, see ref. 55 for full details. Reprinted with
permission from ref. 55, r (2003) American Chemical Society.
the L/L interface with a microporous material. The first is the
ability to control ion-exchange equilibria within zeolite materials through the potential difference imposed on the L/L
interface. The exchange of charge-balancing cations within
zeolites is of considerable technological importance, for example in waste-water purification, catalysis and gas separation.61,62 We have recently illustrated this phenomenon using
an aqueous suspension of zeolite Y in contact with an organic
phase, containing a ligand capable of extracting the native ion
(sodium) from the zeolite.63 The second application is the
concept of size-selectivity at the L/L interface. Clearly, IT can
be controlled potentiostatically through the potential difference applied to the interface. However, if the interface is
modified with a continuous material, of a pore size similar
to the ionic diameter, then ionic size becomes a second degree
of selectivity. We have illustrated this effect with silicalite
membranes (a neutral framework zeolite, with a pore dimater
of ca. 0.6 nm), showing that tetra-alkylammonium ions can be
transferred, or rejected, based wholly on their size and that
information on facilitated IT processes can also be obtained.64–67 Future prospects for this work might include the
deposition of size-selective (or indeed charge or shape selec1874 | Phys. Chem. Chem. Phys., 2006, 8, 1869–1883
Electrochemical techniques can also be used to probe the
transfer of electrons across the L/L interface. Since the mid1990s, there has been enormous progress in our understanding
of heterogeneous ET at this interface, due largely to the
application of the scanning electrochemical microscope and
impedance techniques to dark- and light-induced ET, respectively.69–74 In the light-induced case, the modification of the
interface via the supramolecular assembly of porphyrin species
has been shown to contribute to the overall efficacy of the ET
process.73 Restrictions on space prevent a more complete
discussion of this work, however we note that there are some
prior reviews of these topics.75,76
In the Introduction, it was noted that metal deposition at
the interface between liquid phases (with the metal precursor
and reducing agent located in different phases) could be
considered as a useful intermediate case between solution
phase metal reduction (relevant to many preparations of
metallic nanoparticles) and conventional electrodeposition.
In this section, we discuss this process in more detail, and
consider the ways in which the analogy to these more familiar
metal deposition processes holds.
Michael Faraday studied the formation of colloidal gold
particles at the L/L interface in the 19th century, using a
carbon disulfide solution of phosphorus to reduce aqueous
solutions of gold chloride. Faraday noted that ‘‘dark flocculent’’ deposits were formed, although lacking today’s tools of
structural characterisation, he could only conclude that the
metal deposit must be in a ‘‘fine state of division’’.77 This
experiment represents an interfacial reduction (and associated
metal deposition), at a non-polarised L/L interface, with the
phases remaining electro-neutral overall due to the co-transport of phosphate anions and protons. The spontaneous
reduction of gold at organic/water interfaces is still an active
research area almost 150 years later, with characterisation and
control over the deposit size, along with interfacial assembly of
the deposit, as the lead areas of investigation. In fact, reports
describing the reduction/assembly of metallic and/or semiconductor particles at L/L interfaces are appearing with increasing frequency,15,78 therefore a brief survey of the literature in
this area is timely. We also note that the assembly of diverse
types of (generally pre-formed) particles at L/L interfaces has
recently been the subject of a brief review.79
One of the recently reported examples is the interfacial
reduction and assembly of gold nanoparticles, described by
Rao et al., using toluene as the organic solvent. Nanoparticulate films of gold, silver and copper were formed at the
This journal is
c
the Owner Societies 2006
interface: an increase in the contact time of the two phases led
to an increase in film coverage, but gave no apparent increase
in the mean diameter of the particles.80 Rao has also extended
his approach to the deposition of thin films of metal oxides
and sulfides at the L/L interface. This method appears to rest
upon the metal precursor being present in the organic phase.
Moreover, X-ray photoelectron spectra suggest that the nanoparticulate form of the deposit is due to reaction products
of the phosphonium reducing agent binding to the deposit
surface.81 Rao states that a close-packed film of nanoparticles
can be formed, at least with gold, using this approach. A study
by Vanmaekelbergh et al. has shown how addition of ethanol
to a two-phase (water/heptane) suspension of anionicallystabilised gold nanoparticles leads to progressive formation
of an interfacial gold film.82 Note that in this latter case, the
gold deposit is prepared as an aqueous phase suspension,
which adsorbs at the L/L interface following introduction of
the ethanol. Importantly in this case, the aggregation of the
deposit (see below) to form an ill-defined structure does not
occur, the stabilisation being attributed to electrostatic repulsion between the particles. Another recent communication has
described a L/L interface-based transformation of noble metal
nanoparticles into nanowire ‘‘networks’’, on agitation of the
hydrosol with toluene.83 Similarly, extended silver wire structures have been formed at L/L interfaces, with the ratio of
metal precursor to electron donor reported in this case to
influence the deposit geometry.84 Earlier work by Efrima and
co-workers described the formation of so-called ‘‘liquid-like’’
films of silver by reduction of aqueous silver nitrate solutions
in the presence of certain surfactants. As with the work of
Vanmaekelbergh, the silver deposit is formed with an aqueous
phase reductant, and subsequently assembled at the L/L interface. Subsequent structural studies via transmission electron
microscopy (TEM) and Moiré deflectometry suggested that
the silver films consisted of aggregates of small (several nanometres diameter) particles.85,86 Efrima et al. subsequently
employed reflectance spectroscopy in combination with surface pressure isotherms of the ‘‘liquid-like’’ interfacial silver
films, and re-affirmed that the films consisted of micron-scale
flocs of nanometre scale silver particles.87 A very recent article
by Spain et al. describes the assembly of silver films at the
ethanol–water/chloroform interface: once more the silver particles are formed in the aqueous phase, and assemble on
contacting this phase with the organic layer. In this case, a
specific thiol ligand is required to induce interfacial film
formation, although the factors driving this assembly process
are not entirely clear.88 Another recent report shows that
aqueous suspensions of gold and iron(III) oxide nanoparticles
can also be induced to assemble, in this case at the water/
toluene interface, on particle treatment with bromo-propionate terminated ligands.89 The amphiphilic nature of the
modified particles, with a water/toluene contact angle close
to 901, is reported to drive the interfacial assembly. Nanoparticulate interfacial assembly has been exploited as a means to
form nanoparticulate heterodimers.90 It is important to distinguish between these reports, where an interfacial layer of
metal is formed, and the seminal work of Brust et al. who
formed thiol-protected Au particles through liquid/liquid
phase transfer, followed by a homogeneous reduction proThis journal is
c
the Owner Societies 2006
cess.91 A further distinction is the locus of the particle assembly process: reduction in the above ‘‘film formation’’ cases is
assumed to be homogeneous, with subsequent interfacial
assembly, as opposed to the cases discussed below where both
processes are heterogeneous.
Given the difficulty in probing such complex interfacial
events, it seems logical to investigate metal deposition at the
electrified L/L interface. The great advantage of using the
electrified interface is that the rate of metal deposition can be
obtained from the interfacial flux (current). Specifically, the
voltammetric techniques discussed in the context of IT can be
applied to deposition at the L/L interface. Consequently, the
driving force for deposition can be varied through the potential difference applied to the L/L interface. The first report of
metal deposition at the electrified L/L interface described the
‘‘metallization’’ of the water/1,2-dichloroethane (DCE) interface on passage of a current in the presence of cupric ions
(aqueous phase) and an organic phase reducing agent. A
similar effect was seen when silver ions were present in the
aqueous phase.92 This report remained as something of a
curiosity, until Cheng and Schiffrin described heterogeneous
gold reduction at the water/DCE interface using a fourelectrode potentiostat, permitting accurate control over the
interfacial potential.12 This work attempted to demonstrate
that the process was truly heterogeneous, and to correlate the
structure of the resultant deposits with the growth conditions.
In situ visible absorption experiments were also performed
which suggested, by approximate comparison with Mie
theory, that nanometre scale gold particles were formed. Of
course this type of experiment, where solids nucleate and grow
at a L/L interface has a long pedigree, which can be traced
back to Faraday’s aforementioned work.77 The difficulty that
Faraday encountered remains however, namely characterisation of the solid phase that is formed at the now-modified L/L
interface, given the general lack of applicability of most high
resolution microscopic techniques to this interface (q.v.).
Cheng and Schiffrin provided one electron micrograph (scanning electron microscopy, SEM) of the gold deposits,12 but the
relationship between the deposit morphology in vacuo and that
in solution is not clear. In particular, the interfacial distribution of the particles (mean separation and array structure, if
any) is almost certainly lost on solvent removal. The micrograph of the electrodeposited gold revealed micron scale
particles, but on the basis of subsequent work and the spectroscopic data of Cheng and Schiffrin (vide infra), it seems likely
that these structures are formed from the agglomeration of
smaller, initially formed particles.
Schiffrin noted that the L/L interface should represent a
model system for the investigation of electrochemical phase
formation,12 since in its pristine state the interface is free of
defect sites, as noted in the discussion of IT kinetics above.
Such parameters can be extracted for deposition at metal/
electrolyte interfaces, but because of the strong interaction
between the substrate and the initial form of the deposit, it is
difficult to attach any generality (i.e. substrate independence)
to the parameters derived by this method. Application of
analogous models to deposition at the electrified liquid/liquid
interface could allow the ‘‘substrate-free’’ critical cluster size of
a given metal to be measured, implying that quantitative
Phys. Chem. Chem. Phys., 2006, 8, 1869–1883 | 1875
information derived for deposition at this interface would
yield general information on how metals form and grow from
solution. A subsequent paper by Kontturi, Schiffrin and coworkers derived models for the current response associated
with deposit growth, by solving the transport equations
describing diffusion of the metal precursor and the electron
donor from their respective phases.13 Comparison was made
between the predictions of the model and an experimental
system (reduction of aqueous phase tetrachloropalladate by
organic phase butylferrocene), however the agreement between the two was quite limited. The authors attributed the
discrepancy between theory and experiment to the role of
unconstrained particle aggregation in altering the initial distribution of the deposits. The role of aggregation was highlighted by a subsequent galvanostatic study of palladium
deposition by the same research group.93 This point is important as it once more highlights the need for detailed
structural characterisation, in this case to ascertain the extent
to which aggregates form. An interesting way of probing the
aggregation phenomenon has been proposed by Sastry and coworkers.94 A dynamic L/L interface was formed by injecting
the aqueous phase in a quasi-two-dimensional (Hele-Shaw)
configuration. The aqueous phase contained pre-formed Au
particles, which assembled on contact with the di-amine present in the organic phase. It was shown that a more ‘‘open’’
aggregate was formed when a more viscous organic phase was
employed, highlighting a possible route to control interfacial
particle assembly. It should be noted that Cunnane and coworkers have also described the formation of polymer phases,
under electrochemical control, at the L/L interface.95 Overall,
we can say that the beneficial effect of using a fluid interface as
a ‘‘defect-free’’ system for the nucleation and growth of
metallic phases is counter-balanced by the uncontrolled ‘‘ripening’’ of deposits, which tends to result from the fluidity of
the interface.
A further distinction can be made between the truly heterogeneous process, where the potential is applied directly to the
L/L interface leading to interfacial deposition, and a comparable—yet distinct—‘‘almost heterogeneous’’ process. In the
latter case, a deposit is formed by bringing a (metallic)
electrode close to a L/L interface. As these deposits grow
from the electrode, the proximity of the interface serves to
constrain growth to two dimensions, allowing correlations to
be made between fractal dimension, for example, and deposition potential.96–98 A slightly different case is the recent work
by Mirčeski, using Anson’s approach99 to spread a thin
organic film on a graphite electrode, which is then contacted
with an aqueous phase. Spontaneous formation of silver at the
L/L interface results from contact of a silver solution with a
thin film containing an appropriate reducing agent. Film
formation can be indirectly controlled through the electrochemical regeneration of the organic reducing agent. Dense silver
films have been reported, which show ‘‘blocking’’ behaviour
towards IT at the L/L interface but display a catalytic effect on
ET processes at this interface.100An earlier report describes the
catalytic effect of interfacial Pd particles on L/L ET processes
using impedance techniques.101
To close this section, we consider a study by Unwin et al.,
using both micron and nanometre diameter pipettes to restrict
1876 | Phys. Chem. Chem. Phys., 2006, 8, 1869–1883
the extent of silver deposition at the L/L interface.102 Studies
of the deposition of metals on (solid) electrode surfaces have
used microelectrodes in an attempt to reduce the number of
nucleation sites toward one, and thereby simplify the analysis
of the associated current response by eliminating the need to
consider the overlap between multiple diffusion fields.35,103
The study of silver deposition at the pipette-defined water/
DCE interface included some SEM of the deposits, which
again suggested that the deposits were on the micron scale.
Importantly this communication also provided a preliminary
comparison of electrochemical and structural (SEM) data,
suggesting that around seven silver nuclei were formed in a
two micron radius pipette,102 although it is not immediately
clear if deposit aggregation can be distinguished from the
formation of individual nuclei.
Electron transfer at the L/L interface: metal deposition at the
modified interface
This section deals with the effect of a modifying material, in
particular the membrane structures described earlier, on metal
deposition at the L/L interface. In fact, the distinction between
modified and ‘‘bare’’ interfaces of the previous section may be
considered to be rather arbitrary, because the formation of
even a single metal nucleus at the L/L interface will of course
modify the interface. Prior to the work of Unwin et al.
discussed in the preceding section,102 my group had reported
an alternative method of restricting metal deposition at the L/
L interface.104 This approach consisted of modifying the
interface with the types of membrane material discussed earlier
in the context of IT. A hydrophilic membrane should transform the interface into an ensemble (or array) of small
aqueous phases bounded by the organic phase, whereas an
ensemble of organic phases in contact with the aqueous phase
would result from the use of a hydrophobic membrane material. The studies we have reported to date pursued the former
approach, with g-alumina being used to form ensembles of
interfaces with individual radii of 50 nm (or less), which dictate
the maximum size of the resultant metallic deposits.14,104–106
This approach has a parallel in the electrodeposition studies
on solid surfaces referred to above: g-alumina membranes and
related materials have been coated on one face with a metal,
and thus used to ‘‘template’’ the deposition of metallic ‘‘nanorods’’ from solution by a number of research groups.26,107,108
The approach has been extended to the formation of metallic
‘‘bar-codes’’, by successive immersion of the membrane-modified electrode in solutions of the corresponding metal ions.109
The beauty of applying template deposition to the L/L interface is that contact between the two liquid phases and the
membrane can be made spontaneously. By contrast, the
modification of a solid surface for template deposition using
these mesoporous membranes, generally requires that the solid
phase is deposited via sputter-coating or via an electroless
deposition process to ensure good electrical contact between
this phase and the membrane.110,111
In addition to providing a L/L based route to the formation
of nano-structured metals, our motivation for using the membrane template was three-fold: 1. the restriction on the growth
of the metal imposed by the pore dimensions means that the
This journal is
c
the Owner Societies 2006
Fig. 7 Characterisation of the palladium deposits formed at the
alumina-templated L/L interface, via (a) TEM and (b) SEM. TEM
is performed by microtome slicing of the metal loaded membrane.
SEM is performed by dissolution of the alumina and trapping
the liberated metal on a polyester membrane. See ref. 14 for further details. Reprinted from ref. 14, r (2003) with permission from
Elsevier.
influence of lateral particle aggregation should be reduced;13,93
2. the localisation of the particles, imposed by the pores,
means that the original spatial distribution of the particles is
retained, and can be assessed ex situ; and 3. the thickness of
the hydrophilic g-alumina membrane means that, at least for
short timescales, the diffusion fields of the metal precursors are
retained within the pores of the membrane. This final feature
has two benefits: firstly, adjacent diffusion fields (from nucleation events in neighbouring pores) cannot interact with one
another and are thus de-coupled over these times, and secondly, the de-coupled diffusion process can be approximated
as a linear flux due to the aspect ratio of the pores. These
advantages of template deposition, over deposition at the
unmodified L/L interface, have enabled us to gain some
insight into the L/L deposition process using palladium and
platinum as ‘‘model’’ deposition systems.14,105,106 Specifically,
the first finding was that the g-alumina membranes could
indeed be used to localise particle growth within the pores of
the membrane. This was proven by taking images of the
This journal is
c
the Owner Societies 2006
membrane, in cross-section, using an ultra-microtome to make
TEM possible. Micrographs, from the TEM, of aluminatemplated platinum and palladium deposits are shown in
Fig. 7. The metallic deposits are seen to be located at the
mouth of the pores. In fact, the membrane itself was so
hydrophilic that a gravitationally inverted configuration where
the denser DCE phase was placed above the aqueous phase,
had to be employed to prevent the aqueous phase from
traversing the entire membrane. This conclusion was borne
out both by microscopy of the resultant metal deposits, and by
a separate analysis of the IT flux across the membrane
modified water/DCE interface under diffusion-only and hydrodynamic conditions.28,112 This behaviour had been sought,
however what was less expected was the ‘‘finely divided’’
nature of the metal deposits. TEM and X-ray diffraction
(XRD) measurements on the membranes following deposition
indicated that nanoparticulate aggregates of palladium and
platinum were formed, with the mean particle diameter being 5
and 4 nm, respectively.105 The platinum deposition process
was found to be slower, however the uniformity of the deposits
seen in both cases suggests that the deposits stop growing, on
reaching a certain size.
Noble metal nanoparticles can be formed readily by homogeneous reduction in bulk solution (see Introduction), but a
stabilising ligand is normally required to ‘‘cap’’ particle
growth. Similar observations for L/L deposition were made
by Rao and co-workers, in their aforementioned study of
spontaneous metal reduction using an aqueous phase reductant.80,81 The gold films, in particular, were characterised in
some detail and shown to consist of nano-particulate deposits,
with a mean diameter of 9 nm (at the particular reagent
concentrations used) and a relatively narrow particle size
distribution (see Fig. 8). Again, such deposits were obtained
in the absence of a ligand specifically added to ‘‘cap’’ the solid,
although Rao has noted that the other species present in his
system may fulful this role.81 The mean deposit size appeared
to be time-invariant, suggesting that some factor encourages
the deposits to stop growing on attaining a certain size.
However, an in-depth quantitative analysis of this phenomenon has yet to be performed. The main change observed by Rao
on increased contact time was an increase in the interfacial
coverage of the deposit. In the next paragraph, we give some
consideration to the factors influencing deposit geometry at
the L/L interface, in comparison to deposition at the S/L
interface and in bulk solution phases. To do so requires some
consideration of the growth modes of metallic particles.
Superficially at least, the L/L deposition case may be
considered in the same way as electrodeposition on a solid
substrate. The morphology of electrodeposits is strongly
linked to the interactions between the nascent deposit and
the underlying substrate (the electrode surface). Three types of
classical growth mechanism have been identified:113 1. Volmer–Weber growth, where there is only weak interaction
between the substrate and the new phase. This effect can be
quantified via the binding energy of the atoms of the new
phase (M) to the substrate (S), which would have to be lower
than that of the new phase to itself (M). Consequently, the
deposit tends to grow as 3-D ‘‘islands’’, growing upwards
rather than outwards; 2. Frank–van de Merwe growth results
Phys. Chem. Chem. Phys., 2006, 8, 1869–1883 | 1877
edges, to form nanowires, is possible using appropriate conditions.115,116 Penner has also given considerable thought to
the issue of nanoparticle growth on such surfaces.114 The
Volmer–Weber case represents a good system for studying
particle distribution, because of the tendency to form isolated
clusters, rather than 2-D films, on the substrate. The random
spatial (and temporal) nature of nucleation on a solid surface
means that there will always be a considerable spread in interparticle (e.g. nearest neighbour) separations. This fact means
that particle diffusion fields will become coupled (overlap) at
different times. Given that an isolated particle will be exposed
to a higher flux of material (see foregoing discussion on
microelectrode arrays/ensembles), the particles will tend to
diverge in size, essentially because their respective diffusion
fields become coupled at different stages of their growth. This
hypothesis has been borne out using Brownian dynamics
simulations of particle growth on surfaces.117 The only way
to ensure a uniform flux, and hence a uniform growth rate, to
each growing particle is to ensure that the particles have a
uniform spacing on the substrate, i.e. deposition should
proceed from an array, as opposed to an ensemble, of nuclei.
However, given that nucleation occurs from defect sites,
formation of an array is unrealistic. Penner contrasts the
electrodeposition case with the growth of colloidal particles
in homogeneous solutions (viz. nanoparticle formation),114
quoting laws for surface-reaction controlled growth:
r = kV mc0t
(3)
and diffusion-controlled growth:
r¼
Fig. 8 TEM micrographs for gold deposited at the L/L interface as a
function of contact time: (a) 3 h, (b) 6 h and (c) 9 h. The scale bar
corresponds to 50 nm in each case. See ref. 80 for further details.
Reprinted with permission from ref. 80, r (2003) American Chemical
Society.
where there is strong interaction between the substrate and the
new phase (i.e. the M–S binding energy exceeds the corresponding M–M parameter), with negligible lattice mismatch
between the two materials, hence a smooth 2-D film tends to
result; and 3. Stranski–Krastanov growth is found when the
strong new lattice–substrate interaction is accompanied by a
lattice mismatch, meaning 3-D islands of M tend to form on a
strained 2-D layer of M on S. Clearly cases 2 and 3 correspond
to ‘‘underpotential’’ deposition phenomena, as seen for silver
deposition on both Au(100) and Au(111) surfaces (case 2) or
copper deposited on a Au(111) electrode (case 3).113 The
deposition of noble metals on highly oriented pyrolytic graphite (HOPG) surfaces is a good example of Volmer–Weber
growth.114 The question is, ‘‘does deposition at the L/L interface fit into any of these classes?’’
Penner and co-workers have studied electrodeposition on
HOPG extensively and shown that selective deposition at step1878 | Phys. Chem. Chem. Phys., 2006, 8, 1869–1883
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2Vm Dc0 t
ð4Þ
In the above equations, r represents the particle radius, t is
time elapsed since the nucleation of the particle, Vm is the
molar volume of the deposited metal, c0 is the bulk concentration of the metal precursor, D is the diffusion coefficient of this
species and k is the pseudo-first order rate constant describing
particle growth. The time-derivative of the diffusion controlled
equation follows an inverse square root dependence on time,
thus particle growth should slow with time, permitting a
‘‘focussing’’ of particle dimensions and hence uniform particle
sizes. This uniformity arises from the averaged nature of the
particle distribution: although there will be local fluctuations
in the distributions of nearest neighbours around a given
growing particle, over time each particle will experience the
same mean flux, due to their translational freedom.
These cases can now be compared with the case of metal
deposition at the L/L interface. Deposit–substrate interaction
could be perceived to be weak, since no strong chemical
interaction would be expected between a growing palladium
nucleus and an organic solvent, for example. Thus, L/L metal
deposition might be expected to follow the Volmer–Weber
mode of growth (weak substrate interaction), with small 3-D
clusters of deposits seen, rather than smooth 2-D films. To a
degree, this statement is consistent with some of the experimental evidence available to date,14,105 but when considering
the subsequent growth of the particles, the apparent uniformity
of the deposits formed at the L/L interface is notable. Why is
this difference from the case of deposition on a surface such as
This journal is
c
the Owner Societies 2006
HOPG seen? A plausible explanation is the relative mobility of
the particles grown at the L/L interface. As with the bulk
solution case, the particles are able to attain a uniform mean
separation, which exposes them to the same mean flux of
reactants. This implicitly assumes that the nucleation process
is instantaneous: if it is not (as we suspect, given the observed
slow kinetics of platinum deposition at the water/DCE interface and the increased coverage of gold seen by Rao et al.80,105)
the question is, ‘‘what stops particle growth?’’, given the
apparent uniformity of deposits.
The thermodynamics of the particle–interface interaction
have been overlooked in the preceding discussion. The assumption made implicitly above is that the interactions between the growing metal particle and the fluid substrate are
weak, however this is incorrect: the important factor is not the
particle–solution interaction but the balance of interactions
between the particle and surrounding liquids, compared with
the interactions at the L/L interface alone (i.e. before the
particle formed). In fact, this area has been considered in
some detail by colloid scientists, who have known for many
years that particles can be used to stabilise L/L interfaces,
leading to so-called Pickering (or Ramsden) emulsions.118 In
simple terms, the particle stabilises the interface because it
minimises the extent of interactions between the two immiscible liquids: particles can thus become strongly adsorbed at the
L/L interface. Young’s equation, relating the contact angle
between the solid and the adjacent liquids to the interfacial
tensions of the L/L and S/L interfaces, can be modified to
include the line tension, t, at the zone of three-phase contact
(see Fig. 9):119
gS=W gS=O
t
gO=W ¼ r sin W
cos W
ð5Þ
where, as defined in Fig. 9, the g terms give the interfacial
tension of the solid/aqueous (S/W), solid/organic (S/O) and
organic/water (O/W) interfaces, respectively, and y is the
contact angle (defined with respect to the organic phase) at
the three-phase boundary. The thermodynamic analysis of
Aveyard and Clint has shown that the interfacial adsorption
of solid particles, illustrated in terms of the contact angle and
line tension, leads to three regions where the particle is stable,
unstable or metastable with respect to adsorption. The metastable zone corresponds to a local interfacial minimum, where
an activation energy is required to desorb the particle from the
interface. A problem with the experimental verification of the
adsorption models is the present lack of accurate line tension
data. Notwithstanding this, comparison can be made between
experimental and calculated data for the desorption of particles from L/L interfaces. Simple models of particle adsorption
permit typical adsorption energies at the L/L interface to be
determined.118 Particle adsorption energies are ca. 103 kbT
(where kb is the Boltzmann constant and the gO/W values are
appropriate for the water/DCE or water/toluene interfaces)
for micron-scale particles with contact angles close to 901,
meaning such particles are in effect irreversibly adsorbed at
ambient temperatures. The adsorption energies scale with the
square of the particle radius.6,118 Hence for smaller particles
and/or those with either a very low or a very high contact
angle, adsorption energies are much smaller, lying close to kbT
for particles with diameters of a few nanometres. Adsorption
thus becomes reversible: the size-selective displacement of
nanoparticles by their larger brethren is an interesting consequence of the adsorption equilibria. Dinsmore and co-workers
have exploited this phenomenon experimentally, with CdSe
nanoparticles of ca. 5 nm diameter shown to displace ca. 3 nm
diameter particles from the water/toluene interface.6
The above work deals with adsorption of pre-formed particles at L/L interfaces, but can shed some light on the factors
affecting particle growth (and subsequent adsorption) at the
L/L interface. Conventionally, nanoparticulate materials are
treated as being thermodynamically unstable, because of their
high surface area—hence the common use of ligands to ‘‘cap’’
such materials. At the L/L interface, however, the intrinsic
high surface area may serve to stabilise the particles, the
energetic cost in terms of the higher interfacial tension at the
particle/solution interface being outweighed by the stabilisation induced by pushing the two immiscible phases apart.
Many questions remain unanswered in this area, largely
because of the lack of quantitative studies of deposition at
L/L interfaces. Although we have highlighted the importance
of the interfacial tension in stabilising deposits at the L/L
interface, the effect of this parameter on deposit morphology is
difficult to determine directly for the case of electrodeposition
at the L/L interface. This ambiguity arises because the interfacial tension is a function of the potential (f) applied to the
L/L interface, the simplest relation being Lippmann’s classical
electro-capillarity expression:120
@ 2 gO=W
@f2
Fig. 9 Definition of the line tension, contact angle and interfacial
tension terms used in the analysis of particle adsorption at the L/L
interface (adapted from ref. 119).
This journal is
c
the Owner Societies 2006
¼
@s
¼ Cd
@f
ð6Þ
where Cd is the double layer capacitance of the L/L interface
and s is the interfacial charge density. One report has explicitly
addressed the effect of added surfactant on the electrodeposition of Pd at the L/L interface, showing a correlation between
an enhanced barrier to nucleation (larger applied overpotential) and the presence of an adsorbed phospholipid at the L/L
interface.121 Again, however, no detailed quantitative treatments of surfactant addition have been reported. A parameter
Phys. Chem. Chem. Phys., 2006, 8, 1869–1883 | 1879
that would be of intrinsic interest is the nucleation rate at the
L/L interface and its dependence on applied potential. Once
more, the only published work in this area to date is limited to
placing a lower bound (of 104 cm s1) on the rate constant
for palladium formation at the water/DCE interface.121 In
particular, the application of classical electrodeposition models to the L/L interface to extract critical cluster sizes, would be
of tremendous value in assessing the experimental validity of
these models. In classical models of phase formation, the
Gibbs energy of the growing phase goes through a maximum
as the number of atoms (N) increases due to the competing
bulk (stabilising) and surface (de-stabilising) terms. For electrodeposition, one can vary the driving force (supersaturation)
externally via the applied potential difference, giving rise to the
picture of Fig. 10 where the critical cluster size, Ncrit, is defined
as the Gibbs energy maximum,113 i.e.:
@DG
¼0
@N
ð7Þ
These studies would be of value in their own right, however
this interface is also a useful testing ground for studies of
solution phase nucleation in general, which would lend added
weight to work of this kind. Earlier we noted that particle
agglomeration complicated the analysis of deposition current
transients, due to its effect on the particle distribution (and
resultant mass-transport). Typical voltammetric and chronoamperometric responses for palladium deposition at the
alumina-templated L/L interface are shown in Fig. 11. We
have recently attempted to analyse current transient data from
these templated interfaces, based on the assumption that
agglomeration is restricted by the presence of the template.
The transient current response showed a marked dependence
on applied overpotential. Analysis of the data in terms of
classical electrodeposition models, developed for solid surfaces, gives critical cluster sizes of ca. 0.5 atoms (the exact
value being a function of the overpotential applied). Critical
cluster sizes below unity have been reported in the literature
previously for deposition on solid surfaces, for example a
recent detailed paper deals with the case of palladium deposi-
Fig. 10 The classical model of deposit Gibbs energy as a function of
the number of its constituent atoms (N) for low and high overpotentials (Z), given by the broken and solid lines, respectively.
1880 | Phys. Chem. Chem. Phys., 2006, 8, 1869–1883
Fig. 11 (a) Cyclic voltammetric response for palladium deposition at
the alumina-templated L/L interface, with no organic phase electron
donor (trace a), and added electron donor (trace b); (b) Potential step
response for palladium deposition in the same system. The responses
in both cases are characteristic of nucleation processes, see ref. 14 for
further details. Reprinted from ref. 14, r (2003) with permission from
Elsevier.
tion on HOPG.122 The usual interpretation of this phenomenon is that the defect site of the substrate itself acts as the
nucleus for the formation of the deposit: it is difficult to
conceive how this situation could arise at the L/L interface.
However, at the L/L interface the data interpretation implies
that the new phase is (almost) always more stable than the
bare interface. This statement may be consistent with the
earlier discussion, i.e. the excess surface energy of the particle
is outweighed by the stabilisation achieved from reducing the
contact area of the L/L interface. It will be interesting to
extend this analysis to a wider range of experimental parameters (e.g. other metals, other types of liquid interface, and
the presence of other species such as surfactants to probe the
effect of interfacial tension) to see if similar critical cluster sizes
are obtained.
A second factor requiring further exploration is the timescale of the agglomeration process compared to the rate of
particle growth, and the sensitivity of agglomeration to solution phase composition. It is noteworthy that the gold deposit
formed by Vanmaekelbergh was stable with respect to agglomeration, a fact attributed to the lack of screening electrolyte
present in the organic phase.82 Such electrostatic effects have
also been invoked for the assembly of microscopic polymer
particles at L/L interfaces (see below).19,123 A final comment is
that the assembly of pre-formed particles at the L/L interface
often has to be induced (e.g. by control over capping ligand
identity or through compositional control of the particle
charge,82,89) whereas the assembly of particles formed at the
L/L interface appears to be spontaneous and, apparently,
rather insensitive to composition. If this distinction is indeed
This journal is
c
the Owner Societies 2006
genuine, its origins are not immediately clear. One could
imagine that particle adsorption represents such a deep potential well, that the particles growing at the interface tend to a
form that maximises their extent of adsorption. This effect of
the interface on the particle, invoked above to account for the
nanoparticulate morphology, should be manifested in the
surface chemistry with hydrophilic and hydrophobic domains
apparent on opposite particle faces (i.e. ‘‘Janus’’ particles).
Once more, such detailed structural investigations have yet to
be performed. Alternatively, bearing the work of Aveyard and
Clint in mind,119 the particles are conceivably meta-stable with
respect to their interfacial adsorption.
Particle assembly at liquid/liquid interfaces: some general
considerations
The recent mini-review by Binder surveys some of the latest
work on the assembly of pre-formed particles at the L/L
interface. Binder highlights the dynamic nature of assembly
at the L/L interface which, he believes, should allow ‘‘errors [in
the assembly] to be corrected rapidly’’.79 Until very recently,
the bulk of the work in this area had been undertaken by
colloid scientists and, notably, had shown that spontaneous
long-range ordering could be induced by the adsorption of
micrometer scale particles (e.g. of polystyrene latex) at the
interfaces between water and low polarity organic solvents,
such as n-octane.18,123,124 There has been some debate over the
physical origin of this long-range order: residual surface
charges on the assembled particles, which are un-screened by
the low polarity organic solvent, are believed to underlie the
observed long-range order.123 One study highlighted capillary
distortion of the interface due to the dipolar field as giving rise
to the stable long-range order observed for micron-scale
polymer particles at L/L interfaces.19 Although this work
was subsequently criticised,125 a long-range (logarithmic) dependence of interfacial distortion on distance has been derived
and verified experimentally.126
A related area of debate within this field has been the
importance of added electrolyte in determining the stability
of the particle assemblies. High concentrations of salt in the
aqueous sub-phase tend to collapse assemblies on compression
of the L/L interfaces (by screening inter-particle repulsion),
although the electrolyte effect is more pronounced at aqueous/
air interfaces.18 A more recent report has stated that the interparticle force is insensitive to aqueous electrolyte concentration.124 To the author’s knowledge, no studies to date have
attempted to assess the validity of the screening theory by
assessing the effect of electrolyte added to the organic phase on
particle assembly. It could be extremely interesting to introduce the externally applied electric field along with/instead of
the surface pressure (normally employed by the colloid scientists) as an extra degree of freedom controlling particle assembly. Whether such experiments are, indeed, possible remains to
be seen. The long-range order observed, under certain conditions for micron-scale particles at L/L interfaces contrasts with
the situation for nanometre scale particles. In the latter case,
the existence of close-packed ‘‘films’’ has been demonstrated,
but no order is believed to be present.127 Finally, we note that
the reversible potential-induced assembly of gold and titanium
This journal is
c
the Owner Societies 2006
dioxide nanoparticles at L/L interfaces has been reported.15,78
Electrochemical data, backed up by quasi-elastic light scattering, has been used to probe the local surface excess of the
particles on the aqueous side of the L/L interface. Given the
small particle size, one interesting element is missing thus far,
namely data on inter-particle interactions, specifically whether
an array or merely an ensemble of particles exists.
Conclusions and outlook
The unique properties of the L/L interface make it an ideal
environment for the deposition (via assembly of pre-formed
particles, or interfacial growth) of micron and nano-particulate materials. Experiments of this type have been reported for
insulating, semiconducting and conducting materials. Most
interestingly, spontaneous ordering phenomena on the micron
scale have been demonstrated under certain conditions. Electrification of the L/L interface may prove to be a powerful tool
to understand the growth/assembly process, since the driving
force for particle deposition can be altered at will. The main
challenge in this area is to develop an effective phase diagram,
describing particle ordering as a function of surface area and
surface charge (possibly via the applied potential). A second
challenge is to obtain a better understanding of the dynamics
of the interfacial assembly process. We conclude by noting
that, as in Faraday’s day, the difficulty is to overcome the
general inapplicability of many of the techniques routinely
used for analysis of other classes of interface. However, the
recent imaginative use of in situ X-ray scattering and fluorescent microscopy techniques127 may point the way to future
advances in this area.
Acknowledgements
I would like to thank the members of my research group who
have worked in this area, particularly Mark Platt and Brett
Kralj. I am also grateful to Prof R. M. Penner (University of
California Irvine) and Prof H. H. Girault (Ecole Polytechnique Fédérale de Lausanne) for their hospitality, while I visited
their laboratories during 2004. Finally, I would like to thank
Anne Juel for her assistance with the preparation of this
article.
References
1 A. W. Adamson, Physical Chemistry of Surfaces, Wiley, Chichester, 1982.
2 L. F. Scatena, M. G. Brown and G. L. Richmond, Science, 2001,
292, 908.
3 G. Luo, S. Malkova, S. Venkatesh Pingali, D. G. Schultz, B. Lin,
M. Meron, T. J. Graber, J. Gebhardt, P. Vanysek and M. L
Schlossmann, Faraday Discuss., 2005, 129, 23.
4 H. H. Girault, in Modern Aspects of Electrochemistry, ed. J. O.
Bockris, Plenum, New York, 1993, vol. 25, p. 1.
5 I. Benjamin, Chem. Rev., 1996, 96, 1449.
6 Y. Lin, H. Skaff, T. Emrick, A. D. Dinsmore and T. P. Russell,
Science, 2003, 299, 226.
7 C. Amatore, in Physical Electrochemistry: Principles, Methods
and Applications, ed. I. Rubinstein, Marcel Dekker, New York,
1995, p. 131.
8 R. A. Marcus, J. Chem. Phys., 2000, 113, 1618.
9 C. B. Murray, C. R. Kagan and M. G. Bawendi, Ann. Rev. Mater.
Sci., 2000, 30, 545.
Phys. Chem. Chem. Phys., 2006, 8, 1869–1883 | 1881
10 A. Taleb, C. Petit and M. P. Pileni, Chem. Mater., 1997, 9, 950.
11 C. J. Murphy and N. R. Jana, Adv. Mater., 2002, 14, 80.
12 Y. Cheng and D. J. Schiffrin, J. Chem. Soc., Faraday Trans., 1996,
92, 3865.
13 C. Johans, R. Lahtinen, K. Kontturi and D. J. Schiffrin, J.
Electroanal. Chem., 2000, 488, 99.
14 M. Platt, R. A. W. Dryfe and E. P. L. Roberts, Electrochim. Acta,
2003, 48, 3037.
15 B. Su, J. P. Abid, D. J. Fermı́n, H. H. Girault, H. Hoffmannová,
P. Krtil and Z. Samec, J. Am. Chem. Soc., 2004, 126, 915.
16 C. Wei, A. J. Bard and M. V. Mirkin, J. Phys. Chem. B, 1995, 99,
16033.
17 D. M. Kolb, in Advances in Electrochemical Science and Engineering, ed. R. C. Alkire and D. M. Kolb, vol. 7, Wiley-VCH,
Weinheim, 2002, p. 107.
18 R. Aveyard, J. H. Clint, D. Nees and V. N. Paunov, Langmuir,
2000, 16, 1969.
19 M. G. Nikolaides, A. R. Bausch, M. F. Hsu, A. D. Dinsmore, M.
P. Brenner, C. Gay and D. A. Weitz, Nature, 2002, 420, 299.
20 C. Gavach, T. Mlodnicka and J. Guastalla, Compt. Rendu Acad.
Sci. Ser. C, 1968, 266, 1196.
21 J. Koryta, P. Vanysek and M. Brezina, J. Electroanal. Chem.,
1976, 67, 263.
22 A. J. Bard and L. R. Faulkner, Electrochemical Methods: Fundamentals and Applications, Wiley, 2nd edn, Chichester, 2001.
23 K. Aoki, Electroanalysis, 1993, 5, 627.
24 D. W. M. Arrigan, Analyst, 2004, 129, 1157.
25 P. Ugo, L. M. Moretto and F. Vezzà, ChemPhysChem, 2002, 3,
917.
26 J. C. Hulteen and C. R. Martin, J. Mater. Chem., 1997, 7, 1075.
27 B. Kralj and R. A. W. Dryfe, Phys. Chem. Chem. Phys., 2001, 3,
5274.
28 M. Platt, R. A. W. Dryfe and E.P. L. Roberts, Langmuir, 2003,
19, 8019.
29 G. Taylor and H. H. Girault, J. Electroanal. Chem., 1986, 208,
179.
30 J. A. Campbell and H. H. Girault, J. Electroanal. Chem., 1989,
266, 465.
31 B. Liu and M. V. Mirkin, Electroanalysis, 2000, 12, 1433.
32 C. Cai, Y. Tong and M. V. Mirkin, J. Phys. Chem. B, 2004, 108,
17872.
33 X. Gao and H. S. White, Anal. Chem., 1995, 67, 4057.
34 J. K Campbell, L. Sun and R. M. Crooks, J. Am. Chem. Soc.,
1999, 121, 3779.
35 S. Chen and A. Kucernak, J. Phys. Chem. B, 2003, 107, 8392.
36 R. G. Compton, A. C. Fisher, R. G. Wellington, P. J. Dobson
and P. A Leigh, J. Phys. Chem., 1993, 97, 10410.
37 E. Budevski, G. Staikov and W. J. Lorenz, Electrochim. Acta,
2000, 45, 2559.
38 T. J. Davies, M. E. Hyde and R. G. Compton, Angew. Chem., Int.
Ed., 2005, 44, 5121.
39 J. B. Lewis, Chem. Eng. Sci., 1954, 3, 218.
40 W. J. Albery, A. M. Couper, J. Hadgraft and C. Ryan, J. Chem.
Soc., Faraday Trans. 1, 1974, 70, 1124.
41 G. J. Hanna and R. D. Noble, Chem. Rev., 1985, 85, 583.
42 C. J. Slevin and P. R. Unwin, Langmuir, 1999, 15, 7361.
43 B. Hundhammer, S. K. Dhawan, A. Bekele and H. J. Seidlitz, J.
Electroanal. Chem., 1987, 217, 253.
44 B. Hundhammer, T. Solomon, T. Zerihun, M. Abegaz, A. Bekele
and K. Graichen, J. Electroanal. Chem., 1994, 371, 1.
45 S. Wilke, H. Franzke and H. Müller, Anal. Chim. Acta, 1992, 268,
285.
46 N. Wilke, R. A. Iglesias, S. G. Chesniuk, S. A. Dassie and A. M.
Baruzzi, Bull. Chem. Soc. Jpn., 2002, 75, 235.
47 V. Maraček, H. Jänchenová, M. P. Colombini and P. Papoff, J.
Electroanal. Chem., 1987, 217, 213.
48 S. Wilke, M. D. Osborne and H. H. Girault, J. Electroanal.
Chem., 1997, 436, 53.
49 S. Sawada, M. Taguma, T. Kimoto, H. Hotta and T. Osakai,
Anal. Chem., 2002, 74, 1177.
50 H. J. Lee, C. M. Pereira, A. F. Silva and H. H. Girault, Anal.
Chem., 2000, 72, 5562.
51 C. Amatore, J. M. Savéant and D. Tessier, J. Electroanal. Chem.,
1983, 147, 39.
1882 | Phys. Chem. Chem. Phys., 2006, 8, 1869–1883
52 J. C. Eklund, A. M. Bond, J. A. Alden and R. G. Compton, in
Adv. Phys. Org. Chem., Academic Press, vol. 32, 1999, p. 1.
53 J. E. Vitt, D. C. Johnson and D. E. Tallman, Anal. Chem., 1993,
65, 231.
54 B. Kralj and R. A. W. Dryfe, J. Phys. Chem. B, 2002, 106, 6732.
55 S. S. Hill, R. A. W. Dryfe, E. P. L. Roberts, A. C. Fisher and K.
Yunus, Anal. Chem., 2003, 75, 486.
56 S. Peulon, V. Guillou and M. L’Her, J. Electroanal. Chem., 2001,
514, 94.
57 H. Ohde, A. Uehara, Y. Yoshida, K. Maeda and S. Kihara, J.
Electroanal. Chem., 2001, 496, 110.
58 C. N. Baroud and H. Willaime, Compt. Rendu Phys., 2004, 5, 547.
59 T. Tokimoto, S. Tsukahara and H. Watarai, Chem. Lett., 2003,
32, 940.
60 A. A. Kornyshev and M. Urbakh, Electrochem. Commun., 2004,
6, 703.
61 P. M. M Blauwhoff, J. W. Gosselink, E. P. Kieffer, S. T. Sie and
W. H. J. Stork, in Catalysis zeolites: fundamentals and applications, ed. J. Weitkamp and L. Puppe, Springer, Berlin, 1999, pp.
437–538.
62 I. E. Maxwell and W. H. J. Stork, in Studies in Surface Science
and Catalysis, Vol. 137: Introduction to Zeolite Science and
Practice, ed. H. van Bekkum, E. M. Flanigen, P. A. Jacobs and
J. C. Jansen, Elsevier, Amsterdam, 2nd edn, 2001, pp. 747–820.
63 M. J. Stephenson, S. M. Holmes and R. A. W. Dryfe, Angew.
Chem., Int. Ed., 2005, 44, 3075–3078.
64 R. A. W. Dryfe and S. M. Holmes, J. Electroanal. Chem., 2000,
483, 144.
65 G. C. Lillie, R. A. W. Dryfe and S. M. Holmes, Analyst, 2001,
126, 1857.
66 M. J. Stephenson, S. M. Holmes and R. A. W. Dryfe, Electrochem. Commun., 2004, 6, 294.
67 M. J. Stephenson, A. J. King, S. M. Holmes and R. A. W. Dryfe,
J. Phys. Chem. B, 2005, 109, 19377.
68 V. Maraček and H. Jänchenová, J. Electroanal. Chem., 2003, 558,
119.
69 T. Solomon and A. J. Bard, J. Phys. Chem., 1995, 99, 17487.
70 M. Tsionsky, A. J. Bard and M. V. Mirkin, J. Am. Chem. Soc.,
1997, 119, 10785.
71 J. Zhang and P. R. Unwin, J. Phys. Chem. B, 2000, 104, 2341.
72 D. J. Fermı́n, Z. Ding, H. Dung Duong, P. F. Brevet and H. H.
Girault, J. Phys. Chem. B, 1998, 102, 10334.
73 D. J. Fermı́n, H. Dung Duong, Z. Ding, P. F. Brevet and H. H.
Girault, J. Am. Chem. Soc., 1999, 121, 10203.
74 D. J. Fermı́n, H. Dung Duong, Z. Ding, P. F. Brevet and H. H.
Girault, Phys. Chem. Chem. Phys., 1999, 1, 1461.
75 S. Amemiya, Z. Ding, J. Zhou and A. J. Bard, J. Electroanal.
Chem., 2000, 483, 7.
76 F. Reymond, D. J. Fermı́n, H. J. Lee and H. H Girault, Electrochim. Acta, 2000, 45, 2647.
77 M. Faraday, Philos. Trans. 1, 1857, 147, 145.
78 D. J. Fermı́n, H. Jensen, J. E. Moser and H. H Girault,
ChemPhysChem, 2003, 4, 85.
79 W. H. Binder, Angew. Chem., Int. Ed., 2005, 44, 5172.
80 C. N. R. Rao, G. U. Kulkarni, P. J. Thomas, V. V. Agrawal and
P. Saravanan, J. Phys. Chem. B, 2003, 107, 7391.
81 C. N. R. Rao, G. U. Kulkarni, V. V. Agrawal, U. K. Gautum, M.
Ghosh and U. Tumkurkar, J. Colloid Interface Sci., 2005, 289,
305.
82 F. Reincke, S. G. Hickey, W. K. Kegel and D. Vanmaekelbergh,
Angew. Chem., Int. Ed., 2004, 43, 458.
83 T. Maddanimath, A. Kumar, J. D’Arcy-Gall, P. G. Ganesan, K.
Vijayamohanan and G. Ramanath, Chem. Commun., 2005, 1435.
84 F. Scholz and U. Hasse, Electrochem. Commun., 2005, 7, 541.
85 D. Yogev and S. Efrima, J. Phys. Chem., 1988, 92, 5754.
86 D. Yogev, M. Deutsch and S. Efrima, J. Phys. Chem., 1989, 93,
4174.
87 H. Schwartz, Y. Harel and S. Efrima, Langmuir, 2001, 17, 3884.
88 J. K. Sakata, A. D. Dwoskin, J. L. Vigorita and E. M. Spain, J.
Phys. Chem. B, 2005, 109, 138.
89 H. Duan, D. Wang, D. G. Kurth and H. Möhwald, Angew.
Chem., Int. Ed., 2004, 43, 5639.
90 H. Gu, Z. Yang, J. Gao, C. K. Chang and B. Xu, J. Am. Chem.
Soc., 2005, 127, 34.
This journal is
c
the Owner Societies 2006
91 M. Brust, M. Walker, D. J. Schiffrin and R. Whyman, J. Chem.
Soc., Chem. Commun., 1994, 801.
92 M. Guainazzi, G. Silvestri and G. Serravalle, J. Chem. Soc.,
Chem. Commun., 1975, 200.
93 C. Johans, K. Kontturi and D. J. Schiffrin, J. Electroanal. Chem.,
2002, 526, 29.
94 D. Rautaray, R. Kavathekar and M. Sastry, Faraday Discuss.,
2005, 129, 205.
95 V. J. Cunnane and U. Evans, Chem. Commun., 1998, 2163.
96 M. Matsuahita, M. Sano, Y. Hayakawa, H. Hono and Y.
Sawada, Phys. Rev. Lett., 1984, 53, 286.
97 L. Zeiri, O. Younes, S. Efrima and M. Deutsch, J. Phys. Chem. B,
1997, 101, 9299.
98 T. Mitsui and H. Kaneko, J. Appl. Electrochem., 1997, 27, 1100.
99 C. Shi and F. C. Anson, Anal. Chem., 1998, 70, 3114.
100 V. Mirčeski and R. Gulaboski, J. Phys. Chem. B, 2006, 110, 2812.
101 R. M. Lahtinen, D. J. Fermı́n, H. Jensen, K. Kontturi and H. H.
Girault, Electrochem. Commun., 2000, 2, 230.
102 J. D. Guo, T. Tokimoto, R. Othman and P. R. Unwin, Electrochem. Commun., 2003, 5, 1005.
103 G. Hills, A. K. Pour and B. Scharifker, Electrochim. Acta, 1983,
28, 891.
104 M. Platt, R. A. W. Dryfe and E. P. L. Roberts, Chem. Commun.,
2002, 2324.
105 M. Platt, E. P. L. Roberts and R. A. W. Dryfe, Electrochim. Acta,
2004, 49, 3937.
106 M. Platt and R. A. W. Dryfe, Phys. Chem. Chem. Phys., 2005, 7,
1807.
107 D. Al-Mawlawi, C. Z. Liu and M. Moskovits, J. Mater. Res.,
1994, 9, 1014.
108 L. Sun, P. C. Searson and C. L. Chien, Appl. Phys. Lett., 1999, 74,
2803.
109 S. R. Nicewarner-Peña, R. G. Freeman, B. D. Reiss, L. He, D. J.
Peña, I. D. Walton, R. Cromer, C. D. Keating and M. J Natan,
Science, 2001, 294, 137.
This journal is
c
the Owner Societies 2006
110 C. A Foss, M. J. Tierney and C. R. Martin, J. Phys. Chem., 1992,
96, 9001.
111 V. P. Menon and C. R. Martin, Anal. Chem., 1995, 67, 1920.
112 B. Kralj and R. A. W. Dryfe, J. Electroanal. Chem., 2003, 560,
127.
113 E. Budevski, G. Staikov and W. J. Lorenz, Electrochemical Phase
Formation and Growth, VCH, Weinheim, 1996.
114 R. M. Penner, J. Phys. Chem. B, 2002, 106, 3339.
115 M. P. Zach, K. Ng and R. M. Penner, Science, 2000, 290, 2120.
116 F. Favier, E. C. Walter, M. P. Zach, T. Benter and R. M. Penner,
Science, 2001, 293, 2227.
117 J. L. Fransaer and R. M. Penner, J. Phys. Chem. B, 1999, 103,
7643.
118 B. P. Binks, Curr. Opin. Coll. Int. Sci., 2002, 7, 21.
119 R. Aveyard and J. H. Clint, J. Chem. Soc., Faraday Trans., 1996,
92, 85.
120 H. H. Girault, Electrochimie Physique et Analytique, Presses
Polytechniques et Universitaires Romandes, Lausanne, 2001,
Ch. 5.
121 C. Johans, P. Liljeroth and K. Lontturi, Phys. Chem. Chem.
Phys., 2002, 4, 1067.
122 Y. Gimeno, A. Hernández-Creus, P. Carro, S. González, R. C.
Salvarezza and A. J. Arvia, J. Phys. Chem. B, 2002, 106,
4232.
123 T. S. Horozov and B. P. Binks, Colloids Surf., A, 2005, 267,
64.
124 R. Aveyard, B. P. Binks, J. H. Clint, P. D. I. Fletcher, T. S.
Horozov, B. Neumann, V. N. Paunov, J. Annesley, S. W. Botchway, D. Nees, A. W. Parker, A. D. Ward and A. N. Burgess,
Phys. Rev. Lett., 2002, 88, 246102.
125 M. Megans and J. Aizenberg, Nature, 2003, 424, 1014.
126 K. D. Danov, P. A. Kralchevsky and M. P. Boneva, Langmuir,
2004, 20, 6139.
127 Y. Lin, A. Böker, H. Skaff, D. Cookson, A. D. Dinsmore, T.
Emrick and T. P. Russell, Langmuir, 2005, 21, 191.
Phys. Chem. Chem. Phys., 2006, 8, 1869–1883 | 1883