Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Journal of Non-Crystalline Solids 353 (2007) 3891–3905

www.elsevier.com/locate/jnoncrysol

Liquid, glass, gel: The phases of colloidal Laponite


Herman Z. Cummins *

Department of Physics, The City College of CUNY, New York, NY 10031, United States

Available online 30 August 2007

Abstract

Laponite is a synthetic disc-shaped crystalline colloid that is widely used to modify rheological properties of liquids in applications
such as cosmetics, paints, and inks so that understanding its flow properties and aging behavior is of considerable practical as well as
fundamental importance. However, some recent studies of the liquid–glass and sol–gel transitions in aqueous Laponite suspensions have
produced results that do not fully agree with each other. Because Laponite is sensitive to sample preparation procedures, it is not
straightforward to compare results reported by different groups. We have begun a study of the dynamics of Laponite suspensions during
aging using photon correlation spectroscopy to explore the consequences of specific sample preparation procedures which may underlie
these differences, including: (1) filtration of the sample through filters with different pore sizes before beginning the experiments, (2)
adjusting and monitoring the pH of the solution, (3) varying the Laponite concentration, (4) carrying out the sample preparation in
either ambient air or dry nitrogen atmospheres, (5) baking the ‘dry’ powder to remove adsorbed water, and (6) modifying the ion con-
centration by the addition of salts. We will compare the effects of different methods of preparation on the intermediate scattering function
F(q, t) and its time evolution. In this report we will describe experiments that explore (1)–(3). The other three will be discussed in a future
publication.
Ó 2007 Elsevier B.V. All rights reserved.

PACS: 83.80.Hj; 78.35.+c; 67.40.Fd; 82.70.Gg

Keywords: Rayleigh scattering; Transport properties gel; Transport properties – Liquids; Colloids; Nano-clusters

1. Introduction mode coupling theory (MCT), the crossover with decreas-


ing temperature from cage-effect dominated dynamics to
Most recent experimental studies of the liquid–glass hopping dynamics exhibited by molecular glass-formers
transition and comparisons of the results with various the- does not occur, greatly simplifying the analysis. Also, the
ories have concentrated on fragile molecular glass-forming phase diagrams of colloidal suspension often exhibit rich
materials. However, several groups have explored the structure since, if attractive interactions are present, new
liquid–glass transition in colloidal suspensions and found phases may occur that are absent in molecular glass-
that the data obtained are well suited to testing theories. formers.
There are two particular advantages to these systems. First, Colloidal particles in dilute suspensions initially undergo
their relaxation dynamics occur on a considerably longer independent diffusional dynamics. With increasing particle
time scale than for molecular liquids, and can be followed concentration or with aging, particle interactions can lead
completely with the single experimental light scattering to more complex dynamical behavior and to transforma-
technique of photon correlation spectroscopy (PCS). Sec- tions to various new phases including fractal or compact
ond, in carrying out comparisons with predictions of the clusters, cluster gels, repulsive or attractive glasses, and
liquid-crystal phases. The widespread use of colloidal sus-
pensions and gels in foods, pharmaceuticals, cosmetics,
*
Tel.: +1 212 650 6921; fax: +1 212 650 6923. paints and inks, etc. gives these transformations practical
E-mail address: Cummins@sci.ccny.cuny.edu as well as fundamental interest and has led to an extensive

0022-3093/$ - see front matter Ó 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.jnoncrysol.2007.02.066
3892 H.Z. Cummins / Journal of Non-Crystalline Solids 353 (2007) 3891–3905

literature of studies using various theoretical and experi- by a Yukawa potential rather than the square-well poten-
mental techniques (cf., [1–4]). tial considered previously, then the liquid–glass II transi-
The simplest colloidal system, hard-spheres that interact tion line would extend to very low volume fractions,
only at contact (HSS) suspended in a neutral solvent, suggesting that the sol–gel transition is a low-/ continua-
undergoes crystallization to a FCC close-packed structure tion of the glass II transition.
when the particle concentration (volume fraction /) reaches A modification of standard MCT was proposed by Kroy
0.50. However, if there is sufficient polydispersity, crystal- et al. [30] in which two MCT ergodicity-breaking transitions
lization may be avoided and, at /  0.58, a liquid–glass occur: a first short length-scale transition involving the
transition occurs as each colloidal particle becomes trapped formation of clusters, and a second larger length-scale tran-
in the cage formed by its neighbors. This HSS liquid–glass sition in which the clusters aggregate to form a gel (CMCT).
transition has been studied extensively by the groups of van A related scenario was identified in simulation studies by
Megen and Pusey [5–11] and Bartsch [12] and has provided Sciortino et al. [31,32]. These analyses suggest that the same
critical tests of theories of the liquid–glass transition. mechanism underlying the liquid–glass transition also
These studies showed that with increasing volume frac- underlies the sol–gel transition, so that the characteristic
tion / the relaxation dynamics slows dramatically, and dynamical signatures of the liquid–glass transition should
the quantitative structure of the intermediate scattering also appear at the sol–gel transition. The quantitative
function F(q, t) and its evolution with concentration, as aspects of these theoretical predictions largely remain to
determined by dynamic light scattering (photon correlation be explored experimentally, especially those regarding their
spectroscopy), are closely described by quantitative predic- dynamics.
tions of the mode-coupling theory (MCT) for the hard- If the colloidal particles are electrically charged, addi-
sphere system [13–15]. In the experiments, kinetic arrest tional phases can occur. Kumar and Wu [33] reported
was found to occur at a volume fraction of /  0.58, some- molecular dynamics simulations of colloids interacting
what higher than the MCT ideal glass transition prediction through a short-ranged van der Waals attraction and a
/C = 0.516 for the hard-sphere system (HSS), although longer-ranged electrostatic repulsion. They observed a vari-
recent extensions of MCT to include higher-order terms ety of ‘jammed states’ at volume fractions between / = 0.4
reportedly lead to an increase in this MCT value [16]. and / = 0.1, ranging from nearly uniform glass-like struc-
If, in addition to the hard-sphere repulsive potential tures to network-like gel structures. The relation between
there is a short-range attractive potential, an additional cluster formation and combined short-range attraction
soft-solid phase can occur. In 1999, Fabbian et al. carried and long-range repulsion has been studied by Sciortino
out MCT calculations for a system of colloidal spheres et al. [32]. Lu et al. have reported that suspensions of col-
characterized by a hard-sphere potential supplemented by loids with attractive interactions induced by polymers exhi-
a short-range attractive square-well potential (‘sticky bit a stable phase of clusters even in the absence of long-
hard-spheres’) [17–20]. They found that this system exhibits range repulsion, and that clusters can percolate across the
two glass transitions; first, with increasing strength of sample to form a gel [34]. Also, if the colloidal particles
attraction, the volume fraction /C for the usual cage-effect are electrically charged, a third glass phase can occur, sta-
mediated glass transition increases (glass I). Second, within bilized by electrostatic repulsion. This phase is sometimes
the glass I phase, further increase of the attraction causes called the ‘Wigner glass’.
pairs of particles to move together, opening holes in the In the colloidal systems described so far, the individual
cages, and causing the glass to melt. Finally, as the attrac- particles are assumed to be spherical. In the case of asym-
tion increases further, a second transition dominated by metric particles (e.g. round discs as in Laponite), there is
attractive forces occurs (glass II). These predictions were also the possibility of orientational order and liquid-crystal
verified in experiments in which the attractive interaction phases. Also, asymmetry of the charge distribution can
was produced by adding small polymers to the colloidal produce dense soft-solid phases stabilized by electrostatic
suspension, which causes a short-range depletion attraction interactions.
[21–23].
Soft-solid phases of colloids held together by attractive 1.1. Laponite
forces are usually considered as gels, but the distinction
between gels and attractive glasses is not clear. Analogies Many recent studies of the liquid–glass and liquid–gel
between the two have been studied by several groups, e.g. transitions have employed the synthetic colloid Laponite
[24,25]. Segre et al. have shown that relaxation dynamics which has all the characteristics discussed so far: both
near gelation and near the liquid–glass transition are attractive and repulsive interactions, anisotropy and net
remarkably similar [26]. Bergenholtz and Fuchs [27–29] charge, as well as an anisotropic charge distribution. It
examined the mode-coupling theory predictions for the exhibits an array of different phases and behaviors and
behavior of colloidal suspensions with attractive interac- has become a widely used model system for testing theories
tions at low volume fractions and concluded that the sol– of liquid–glass and liquid–gel transitions as well as various
gel transition could also be described by MCT. They noted aspects of aging phenomena. However, Laponite is not a
that if the short-range attractive interaction is represented simple material to handle, since it is sensitive to the meth-
H.Z. Cummins / Journal of Non-Crystalline Solids 353 (2007) 3891–3905 3893

ods of sample preparation. Comparing the different When dispersed in water, Laponite hydrates and swells to
reported studies therefore requires evaluating the impor- form a clear colloidal dispersion with the Na+ ions forming
tance of the different methods of preparation employed double layers on the faces. The pH for a 2% Laponite sus-
by different authors. There are potentially many different pension in pure water is 9.8.
Laponite phases possible, and a major challenge is to sort At low ionic strength, electrostatic repulsion keeps the
out which of these phases are observed under particular particles apart. Laponite is decomposed by acids, leading
experimental conditions, and how they are influenced by to an increase in ion concentration with time at low pH.
the sample preparation method followed. At concentrations of 2% or greater in water a gel will form
Laponite (hydrous sodium lithium magnesium silicate) rapidly. However, gel formation has been observed at con-
is a synthetic crystalline layered silicate colloid with crystal centration well below 2% in several studies including the
structure and composition closely resembling the natural present one.
smectite clay hectorite. It is manufactured by Rockwood Laponite gel is strongly thixotropic, i.e. its viscosity
Additives Ltd (formerly Laporte Ind. Ltd.), Cheshire decreases rapidly under shear. After the shear stress is
UK, and Southern Clay Products, Inc., Gonzales, Texas. removed, the gel reforms; the rate of restructuring depends
Chemical analysis of Laponite RD by Levitz et al. [35] gave on composition, electrolyte level, age of the dispersion, and
mean chemical composition: SiO2, 65.82%; MgO, 30.15%; temperature. The addition of salts reduces the thickness of
Na2O, 3.20%; LiO2, 0.83%. The melting point is 900 °C. the electrical double layer, promoting gel formation.
Extensive information on the structure and applications Laponite contains approximately 8 wt% water which is
of Laponite can be found on the manufacturer’s websites chemically absorbed into the crystal structure and can only
http://www.laponite.com and http://www.scprod.com. be removed by baking at temperatures above 150 °C. In
The density of Laponite is 2.53 g m/cm3. Single Lapo- addition, Laponite is hygroscopic and will adsorb addi-
nite crystals are disc shaped and nearly uniform, typically tional water from the atmosphere, typically up to 15% at
25 nm in diameter by 0.92 nm thick, much smaller than 50% relative humidity. The structure of individual Lapo-
natural clays. Within a single crystal, each sheet of octahe- nite particles and a schematic drawing of the proposed
drally coordinated aluminum or magnesium oxide is sand- ‘house of cards’ soft-solid phase are illustrated in Fig. 1.
wiched between two layers of tetrahedrally coordinated There are several different grades of Laponite available
silica.The crystal faces have negative charge; the edges have for different commercial applications. Laponite RD, the
small pH-dependent positive charge, typically 10% of the most frequently studied grade, is used in many household
negative charge. The overall net negative charge of a single and industrial products including cleansers, surface coat-
Laponite disc is approximately 700 electron charges. The ings, and ceramic glazes. Laponite XLG is a high-purity
charge is balanced by interlayer cations which are predom- grade of Laponite RD, processed to remove impurities
inantly Na+. In the dry powder, the Laponite crystals form such as heavy metals e.g. lead and arsenic. This grade is
into stacks with the crystals sharing interlayer Na+ ions. used in personal care and cosmetic products including

Fig. 1. Structure of individual Laponite particles and schematic house of cards structure of Laponite gel stabilized by electrostatic interaction between the
negatively charged faces and positively charged edges of the disc-shaped colloidal particles (from southern clay products product information website).
3894 H.Z. Cummins / Journal of Non-Crystalline Solids 353 (2007) 3891–3905

shampoos and sunscreens. Laponite XLG was used in the indicative of nematic orientational order, for Laponite con-
studies of Thompson and Butterworth [36] and in the work centrations above 2 wt%. Gabriel et al. [43] observed sus-
discussed in this report. pensions of Laponite (Laponite B) between crossed
polarizers and found optical birefringence for concentra-
1.2. Laponite phases discussed in the literature include the tions above 2.4 wt%, again indicative of nematic order.
following Agra et al. [44] have shown theoretically how a rich variety
of orientational ordered phases in colloidal crystals can be
1.2.1. Liquid with clusters understood.
Some studies of Laponite suspensions at concentrations Previous light scattering studies of Laponite have been
of 0.9–5 wt% using angle resolved static light, neutron, and reported in numerous references including [37–41,45–57].
X-ray scattering suggested that they contain clusters that Sample preparation methods differ widely among the
have fractal structure [37,38]. Bonn et al. [39] compared published studies. Some of the specific aspects of the prep-
light scattering from two samples prepared with 3.5 wt% aration procedures whose importance we are investigating,
Laponite and found that the angle-dependent intensity are:
characteristic of fractal structures was present in the freshly
prepared solutions, but that filtration through a 0.8 lm 1. Sample filtration: What type and pore size filter was
millipore filter resulted in the complete absence of the angle used? Was there a delay between mixing and filtration?
dependence. Clusters should be more likely to form in sus- 2. Is the water pH adjusted before/after adding the Lapo-
pensions having higher ionic concentration since at low ion nite? Is it monitored later?
concentrations repulsive electrostatic interactions will keep 3. What is the Laponite concentration?
the particles apart. 4. Is the sample prepared under nitrogen or in a normal
ambient atmosphere?
1.2.2. Wigner glass 5. Is the sample dried to remove moisture?
At low ionic strength, electrostatic repulsion keeps the 6. What is the ion concentration (possible modification by
colloidal particles apart and can produce a transition to addition of salt)?
an arrested state stabilized by long-range electrostatic
repulsion [40]. In this report we will concentrate on points 1–3. The
others are currently under study and will be discussed in
1.2.3. High-density gel a future publication.
At higher ionic strength, as the screening length
decreases, the positive double layers at the edges of plate- 2. Experimental
lets can approach the negatively charged double layers on
the faces. The high-density gel state of Laponite, called 2.1. Sample preparation
the ‘house of cards’ structure, occurs when the screening
length is sufficiently short so that this attractive interaction The Laponite XLG used in the experiments described in
dominates. This structure is readily observed if dry Lapo- this report was lot 04-239, purchased from Southern Clay
nite powder is mixed with tap water which typically has a Products in Feb 2005. The certificate of analysis indicates
high ion concentration. 6.8% moisture content, although this should be expected
to increase during handling and transfer to storage jars.
1.2.4. Low-density gel The moisture content was measured during preparation
Ruzicka et al. [41] studied Laponite suspensions with of the samples with a Sartorius MA100C moisture analyzer
concentrations between 0.3 and 3.1 wt%. They found that and was found to be 9.8%.
even for the lowest concentrations a transition to an Samples for the PCS experiments were prepared with the
arrested phase occurs after a sufficiently long time Laponite as provided without further drying. Samples were
(6 months for 0.3 wt%). They also found that the time loaded in screw-top cylindrical glass vials with outside
evolution of the dynamics differed for concentrations diameters of either 20 or 28 mm. Three different series of
above and below 0.17 wt% suggesting that there are Laponite samples were prepared. Each series included sev-
two different gel structures for this material. One possibility eral stock solutions with different concentrations prepared
is that the high-density gel is the ‘House of Cards’ structure following the same procedure. From each stock solution,
while the low-density gel consists of a network of chains as three (or more) samples were prepared by extracting some
one finds in polymer gels, which can form gels at very low of the stock solution with a syringe and forcing it through
concentrations. Alternatively, the low-density gel may con- various Millipore millex sealed syringe filters with 33 mm
sist of a network of clusters as discussed by Lu et al. [34]. mixed cellulose ester membranes. For each such prepara-
tion, one sample was prepared without a filter. The samples
1.2.5. Nematic phases are listed in Table 1. Concentrations are given in weight
Lemaire et al. [42] studied Laponite gels with SAXS and percent of Laponite, uncorrected for water content of the
found evidence of anisotropy in the scattering patterns, powder. Using the Laponite density of 2.53 g m/cm3 and
H.Z. Cummins / Journal of Non-Crystalline Solids 353 (2007) 3891–3905 3895

Table 1
Laponite samples and the results of PCS measurements as discussed in the text
Series Sample Filter Loaded Gelled Bad KWW Last PCS 7/18/06 since load (days)
A (measure pH later) pH = 9.53
AA (0.89%) Mix: 6/30/05 AA2 None 7/6/05 1/4/06 9/14/05 GEL (G)182
AA3 0.1 7/6/05 1/4/06 9/14/05 GEL (G)182
AA4 0.8 7/14/05 7/10/06 liq(7/17) 368a
AA5 0.45 7/14/05 9/12/05 9/10/05 GEL (G)60
pH = 9.28
AB (0.06%) Mix: 7/18/05 AB1 None 7/18/05 2/16/06 9/22/05 liq(7/17) 364
AB2 0.45 7/19/05 2/9/06 2/9/06 liq(7/17) 363
AB3 0.8 7/19/05 1/5/06 9/14/05 liq(7/17) 363
pH = 9.90
AC (1.50%) Mix: 8/8/05 AC1 None 8/9/05 9/15/05 9/15/05 GEL (G)37
AC2 0.8 8/9/05 6/13/06 3/2/06 3/6/06 GEL (G)308
AC3 0.45 8/9/05 6/13/06 3/2/06 GEL (G)308

B (adjust pH after mixing if pH < 10.0) pH = 10.12


BA (1.00%) Mix 1/2/06 BA1 None 1/2/06 7/10/06 liq(7/17) 196
BA2 0.8 1/2/06 7/10/06 liq(7/17) 196a
BA3 0.45 1/2/06 7/10/06 liq(7/17) 196
pH = 10.37
BB (0.18%) Mix: 2/6/06 BB1 None 2/6/06 7/10/06 liq(7/17) 161
BB2 0.8 2/6/06 7/10/06 liq(7/17) 161
BB3 0.45 2/6/06 7/10/06 liq(7/17) 161

C (use water with pH = 10) pH = 10.27


CA (0.98%) Mix: 1/3/06 CA1 None 1/3/06 7/10/06 liq(7/17) 195
CA2 0.8 1/3/06 5/3/06 4/11/06 4/11/06 GEL (G)123
CA3 0.45 1/3/06 7/10/06 6/5/06 6/26/06 Soft GEL (G)194
pH = 10.29
CB (0.04%) Mix: 1/9/06 CB1 None 1/10/06 7/10/06 liq(7/17) 188
CB2 0.8 1/10/06 7/10/06 liq(7/17) 188
CB3 0.45 1/10/06 7/10/06 liq(7/17) 188
pH = 10.42
CC (0.18%) Mix 1/12/06 CC1 None 1/13/06 7/10/06 liq(7/17) 185
CC2 0.8 1/13/06 7/10/06 liq(7/17) 185
CC3 0.45 1/13/06 7/10/06 liq(7/17) 185
The final column shows the elapsed time (in days) before sample gelation (G) or, if gelation was not observed by 7/18/06, the elapsed time since it was
prepared.
a
Note: by 10/25/06 samples AA4 and BA2 had also gelled.

water content of 9.8%, the relation between volume frac- B series Laponite powder was added slowly while stir-
tion / and concentration C is / = 0.9C/(2.5  1.35C) (with ring; after mixing was complete the pH was adjusted
C = 0.01*C (wt%)). For the samples studied / ranges from to pH > 10 by addition of 1% NaOH solution, if
a maximum of 5.4E3 for the 1.5 wt% samples to 1.4E4 required. Because the DIUF water pH is 4, some acid
for the 0.04 wt% samples. The samples were all prepared dissociation of these B series samples may have occurred
under normal ambient atmosphere. Preparation of samples before the pH was adjusted. Therefore, for the C series,
in a glove box under dry nitrogen atmosphere is currently the water pH was adjusted before adding the Laponite.
in progress and will be discussed in a future publication. C series Laponite powder was added slowly to DIUF
Three series of samples (A, B, C) were prepared, each water with pH adjusted to >10 by addition of 1% NaOH
following a different pH adjustment protocol. The pH val- solution before mixing.
ues measured after completion of the mixing procedures
are shown in the second column of Table 1. Periodically all samples were removed from the storage
rack and tilted slightly to see if gelation had occurred. This
A series Laponite powder was added slowly to distilled tilting may have caused some slight mixing in those sam-
or DIUF water while stirring. There was no measure- ples that had not gelled. In the right-hand column of Table
ment or control of pH during preparation. The pH of 1 we show the elapsed time (in days) from preparation until
each stock solution listed in the table was measured a gel was observed. For samples that had not gelled, we
later. give the elapsed time from sample preparation until the last
3896 H.Z. Cummins / Journal of Non-Crystalline Solids 353 (2007) 3891–3905

2 2
observation of the liquid. Note that for samples with g2 ðtÞ ¼ 1 þ ajg1 ðtÞj ¼ 1 þ a½F ðq; tÞ=F ðq; 0Þ ð1Þ
C < 0.2 wt% no gelation was observed.
The elapsed time (in days) from sample preparation In the simplest case of uncorrelated spherical particles of
until a gel was observed is shown for all samples by the radius r undergoing independent translational diffusion,
solid symbols in Fig. 2. For the samples where gelation g2 ðtÞ ¼ 1 þ a expð2t=sÞ ¼ 1 þ a expð2Dq2 tÞ ð2Þ
was not seen, the sample is represented by an open symbol.
where the translational diffusion constant D = kT/6pgr.
We note that the filters were used as provided by the
For a distribution of particle sizes, a simple generalization
manufacturer. Some surprising inconsistencies that we
(which we will use here) is to replace the exponential in Eq.
observed, especially with the 0.8 lm filters, may be related
(2) with a KWW stretched exponential function and to use
to residual traces of detergent or solvent in the filters. This
a free baseline b  1:
could be checked by rinsing the filters with pure water h i
before passing the Laponite solution through them, but g2 ðtÞ ¼ b þ a exp 2ðt=sÞ
b
ð3Þ
this has not yet been done.
For 4880 Å light and 90° scattering, the mean hydrody-
namic radius rh is approximately related to the measured
2.2. PCS measurements correlation time s by
PCS measurements were carried out with a Brookhaven rh ðnmÞ ¼ sðlsÞ=7:76 ð4Þ
Instruments BI-9000AT digital correlator. Excitation was We used Eqs. (3) and (4) to find approximate scatterer sizes
provided by a Coherent Innova I306C Argon laser operat- from the PCS data. For independent single Laponite parti-
ing in single-mode at 488 nm with typical output power of cles we expect rh  13 nm. We emphasize that this fitting
150 mW. Power at the sample was approximately 50 mW. procedure was a simple approximation used to provide a
All experiments were performed at a 90° scattering angle rough estimate of the time evolution of cluster sizes and
with data collection time of 10 min. polydispersity under different preparation procedures.
For ergodic samples the normalized intensity correlation Since the experiments were performed at fixed q, the q2
function g2(t) = C(t)/B (where B is the background) is dependence of Eq. (2) was not tested. Furthermore, the
related to the intermediate structure factor F(q, t) by cluster size was estimated from Eq. (4) using the value of
s from the fits to Eq. (3). The mean value of s would be in-
creased for b < 1, reaching hsi = 2s for b = 0.5.
If the colloidal sample is a gel, then extracting dynamical
information from PCS data is much more difficult as dis-
cussed in detail by Pusey and van Megen in 1989 [58].
We will discuss the PCS data analysis problem for gels
briefly in Section 3.4 below.
If the correlation function of a monodisperse solution is
fit to Eq. (3), the KWW stretching coefficient should be
b = 1. If the sample is polydisperse then b < 1. To explore
the dependence of b on polydispersity, we constructed syn-
thetic g2(t) data and performed KWW fits for theoretical
polydisperse solutions with radii ranging from 13 nm to a
maximum rmax between 13.1 nm and 300 nm, assuming
that the product of particle concentration and particle scat-
tering strength was constant across the range of sizes
included. For a size distribution whose width is 0.4 times
the mean size, b is 0.99, still very close to 1. For the most
polydisperse C(t) considered, with width 1.8 times the
mean size, b decreased to 0.79. Also, for that fit, there is
a small systematic error as C(t) begins to decay from the
initial plateau; the error is very similar to fitting errors seen
in the PCS experiments as discussed below.

3. Results
Fig. 2. Elapsed time in days from sample preparation until first
PCS experiments on the Laponite samples listed in
observation of a gel (solid symbols) vs concentration in wt%. Open
symbols indicate samples that were still liquid at the latest observation. Table 1 were performed frequently, beginning soon after
Series A-circles, series B-Squares, series C-triangles. No filter: large each sample was prepared. The PCS data was analyzed
symbols; 0.8 lm filter: medium symbols; 0.45 lm filter: small symbols. with the four-parameter KWW function (Eq. (3)) from
H.Z. Cummins / Journal of Non-Crystalline Solids 353 (2007) 3891–3905 3897

which an estimate of the average size rh of the scatterers included for the 137-day data. The relaxation slows with
was obtained with Eq. (4). The relaxation time s (and increasing time for all three samples as expected due to
estimated radius rh) increased with time for all samples growth of clusters. From the KWW fits with Eq. (3), we
studied, although in some cases a small decrease was obtained average sizes rh at 6 days and 137 days for the
observed during the first few days after sample preparation. three samples: (CC1) 14.8 nm, 360.4 nm; (CC2) 14.6 nm,
In Fig. 3 we show PCS data for samples CC1, CC2, and 154.0 nm; (CC3) 14.3 nm, 53.9 nm. At 6 days, all three
CC3 (0.18 wt%), 6 days after preparation (squares) and 137 samples had correlation times of about 100 ls and corre-
days after preparation (circles). The KWW fits are also sponding estimated radii of rh  14 nm, indicating that

Fig. 3. PCS data for 0.18 wt% Laponite samples CC1, CC2, and CC3 six days (squares) and 137 days (circles) after preparation. The solid lines are KWW
fits used to extract estimates of the average radius of the scatterers as described in the text.
3898 H.Z. Cummins / Journal of Non-Crystalline Solids 353 (2007) 3891–3905

the scatterers were individual Laponite particles or very The systematic departures from the KWW fit in Fig. 3
small clusters. (Rosta and von Gunten concluded that their for sample CC1 closely resemble those seen in our fits to
Laponite suspensions contained very small clusters of synthetic data for the most polydisperse case, indicating the
between two and four platelets [57]). Note that at 6 days presence of considerable polydispersity in this sample.
rh does not depend on the filter used, but after 137 days At longer times, the correlation functions of samples
rh of the unfiltered sample has increased the most, while that remain liquid often evolve into shapes with long tails,
the 0.45 lm filtered sample has increased the least. How- signaling the existence of large slow-moving clusters with
ever, as we shall see, the correlation between filter pore size large polydispersity and limiting the utility of KWW fits.
and cluster size is not generally consistent. Fig. 4 shows PCS data for samples CC1, CC2, and CC3

Fig. 4. PCS data for 0.18 wt% Laponite samples CC1, CC2, and CC3 166 days after preparation showing the ‘tails’ on C(t) for CC1 and CC2, but not for
CC3. The insets for CC1 and CC3 show the counts accumulated during each second of the 10-min runs.
H.Z. Cummins / Journal of Non-Crystalline Solids 353 (2007) 3891–3905 3899

recorded 166 days after preparation. For CC1 and CC3 we increasing polydispersity, that precedes gelation (see, e.g.,
also include the count rate histories as insets which show BA2, BB2, and CA2).
the number of photocounts collected each second during
the 10-min runs. Sample CC1 (unfiltered) has a prominent A series: Of the samples with 0.89 wt% concentration, all
PCS tail, and the count rate exhibits large fluctuations on a but one (AA4) had gelled within 75 days of prep-
time scale of 1 min, indicating (via Eq. (4)) the presence aration while the third sample (AA4) was still
of large clusters with rh  10 lm. Sample CC2 also has a liquid after 250 days. The 1.50 wt% AC samples
prominent tail, but CC3, which was passed through the all gelled, with the unfiltered sample AC1 after
0.45 lm filter, has no tail and the count rate history shows 60 days and the other two after 340 days.
no slow fluctuations. The AB 0.06 wt% samples showed constantly
The evolution of rh and b obtained from the KWW fits increasing cluster sizes but did not gel within
for all samples studied is shown in Figs. 5–7. From these fig- one year.
ures, and from Fig. 2, some general observations can be B series: Samples BB1, BB2, and BB3 (0.18 wt%) PCS
made. First, for the lowest concentration samples data obtained for up to 154 days after prepara-
(C < 0.2 wt%), no gelation was observed within the obser- tion. Note that the unfiltered sample (BB1) and
vation time of 1 year. Second, for the highest concentra- the 0.45 lm filtered sample (BB3) have
tion samples AC, the unfiltered sample gelled first (37 rh  12.3 nm indicating that no significant aggre-
days) while the two filtered samples did not gel until 308 gation has occurred while sample BB2, filtered
days. But for the AA 0.89% samples, the unfiltered and with a 0.8 lm filter, has rh  56 nm indicating
0.45 lm filtered samples gelled at 182 days and 60 days, considerable aggregation.
respectively, while the 0.8 lm filtered sample was still liquid C series: Aggregation of the C samples proceeded as least
after a full year. as fast as the B samples. This indicates that there
Third, for the samples that gelled, there was a rapid is no advantage to adjusting the water pH before
increase in size and corresponding decrease in b, indicating addition of the Laponite.

Hydrodynamic radius vs elapsed time - samples AA,AB,AC (12 JULY06) A_ rh-v s-et .pxp

70 KWW parameter β vs elapsed time - samp les AA, AB, AC ( 12 July 06) A_ bet a-v s-et .pxp
AA : 0.89 wt% - mixed inpure
0.89 wt%
radius (nm)= tau (microsec)/7.76

AA5 gelled
DIUF water - no pH control 1. 0
12 Sept red: AA2 (no filter)
60 red AA2 (no filter)
KWW stretching parameter β

green: AA4 (0.8micron filter)


green AA4 (0.8micron filter)
blue: AA5 (0.45 micron filter)
blue AA5 (0.45 micron filter) 0. 9 black: AA3 (0.1micron filter)
50 black AA3 (0.1micron filter)

AA3 gelled 0. 8
40 10July06
after 14 Sept
( 361 days)
30 0. 7
AA2 gelled AA4
after 14 Sept
20 0. 6 10July06
t=361 days

10 0. 5

0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350
elapsed time since loading (days) elapsed time since sample loading (days)
4 AB: 0.06 wt% - mixed in DIUF
water (no pH adjustment)
0. 06 wt % red: AB1 (no filter)
2 green: AB3 (0.8 micron filter)
5 Jan 0.8
blue: AB2 (0.45 micron filter)
1000
6 9 Feb
0.6
4

2
red AB1 (no filter)
0.4
100 green AB3 (0.8micron gilter) 9 Feb06
6 blue AB2 (0.45 micron filter)
4 NOTE: Beyond ~ 60 days, PCS datanot 0.2
described by KWW - have big tails
2
BUT AB SAMPLES DONOT GEL
0.0
50 100 150 200 20 40 60 80 100 120 140 160

0.7 red: AC1 (no filter)


green: AC2 (0.8 micron filter)
blue: AC3 (0.45 micron filter)
1000 AC 1.50 wt % - mixed in oure DIUF water - no pH adjustment 0.6
8
red AC1 ( no filter)
1. 50 wt %
6
green AC2 (0.8micron filter)
4
blue AC3 (0.45 micron filter) 6 March 0.5
2

100 gel ( 60 days) 0.4


8
6
4
AC2 & AC 3
Poor KWW
0.3
2
fi ts ; gelled after 6 Marc h06
~340 days
10 0.2
0 50 100 150 200 250 0 50 100 150 200 250 300

Fig. 5. Approximate hydrodynamics radius (left) and KWW stretching parameter b (right) vs elapsed time since sample preparation in days from KWW
fits for all samples in series A.
3900 H.Z. Cummins / Journal of Non-Crystalline Solids 353 (2007) 3891–3905

Stretching coefficient beta vs elapsed time - samples BA, BB [10 July 06] B_beta -v s-et .pxp
Hydrodynamic radius vs elapsed time - samples BA, BB (10July06) B_rh-v s-et .pxp
0.8
8
7
6 BA: 1.0 wt% mixed in pure DIUF water
5
BA:1.0 wt % adjust pH = 10 afterwards
4 if needed with 1% NaOH
red: BA1 (no filter) 0.7
3 green: BA2 (0.8 micron filter) BA: beta vs elapsed time
radius(nm) = tau(microsec)/7.76

blue: BA3 (0.45 micron filter) red: BA1 (1.0 wt%)


2 green: BA2 10 July06
blue: BA3
0.6

100
8
7
6 10 July 06 0.5
5 189 days
4

3
0.4
2

10 0.3
0 50 100 150 200 0 50 100 150
elapsed time since loading (days)

BB:0.18 wt% mixed in pure DIUF water, 0.90


BB: beta vs elapsed time
100 BB: 0. 18 wt % adjust pH = 10 afterwards if needed
red: BB1 (no filter)
red: BB1 (0.18 wt%)
9 gr een: BB2
8 green: BB2 (0.8 micron filter) 0.88
blue: BB3 (0.45 micron filter) blue: BB3
7

6 0.86 10 July 06
5
0.84
4

3 0.82

0.80
2

10 July 06 0.78
154 days

0.76
10

0 20 40 60 80 100 120 140 160 0 20 40 60 80 100 120 140 160

Fig. 6. Approximate hydrodynamics radius (left) and KWW stretching parameter b (right) vs elapsed time since sample preparation in days from KWW
fits for all samples in series B.

3.1. Cluster formation vs gel formation and can be characterized as a fractal structure with fractal
dimension D in the range 1.7 < D < 2.2. When coagulation
Most samples showed increasing s (and rh) and decreas- is slow, the aggregates tend to be much more dense [2]. This
ing b with increasing time, in some cases following an ini- distinction was discussed by Lin et al. [2] for colloid aggre-
tial short-time decrease in rh, demonstrating that both gation and may underlie the two routes to gelation
mean cluster size and polydispersity generally increase as reported by Ruzicka et al. [41].
aging proceeds. For some samples, the intercept/back-
ground ratio a/b suddenly decreased from 1 to 0.5 or less, 3.2. When does aging begin?
and tipping these samples then showed that a gel had
formed. The dates and corresponding elapsed times since It has sometimes been asserted that when stock Lapo-
preparation when a gel was first observed for each sample nite solutions are passed through a filter into a sample cell,
are also shown in Table 1. For other samples, especially all clusters are broken up and the sample aging process
those prepared at low concentrations, s (and rh) continued effectively starts over, so the aging time clock is reset to
to increase while a/b remained at 1. For these samples, zero. However, we observed three effects that appear to
the correlation function C(t) usually developed a long high contradict this claim:
tail indicating that cluster size and polydispersity continue
to increase, but the samples remained liquid. Also, the (1) Stock solution AA was mixed on 6/30/05 with concen-
count rate record for these samples show very slow fluctu- tration C = 0.89 wt%. Samples AA2 through AA5
ations, indicating the presence of very large clusters. These were loaded within the next two weeks. Another sam-
two distinct patterns of time evolution of the PCS data are ple, AA6, was loaded 88 days after mixing. Samples
illustrated in Fig. 8. AA5 and AA6 were both prepared using 0.45 lm fil-
For polymer suspensions, as the particles aggregate, the ters. The first PCS run for sample AA6, carried out
form of the aggregates (or clusters) can take on different on the same day that the sample was prepared, gave
structures depending primarily on the coagulation rate. an initial s value of 1020 ls, 8 times larger than the
When coagulation is rapid, the cluster structure is open 125 ls found for sample AA5. Presumably, some clus-
H.Z. Cummins / Journal of Non-Crystalline Solids 353 (2007) 3891–3905 3901

Hydr odynamic radius vs elapsed ti me - samp les CA, CB, CC (13July06) C_rh-v s-et .pxp Dependence of KWW beta on elapsed time - Series C (13July2006) C_beta -v s-et .pxp

6 CA : 0.89 wt% mixed in pH=10 DIUF water CA3 1.0 CA: beta vs elapsed time
radius (nm) = tau (microns)/7.76

(adjusted with 1% NaOH) CA2 Gel - 13July 0.89 wt % red - CA1 (0.89%)
4
red: CA1 (no filter) Gel-3May 174 days 0.9 green - CA2 (stop at 98)
green: CA2 (0.8 micron filter) 120 days) blue - CA3
2 0.8
(98days BIG TAIL - stop KWW)
blue: CA3 (0.45 micron filter) 10July06
100 0.7
8
6 CA: 0.89 wt% 0.6
4
0.5 CA2 10July06
Gel-3May
2 0.4 (120 days)
CA3
10 0.3 GEL-13July
(188 days)
0 50 100 150 0 50 100 150
elapsed time since loading (days)

CB: 0.04 wt% mixed in pH=10 DIUF water 1.0


2
red: CB1 (no filter)
0.04 wt %
1000 green: CBA2 (0.8 micron filter) 0.9
6 blue: CB3 (0.45 micron filter)
0.8 10July06
4

2
CB: 0.04 wt% 0.7

100 0.6
6 CB : beta vs elapsed time
4 0.5 red - CB1 (0.04%)
green - CB2
2 10July06 0.4 blue - CB3
10 0.3
0 50 100 150 200 0 50 100 150

1000 1.0
8 CC: 0.18 wt% mixed in pH=10 DIUF water 0. 18 wt % CC: beta vs elapsed time
6 red: CC1 (no filter) red - CC1 (0. 18% )
green: CC2 (0.8 micron filter) 0.9 gr een - CC2
4 blue: CC3 (0.45 micron filter) blue - CC3

0.8
2 CC: 0.18 wt%
100 0.7 10July06
8 13July06
6
0.6
4

2 0.5

10 0.4
0 50 100 150 0 50 100 150

Fig. 7. Approximate hydrodynamics radius (left) and KWW stretching parameter b (right) vs elapsed time since sample preparation in days from KWW
fits for all samples in series C.

ters that had formed in the AA stock solution during increases (for PCS spectra of a standard 22 nm polystyrene
the 88 days after it was mixed were not fully broken suspension, the same procedure gives excellent KWW fits).
up by filtration in the preparation of sample AA6. To see if this effect is due to anisotropy or polydispersity,
(2) For many of the samples (e.g. BA) the value of s we carried out several runs with polarization selection,
decreased for several days after the sample was using samples contained in square optical cuvettes to avoid
loaded and then began to increase again (see polarization distortion. The experiments were performed
Fig. 6). This observation suggests that some small with the incident light polarized vertically, perpendicular
aggregates present in the dry powder survive several to the scattering plane (V) and the scattered light polariza-
hours of mixing and filtration but do dissolve slowly tion was selected as either vertical (VV), horizontal (VH) or
in the sample cells after several days. all scattered light was collected (VT).
(3) The records of radius vs elapsed time shown in Figs For a 1% Laponite dispersion (CA) the Laponite VT and
5–7 allow a comparison of results for different filter VV fits were nearly identical, giving rh = 12.6 and 12.7 nm,
sizes. From the figures, there is no clear correlation respectively. The VH spectrum was very weak, with intensity
of radius with filter size. In fact, for some samples about 3% of the VV intensity. This indicates that the anisot-
prepared with no filter (e.g. CA1 and CB1) the mean ropy of the Laponite particles is not a significant factor in the
cluster size increases less with time than the samples PCS data, and that the typical departure from the KWW fit,
prepared with 0.45 or 0.8 lm filters. The origin of this visible in the short-time behavior of the VT and VV spectra,
inconsistency is currently unknown. is due to polydispersity and not to anisotropy.

3.3. Anisotropy vs polydispersity 3.4. Gels

The fits of Laponite PCS data to Eq. (3) were primarily As the colloidal solution transforms from a sol to a gel
used to estimate rh, but some of the fits were poor, espe- there are dramatic changes in the structure and dynamics
cially in the region of the initial decay away from the pla- that continue to evolve as the sample ages. We intend to
teau. Departures becomes more visible as the mean size explore this aspect of Laponite in detail. So far, however,
3902 H.Z. Cummins / Journal of Non-Crystalline Solids 353 (2007) 3891–3905

Fig. 8. Correlation data with KWW fits for sample AA5 (top) and AB1 (bottom). AA5 was a gel by 62 days after preparation and shows a drop in its a/b
ratio. AB1 remained liquid, but developed a long tail indicating the presence of large clusters.

we have only carried out a preliminary study of one aspect In their 1989 paper, Pusey and van Megen [58] suggested
of the gel transition: how the PCS data are affected by the another way to overcome the nonergodicity problem by
onset of nonergodicity as described briefly below. looking for a place in the scattering volume where the static
When a colloidal dispersion gels, the range of motion of component of the scattering is very weak. Fig. 9 shows PCS
each particles becomes limited and the dynamics becomes spectra of Laponite sample AA5 (0.89 wt%). In the upper
nonergodic. Time averages and ensemble averages are no panel, the sample is a liquid with a/b ratio 0.95, at times
longer equivalent and Eq. (1) and (2) are then not valid. To of 0, 20, and 25 days after loading the sample. The initial
overcome the problem of nonergodicity, several methods hydrodynamic radius is 15 nm, increasing to 70 nm by
have been described. First, the sample can be slowly rotated 25 days. By 62 days after loading, the sample has gelled
[23] or translated during the PCS measurement so that many and the a/b ratio has dropped from 0.95 to 0.35. The
independent scattering volumes are sampled sequentially, PCS spectra shown in the lower panel are all at 62 days
making the time-averaged PCS data effectively an ensemble or later. The a/b ratio varies between a maximum of
average. Second, scattered light can be collected simulta- 0.8 to a minimum of 0.05, depending on location in
neously over a range of scattering vectors and the multispec- the sample. The higher a/b ratios corresponded to lower
kle correlation functions averaged over the different spots, average count rates. This extreme variation occurs because
again resulting in an ensemble average [5,6,58–61]. the detected signal consists of a dynamical component
H.Z. Cummins / Journal of Non-Crystalline Solids 353 (2007) 3891–3905 3903

Fig. 9. PCS data for sample AA5. Upper panel: C(t) and KWW fits after zero days (circles). 20 days (triangles), and 25 days (squares) when the sample is a
liquid. Lower panel: C(t) after 62 days the sample has gelled. Different data correspond to different heights in the cell and show the large variation in a/b
ratio caused by the random nature of the static scattering intensity.

superimposed on a static component which, if the particles largest for the small a/b ratio runs (large static intensity
were immobile, would produce the familiar speckle pattern causes a small a/b ratio) and is relatively constant. For
characteristic of scattering from a system of random fixed the translated samples, the count rate is very large and fluc-
scatterers. As Pusey and van Megen noted, this random tuates wildly as the sample moves. The decay of C(t) at
spatial property of the static component is the reason times of 0.1 s observed for these translated samples is
why the a/b ratio is so variable, and can be exploited by due to the motion of the sample and does not relate to
moving the sample around until a value near zero is found the intrinsic dynamics of the colloidal particles.
for the static component, resulting in an a/b ratio near to
1.0. As shown in Fig. 9, one spectrum has an a/b ratio of 4. Discussion
0.8 and is therefore close to the case they described. Also,
it appears that the apparent decay time becomes longer as We have carried out PCS measurements on aqueous
the a/b ratio decreases, but we have not attempted to verify solutions of Laponite XLG for three different preparation
this correlation quantitatively. methods, for a range of concentrations, and with different
We also recorded PCS spectra of some gelled samples filtration procedures. As in previous studies we found that
while slowly translating the sample tube vertically. The at concentrations below 1 wt% the aging process is very
a/b ratio was then nearly 1.0 as expected if ergodicity is slow and the PCS data are still evolving at times approach-
restored, and C(t) exhibits a high plateau that decays at ing one year. Samples prepared without pH control aggre-
long times. We also recorded count rate histories for these gated fastest in general, although one sample in this series
spectra. For the stationary sample cases, the count rate is (AA4) had not gelled a full year after preparation. For
3904 H.Z. Cummins / Journal of Non-Crystalline Solids 353 (2007) 3891–3905

the samples prepared with pH control, there was little differ- [9] W. van Megen, S.M. Underwood, Phys. Rev. E 47 (1) (1993) 248.
ence between those for which the pH was adjusted to a value [10] W. van Megen, S.M. Underwood, Phys. Rev. E 49 (5) (1994)
4206.
>10 after mixing was complete and those mixed with [11] W. van Megen, Transport Theor. Stat. Phys. 24 (6–8) (1995) 1017.
water whose pH had already been adjusted to a value >10. [12] E. Bartsch, M. Antonietti, W. Schup, H. Sillescu, J. Chem. Phys. 97
Comparing samples prepared without filtration, filtra- (6) (1992) 3950.
tion with a 0.8 lm pore size filter or with a 0.45 lm filter [13] W. Götze, in: J.-P. Hansen, D. Levesque, J. Zinn-Justin (Eds.),
gave ambiguous results. In some cases the unfiltered sam- Liquids, Freezing and the Glass Transition (Les Houches Summer
Schools of Theoretical Physics Session LI (1989)), North-Holland,
ple gelled first, the 0.8 lm sample second, and the Amsterdam, 1991, p. 287.
0.45 lm sample last, with the rate of increase in cluster size [14] W. Götze, L. Sjögren, Rep. Prog. Phys. 55 (1992) 241.
following the same sequence. But in some samples this [15] W. Götze, J. Phys.: Condens. Mat. 11 (1999) A1.
order was permuted or reversed. The filters were used as [16] J. Wu, J. Cao, Phys. Rev. Lett. 95 (2005) 78301.
obtained from the manufacturer (Millipore) and may con- [17] L. Fabbian, W. Götze, F. Sciortino, P. Tartaglia, F. Thiery, Phys.
Rev. E 59 (2) (1999) R1347.
tain small residues of detergent or solvent that influences [18] L. Fabbian, W. Götze, F. Sciortino, P. Tartaglia, F. Thiery, Phys.
the cluster growth and gelation processes. In future exper- Rev. E 60 (2) (1999) 2430.
iments the effect of flushing the filters with pure water [19] K. Dawson, G. Foffi, M. Fuchs, W. Götze, F. Sciortino, M. Sperl, P.
before use will be explored as will the effects of preparing Tartaglia, T. Voigtmann, E. Zaccarelli, Phys. Rev. E 63 (2000) 11401.
samples under a dry nitrogen atmosphere. [20] W. Götze, M. Sperl, J. Phys. Condens. Mat. 15 (2003) S869.
[21] T. Eckert, E. Bartsch, Phys. Rev. Lett. 89 (12) (2002) 125701.
[22] K.N. Pham, A.M. Puertas, J. Bergenholtz, S.U. Egelhaaf, A.
5. Conclusions Moussaı̈d, P.N. Pusey, A.B. Schofield, M.E. Cates, M. Fuchs,
W.C.K. Poon, Science 296 (2002) 104.
We conclude that the aging behavior of Laponite sus- [23] K.N. Pham, S.U. Egelhaf, P.N. Pusey, W.C.K. Poon, Phys. Rev. E 69
pensions is strongly affected by the sample preparation pro- (2004) 11503.
[24] V. Trappe, P. Sandkühler, Curr. Opin. Colloid Interf. Sci. 8 (2004)
cedure, making it essentially impossible to compare the 494.
results of experiments that follow different methods of [25] S. Mossa, F. Sciortino, P. Tartaglia, E. Zaccarelli, Langmuir 20
preparation. First, the speed with which Laponite particles (2004) 10756.
aggregate to form growing clusters is significantly higher [26] P.N. Segre, V. Prasad, A.B. Schofield, D.A. Weitz, Phys. Rev. Lett.
for samples with no pH adjustment than for those with 86 (2001) 6042.
[27] J. Bergenholtz, M. Fuchs, Phys. Rev. E 59 (5) (1999) 5706.
the pH >10. Second, filtration affects the rate of aggrega- [28] J. Bergenholtz, M. Fuchs, J. Phys. Condens. Mat. 11 (1999) 10171.
tion, but the relation between filter pore size and aggrega- [29] A.M. Puertas, M. Fuchs, M.E. Cates, Phys. Rev. Lett. 88 (9) (2002)
tion rate is not consistent. It is possible that residual 98301.
impurities in the filters used play a role, a possibility that [30] K. Kroy, M.E. Cates, W.C.K. Poon, Phy. Rev. Lett. 92 (2004)
requires further study. Finally, in contrast to previous 148302.
[31] F. Sciortino, S. Mossa, E. Zaccarelli, P. Tartaglia, arXiv:cond-mat
claims, we conclude that filtration does not completely (0312161v1) (2003).
break up the existing clusters and that aging that takes [32] F. Sciortino, S. Mossa, E. Zaccarelli, P. Tartaglia, Phys. Rev. Lett. 93
place between mixing and filtration is not completely (5) (2004) 55701.
reversed by filtration. [33] A. Kumar, J. Wu, Colloid. Surface A 247 (2004) 145.
[34] P.J. Lu, J.C. Conrad, H.M. Wyss, A.B. Schofield, D.A. Weitz, Phys.
Rev. Lett. 96 (2006) 28306.
Acknowledgement [35] P. Levitz, E. Lecolier, A. Mourchid, A. Delville, S. Lyonnard,
Europhys. Lett. 49 (5) (2000) 672.
This research was supported by the NSF under Grant [36] D.W. Thompson, J.T. Butterworth, J. Colloid Interf. Sci. 151 (1)
No. DMR-0243471. (1992) 236.
[37] F. Pignon, J.-M. Piau, A. Magnin, Phys. Rev. Lett. 76 (1996) 4857.
[38] F. Pignon, A. Magnin, J.-M. Piau, B. Cabane, P. Lindner, O. Diat,
References Phys. Rev. E 56 (3) (1997) 3281.
[39] D. Bonn, H. Kellay, H. Tanaka, G. Wegdam, J. Meunier, Langmuir
[1] W.B. Russel, D.A. Saville, W.R. Schowalter, Colloidal Dispersions, 15 (1999) 7534.
Cambridge University, Cambridge, 1991. [40] D. Bonn, H. Tanaka, G. Wegdam, H. Kellay, J. Meunier, Europhys.
[2] R.J. Hunter, Introduction to Modern Colloid Science, Oxford Lett. 45 (1) (1998) 52.
University, Oxford, 1993; [41] B. Ruzicka, L. Zulian, G. Ruocco, Phys. Rev. Lett. 93 (2004) 258301.
M.Y. Lin, H.M. Lindsay, D.A. Weitz, R.C. Ball, R. Klein, P. [42] B.J. Lemaire, P. Panine, J.C.P. Gabriel, P. Davidson, Europhys. Lett.
Meakin, Phys. Rev. A 41 (1990) 2005. 59 (1) (2002) 55.
[3] S.-H. Chen, F. Mallamace, F. Sciortino (Eds.)J. Phys. Condens. Mat. [43] J.-C.P. Gabriel, C. Sanchez, P. Davidson, J. Phys. Chem. 100 (1996)
16 (42) (2004). 11139.
[4] L. Cipelletti, L. Ramos, J. Phys. Condens. Mat. 17 (2005) R253. [44] R. Agra, F. vanWijland, E. Trizac, Phys. Rev. Lett. 93 (1) (2004)
[5] P.N. Pusey, W. van Megen, Phys. Rev. Lett. 59 (18) (1987) 2083. 18304.
[6] W. van Megen, P.N. Pusey, Phys. Rev. A 43 (10) (1991) 5249. [45] B. Abou, D. Bonn, J. Meunier, Phys. Rev. E 64 (2001) 21510.
[7] W. van Megen, S.M. Underwood, P.N. Pusey, Phys. Rev. Lett. 67 [46] R.G. Avery, J.D.F. Ramsay, J. Colloid Interf. Sci. 109 (1986) 448.
(12) (1991) 1586. [47] M. Bellour, A. Knaebel, J.L. Harden, F. Lequeux, J.-P. Munch, Phys.
[8] W. van Megen, S.W. Underwood, Phys. Rev. Lett. 70 (18) (1993) Rev. E 67 (2003) 31405.
2766. [48] S. Bhatia, J. Barker, A. Mourchid, Langmuir 19 (2003) 532.
H.Z. Cummins / Journal of Non-Crystalline Solids 353 (2007) 3891–3905 3905

[49] D. Bonn, S. Tanase, B. Abou, H. Tanaka, J. Meunier, Phys. Rev. [55] T. Nicolai, S. Cocard, Eur. Phys. J. E5 (2001) 221.
Lett. 89 (2002) 15701. [56] T. Nicolai, S. Cocard, J. Colloid Interf. Sci. 244 (2001) 51.
[50] R. DiLeonardo, F. Ianni, G. Ruocco, Phys. Rev. E 71 (2005) 11505. [57] L. Rosta, H.R. vonGunten, J. Colloid Interf. Sci. 134 (2) (1990) 397.
[51] A. Knaebel, M. Bellour, J.P. Munch, V. Viasnoff, F. Lequeux, J.L. [58] P.N. Pusey, W. van Megen, Physica A 157 (1989) 705.
Harden, Europhys. Lett. 52 (1) (2000) 73. [59] A.P.Y. Wong, P. Wiltzius, Rev. Sci. Instrum. 64 (1993) 2547.
[52] M. Kroon, G.H. Wegdam, R. Sprik, Phys. Rev. E 54 (1996) 6541. [60] S. Kirsch, V. Frenz, W. Schartl, E. Bartsch, H. Sillescu, J. Chem.
[53] M. Kroon, W.L. Vos, G.H. Wegdam, Phys. Rev. E 57 (2) (1998) 1962. Phys. 104 (1995) 1758.
[54] T. Nicolai, S. Cocard, Langmuir 16 (2000) 8189. [61] V. Viasnoff, F. Lequeux, Rev. Sci. Instrum. 73 (2002) 2336.

You might also like