10298 Langmuir 2002, 18, 10298-10304

NMR Self-Diffusion Studies on PDMS Oil-in-Water

Emulsion
Tatjana Garasanin and Terence Cosgrove*
School of Chemistry, University of Bristol, Cantock’s Close, U.K. BS8 1TS
Leon Marteaux
Dow Corning S.A., 83 Av. Mounier, 1200 Brussels, Belgium
Axel Kretschmer and Andy Goodwin
Dow Corning Ltd, Cardiff Road, Barry, South Glamorgan, U.K. CF63 2YL
Klaus Zick
Bruker Analytik GMBH D-76287 Rheinststten Silbersteifen, Germany

Received June 21, 2002. In Final Form: September 13, 2002

In this paper, standard spin-echo pulsed field gradient (SE-PFG) and stimulated spin-echo pulsed field
gradient (STE-PFG) 1H NMR methods have been used to study the translational diffusion of polymeric-
oil-in-water (O/W) emulsions, made with poly(dimethylsiloxane) PDMS fluids and two nonionic surfactants
polyoxethylene (4) lauryl ether (Brij-30) and polyoxethylene (23) lauryl ether (Brij-35p). A detailed analysis
of the free (Gaussian) diffusion of the PDMS fluid, its restricted diffusion inside the emulsion droplets,
and the Brownian diffusion of the droplets themselves are presented. A graphical representation, which
relates the square root of the mean-squared displacement of the fluid and emulsion droplets with the
experimental diffusion times and the radius of the droplets, helps distinguish the different diffusion regimes.
The Brownian diffusion of the emulsion particles becomes the dominant diffusion process with dilution
of the system and increase in the molecular weight (viscosity) of the PDMS. These different processes
which contribute to diffusion in the system need to be considered if a correct size distribution of the
emulsion droplets is to be determined. For a concentrated emulsion system, the condition for restricted
diffusion is reached by increasing the diffusion time. Under this condition, the Murday and Cotts model
for diffusion inside a spherical droplet can be used to determine the size distribution of the emulsion. In
very dilute systems by increasing the diffusion time, the Brownian diffusion becomes the dominant process
so that the Stokes-Einstein model can be applied to obtain the size distribution.

Introduction the contribution from the diffusion of the droplets them-

The PFG NMR technique has been used extensively to
investigate the size distribution in a variety of emulsion
systems by measuring the diffusion of fluid confined inside
emulsion droplets.1-14 A detailed review of the SE-PFG
NMR technique has been published by Price.15 However,
* To whom correspondence should be addressed. E-mail:
terence.cosgrove@bris.ac.uk.
(1) Stejskal, E. O.; Tanner, J. E. J. Chem. Phys. 1965, 42, 288.
(2) Tanner, J. E.; Stejskal, E. O. J. Chem. Phys. 1968, 49, 1768.
(3) Neuman, C. H. J. Chem. Phys. 1974, 60, 4508.
(4) Murday, J. S.; Cotts, R. M. J. Chem. Phys. 1968, 48, 4939.
(5) Packer, K. J.; Rees, C. J. J. Colloid Interface Sci. 1972, 40, 206.
(6) Callachan, P. T.; Jolley, K. W. J. Colloid Interface Sci. 1983, 93,
521.
(7) Van den Enden, J. C.; Waddington, D.; Van Aalst, H.; Van
Kralingen, c. G.; Packer, K. J. J. Colloid Interface Sci. 1990, 140, 105-
113.
(8) Lonnqvist, I.; Khan, A.; Söderman, O. J. Colloid Interface Sci.
1991, 144, 401.
(9) Söderman, O.; Lonnqvist, I.; Balinov, B. Emulsions - A Funda-
mental and Practical Approach; Kluwer Academic Publishers: Neth-
erlands, 1992; p 239-258.
(10) Li, X.; Cox, J. C.; Flumerfelt, W. AICHE J. 1992, 38, 1671.
(11) Balinov, B.; Jonsson, P.; Linse, P.; Söderman, O. J. Magn. Reson.
1992, 104, 17.
(12) Balinov, B.; Söderman, O.; Warnheim, T. JAOCS 1994, 71, 513.
(13) Appel, M.; Fleischer, G.; Karger, J. Macromolecules 1995, 28,
2345.
(14) Hakansson, B.; Pons, R.; Söderman, O. Langmuir 1999, 15, 988.
selves has not been considered explicitly. Callaghan and
Jolley6 studied water and fat diffusion in cheese, and van
den Enden et al.7 obtained the size distribution of water
droplets from the restricted diffusion model in margarine
and halvarine products. Balinov et al.11 used computer
simulation to compare the Gaussian phase distribution
(GPD) and the short gradient pulse (SGP) approximation15
for the cases of diffusion within spheres and between two
parallel and infinite planes. The GPD approximation was
found to be accurate for most experimental situations,
whereas the usefulness of the SGP limit is more restricted.
Subsequently Balinov, together with Söderman, and
Warnheim12 determined the size distribution of margarine
and low calorie spreads using the GPD approximation. In
this paper, we give a general picture of the diffusion
processes in emulsion systems where several distinct
diffusive processes can be observed.
Experimental Section
Materials. The nonionic surfactants used were commercial
products. Polyoxethylene (4) lauryl ether (Brij-30) was obtained
from ICI Surfactants Germany, and polyoxethylene (23) lauryl
(15) Price, W. S. Concepts Magn. Reson. Part I, 1997, 9, 299-336.

10.1021/la026109x CCC: $22.00 © 2002 American Chemical Society

Published on Web 12/02/2002
Studies on PDMS Oil-in-Water Emulsion Langmuir, Vol. 18, No. 26, 2002 10299

Figure 1. Standard spin-echo pulsed field-gradient experi- Figure 2. Stimulated spin-echo pulsed field-gradient experi-
ment (SE-PFG) with rectangular gradient pulses of duration ment (STE-PFG). The separation between the second and the
δ and magnitude g inserted into each τ delay. The separation third π/2 pulse is T. Remaining parameters are defined as in
between the leading edges of the gradient pulses is ∆. Figure 1.

ether (Brij-35p) was obtained from ICI Surfactants U.K. They intensity at time 2τ in the presence of gradient pulses of
were used as received. The poly(dimethylsiloxane) (PDMS) 1000 strength g in the case of free (Gaussian) diffusion is given
cS (MW ) 36.k kg/mol, MW/MN ) 2.43) and 350 cS (MW ) 21.9 by1
kg/mol, MW/MN ) 2.05) were obtained from Dow Corning Europe,
Brussels, Belgium. The deuterium oxide (D2O) 99.9% (min) was
obtained from Aldrich Ltd.
Sample Preparation. Concentrated 1000cS emulsions were
I(δ,∆,g,τ) ) I0 exp - ( ) (
2τ
T2
δ
exp -γ2g2δ2 ∆ - D
3 ( )) (1)

prepared using the following quantities: 15 g (69.8 wt %) Dow

Corning 1000 cS PDMS fluid, 0.6 g (2.8 wt %) Brij-35p, 0.9 g (4.2
wt %) Brij-30, and 5 g (23.2 wt %) D2O. A total of 1.5 g of Brij-35p
surfactant was heated at 60 °C in an oven until it melted, and
0.6 g of it was then added to 0.9 g of Brij-30 surfactant in 15 g
of the 1000 cS PDMS fluid. The mixture was homogenized with
a high speed Silverson mixer (7500 rpm) for 5 min. The solution
obtained had a low viscosity and was opaque (white). Water was
added to the dispersion while continuously mixing at a fixed
speed (max 7500 rpm). Beyond the addition of ∼1 g (5%) of water,
the solution became very viscous and gray, and a phase inversion
occurred. The emulsion changed from a water-in-oil type (W/O)
to an oil-in-water type (O/W).16 Adding more water (drop by drop)
and mixing, the solution gradually became less viscous and white
again. The preparation of the emulsion was carried out at room
temperature ∼25 °C. The weight percent of the emulsion (PDMS)
in the sample was 10.0 wt %.
where I0 is the signal intensity observed immediately after
the 90°x pulse, τ is the time between the 90°x and 180°y
pulses, ∆ is the distance between the leading edges of the
gradient pulses, δ is the length of the gradient pulse, D
is the self-diffusion coefficient, T2 is the spin-spin
relaxation time, and γ is the gyromagnetic ratio (1H, γ )
26.752 × 107 rad T-1 s-1).
The stimulated spin-echo NMR experiment (Figure 2)
has the advantage that the time ∆ is made up of two parts,
a shorter time (τ) in which T2 spin-spin relaxation takes
place and a longer time (T) in which T1, longitudinal
relaxation,15 occurs. For systems in which T2 is short but
T1 is long, this method makes it possible to extend the
diffusion time, ∆, considerably. The signal intensity at
time 2τ + T in the case of free diffusion is given by

( ) ( )
A commercial emulsion was prepared by Dow Corning Europe
in Brussels, Belgium, and contained 50 wt % Dow Corning 350 1 T 2τ
cS PDMS fluid, 3.86 wt % polyoxethylene (2-3) lauryl ether (Volpo
I(δ,∆,g,τ,T) ) I0 exp - exp -
2 T1 T2

( ( ))
L3), 1.54 wt % Brij-35p, 43.25 wt % H2O, and 1.35 wt % of other
δ
additives. exp - γ2g2δ2 ∆ - D (2)
NMR. The STE-PFG 1H NMR measurements were carried 3
out on a Bruker AVANCE 300 spectrometer using a 7.1 T wide
bore (89 mm) magnet and a Diff30 diffusion probe with actively where I0 is the signal intensity observed immediately after
shielded gradients. This configuration gives a maximum gradient the 90°x pulse.
strength of about 12 T/m with a current of 40 A. Measurements The mean-squared displacement15 is given by
were also performed on a JEOL FX100 high-resolution NMR
spectrometer operating at 2.3 T modified to carry out self-diffusion
measurements. The current amplifier used to generate the field
〈(r1 - r0)2〉 ) ∫-∞∞ (r1 - r0)2F(r0)P(r0,r1,t) dr0 dr1 (3)
gradients was an Amkron M-600 (up to 20 A). The unit was
calibrated with a sample of known diffusion coefficient (75/25 wt F(r0) is the probability that the spin trajectory begins at
% D2O/H2O mixture at 25 °C with a diffusion coefficient of 2.0 r0. P(r0, r1, t) is the diffusion propagator. For the case of
× 10-9 m2 s-1).17 The maximum value of the field gradient of the (three-dimensional) diffusion in an isotropic and homo-
spectrometer that can be applied is 1 T/m. All experiments were
geneous medium is
carried out at 25 °C.

( )
PCS. The photon correlation spectroscopy (PCS) equipment
used to measure the particle size was a “ZetaPlus” (Brookhaven (r1 - r0)2
Instruments Corporation). The wavelength of the laser was 678 P(r0,r1,t) ) (4πDt)-3/2 exp - (4)
4Dt
nm. Measurements were taken at room temperature on very
dilute dispersions (∼0.1 wt %).
Evaluation of eq 3 gives the Einstein equation,18 which in
three dimensions is
Theoretical Background
Free Diffusion. In the SE-PFG NMR experiment 〈(r1 - r0)2〉 ) 6D∆ (5)
(Figure 1), the echo attenuation gives information on
molecular displacement along the gradient axis (z axis) This result can be used as a variable (r ) r1 - r0) to describe
that has occurred during the time period ∆. This can be the various diffusion regimes in the system.
related to the self-diffusion coefficient D. The signal In the PFG NMR experiment, it is the molecular
displacement along the gradient axis (z axis) that is
(16) Becher, P. (Ed.), Encyclopedia of Emulsion Technology; Dekker:
New York, 1981;Vols. 1 and 2. (18) Johansson, L. Diffusion and interaction in gels and solutions;
(17) Mills, R. J. Phys. Chem. 1973, 77, 685. Chalmers University of Technology: Göteborg, Sweden, 1993.
10300 Langmuir, Vol. 18, No. 26, 2002 Garasanin et al.

measured. The respective mean-squared displacement is

〈(z1 - z0)2〉 ) 2D∆ (6)

Restricted Diffusion. In systems, where the diffusion

of molecules is restricted by physical boundaries, eqs 1
and 2 are no longer valid. The attenuation of the signal
intensities for the spin bearing molecules confined in a
spherical cavity of radius R was first derived by Murday
and Cotts,4 using the GPD approximation,19 and is given

[
by

I(R)
I /0
) exp
D(
-2γ2g2
)∑ ∞
m)1
Rm-4
Rm2R2 - 2
×

({ })]
Figure 3. Condition for the restricted diffusion. The x, y, and
z axes are the diffusion coefficient of the fluid, D, the diffusion
time, ∆, and the RMS displacement, (〈r2〉)1/2, respectively.
2 + exp(-R2mD(∆ - δ)) - 2 exp(-R2mDδ)
(gRγδ)2. In the case of the emulsion samples studied in
Rm2D this paper, the condition (13Dδ/3R2) , 1 could not be fully
2δ - reached.
-2 exp(-RmD∆) + exp(-R2mD(∆ + δ))
2
A graphical presentation of these regimes in concen-
Rm2D trated systems is given in Figure 3. In x and y, we have
two independent variables, the diffusion time (∆) and the
(7) diffusion coefficient of the fluid (D), respectively. The z
axis is the square root of the mean-squared displacement
We have omitted the decay due to spin relaxation for of the fluid ((〈r2〉)1/2), RMS) and it can be compared with
simplicity so that I /0 is the initial intensity as in eq 1 (SE- the radius of the emulsion droplet. If the radius of the
PFG) or eq 2 (STE-PFG) but including the relaxation term. particle is larger than the RMS displacement, there is a
Rm is the mth root of the Bessel equation shown below low probability that the fluid will travel far enough to feel

( )
the boundary of the droplet. The diffusion is unrestricted:
1
J (R R) ) 2J′3/2(RmR) (8) this is the region above the shaded surface. Close to the
RmR 3/2 m shaded surface, the diffusion is partially restricted as R
∼ (〈r2〉)1/2. Under this surface, the diffusion of the fluid is
Equation 7 has been evaluated using the software restricted as the RMS displacement becomes larger than
Mathematica v 3.0.20 It was found that the series in eq 7 the radius of the particle; that is, there is an increasing
converges very rapidly so that only the first term needs probability that fluid will be reflected from the walls of
to be retained. This allowed us to approximate the the droplets. By increasing the diffusion time, the
equation, with the first root of the Bessel equation (R1R displacement of the fluid is increased. For example, if we
) 2.08). The result is shown below (eq 9) again omitting have a fluid with a diffusion coefficient of D ) 10-12 m2
the decay because of spin relaxation for simplicity: s-1 confined inside a 1 µm emulsion droplet, effectively,
all of the fluid will feel the restriction if the diffusion time
( )
ln
I(R)
I0
)
2 2 2 2
(∆) is longer than 1 s.
Droplet Diffusion. In a concentrated system, we
54g2R4γ2(3(e13D(δ-∆)/3R - 2e-13Dδ/3R - 2e-13D∆/3R + e-13D(δ+∆)/3R + 2)R2 - 26Dδ)
2
assume that the Brownian diffusion of the particles is
15379D
(9) negligible, but in dilute systems, this is not always the
case. For a dilute suspension of spherical particles, the
If the product of the diffusion coefficient of the bulk fluid, diffusion coefficient (D) is given by the Stokes-Einstein
D, and the diffusion time, ∆, is much smaller than the equation:18
radius squared of the emulsion droplet, R2, the diffusing
fluid cannot diffuse far enough to feel the effect of the kBT
boundary, the measured diffusion coefficient is the same D) (10)
6πηRH
as that observed for a freely diffusing species (i.e. D∆ ,
R2) and is described by eq 1. When D∆ ≈ R2, a certain
where kB is the Boltzmann constant, T is the temperature,
fraction of the fluid feels the effects of the boundary and
η is the viscosity of the surrounding fluid (which in our
we measure an “apparent” diffusion coefficient (Dapp),
case is water), and RH is the hydrodynamic radius of the
which is ∆ dependent. When D∆ . R2, virtually all of the
emulsion droplets. For a hard sphere, RH ) R. From the
emulsified fluid feels the effects of restriction, and in this
diffusion coefficient, the hydrodynamic radius of the
case, the signal attenuation is sensitive to the diffusion
particles can be estimated. By replacing D in eq 1 with
coefficient, shape and dimensions of the droplet and is
the Stokes-Einstein equation (eq 10), the attenuation of
described by eq 7 and 9. Under this condition, eqs 7 and
the signal intensities becomes
9 can be further simplified if Rm2Dδ , 1 or expressed with

( ) ( )
the first root of the Bessel equation as (13Dδ/3R2) , 1.
I(δ,∆,g,τ) 2τ δ kBT
This reduces eq 7 to exp(-0.2(gRγδ)2) and eq 9 to 0.1978-
I0
) exp -
T2
exp -γ2g2δ2 ∆ - (
3 6πηRH )
(19) Douglas, D. C.; McCall, D. W. J. Phys. Chem. 1958, 62, 1102.
(20) www.wolfram.com. (11)
Studies on PDMS Oil-in-Water Emulsion
Figure 4. Diffusion regimes in a dilute oil-in-water emul-
sion system. The light gray plane is a simulation of the
Brownian diffusion of the particles (plane A). The medium
gray plane is a simulation of restricted diffusion of the
fluid when the condition for the total restriction is satisfied
(plane B), plot of R ) RMS. The dark gray plane is a simulation
of the diffusion of the bulk fluid with a diffusion coefficient of
D ) 10-12 m2 s-1 (plane C). The x, y, and z axes are the radii
of the particles, R, the diffusion time, ∆, and the RMS
displacement, (〈r2〉)1/2, respectively. The effective RMS displace-
ment (x〈rINT2〉 + 〈rEXT2〉) presented as a mesh surface.
Polydispersity of the droplets can also be taken into
account by numerically integrating over all droplet radii.
In the expression shown below, the contributions from
the various droplets is weighted by the volume of each
droplet:9
I ∫0∞ I(R)R3N(R) dR
) (12)
I0
∫0∞ R3N(R) dR
where, I(R) is found from eq 9 for the restricted model and
from eq 11 for the Brownian model. N(R) is the relative
frequency and is described in our case by the log-normal
distribution function as proposed by Packer and Rees5
( )
(ln 2R - ln 2R0) 2 always the dominant process and that the degree of
1 domination depends strongly on the particle size. If
N(R) ) exp - (13)
2Rσx2π 2σ2
where σ is the standard deviation of the ln 2R values.
The resultant mean radii, R0, and polydispersities, σ,
are obtained from a nonlinear least-squares fit (Math-
ematica v 3.020) to eq 12. To visually compare the results,
relative frequencies, N(R), are plotted as a function of the
droplets radius. The area under each curve is normalized
as
∫0∞ N(R) dR ) 1 (14)
In this way, plots are relative to the number of particles.
However, N(R) can also be normalized to be relative to
the volume of the particles if R3 is included in the integral.
In Figure 4, the diffusion of oil in a very dilute oil-in-
water emulsion system has been simulated. The x axis is
the diffusion time (∆), the y axis is the radius of the
emulsion droplets (R), and the z axis is the RMS displace-
ment.
The plane, which defines the boundary for Brownian
diffusion, is marked with a light gray color (plane A). Here,
Langmuir, Vol. 18, No. 26, 2002 10301
the RMS displacement of the emulsion droplet is calculated
from eqs 10 and 5 (for diffusion in three dimensions) and
plotted as a function of its radius (R) and the diffusion
time (∆).The height of this boundary surface increases
with an increase of the diffusion time and decreases with
an increase in the radius (R) of the droplets.
If the polymer inside the droplets is very viscous (D ∼
10-14 m2 s-1), it is effectively immobile inside the droplet.
For the whole radius range shown in the graph, Brownian
diffusion is the dominant diffusion process in the system.
In Figure 4, plane A illustrates the Brownian diffusion in
this system.
For less viscous polymers, the diffusion of the fluid itself
needs to be considered. How much this internal diffusion
contributes in the total diffusion measured depends on
the radius of the emulsion droplet (R) and the diffusion
time (∆). First, we will consider low viscosity polymers (D
∼ 10-9 m2 s-1, order of water) wherein diffusion is restricted
for all of the radius range given in the graph when ∆ >
0.1. This implies that the RMS displacement of the polymer
is of the order of the radius of the emulsion droplet. The
surface plot of the restricted diffusion is presented by the
darker gray color (plane B, plot of R ) RMS). The region
close to the intercept between plane A and B is critical.
Here, the diffusion of the fluid inside the emulsion droplets
is of the same order of magnitude as the Brownian diffusion
of the droplets. This region corresponds to the range of
the droplet radii from ∼100 nm to 5 µm, which is typical
for macroemulsions.16 The resultant mean-squared dis-
placement, which corresponds to the diffusion measured
by NMR, is a combination of Brownian diffusion (external
diffusion) and self-diffusion of fluid confined inside the
droplet (internal diffusion). The resultant displacement
can be expressed as the sum of their respective mean-
squared displacements:21
〈r2〉 ) 〈rINT2〉 + 〈rEXT2〉
(15)
However, out of this radius range, the situation is much
simpler. For very small droplets (smaller than 100 nm),
Brownian diffusion dominates, as 〈rINT2〉 is negligible and
for large droplets (∼10 µm) restricted diffusion can
dominate (〈rEXT2〉 is negligible). We can conclude that for
low viscosity emulsified fluids Brownian diffusion is not
particles are very large (∼10 µm), and restricted diffusion
dominates, the size distribution can be determined using
Murday and Cotts4 model. For the critical range (from
∼100 nm to 5 µm), by increasing the diffusion time (∆),
the contribution from the inner diffusion of the polymer
can be reduced so that Brownian diffusion becomes more
dominant. This is especially important for the case of
polydisperse emulsion systems.
For the intermediate viscosity polymers, the diffusion
process in the system is further complicated. In addition
to the Brownian and restricted diffusion, depending on
the droplets size, we can have partially restricted or
unrestricted diffusion. To emphasize this in Figure 4, the
surface plot of the bulk diffusion of a fluid with diffusion
coefficient of 10-12 m2 s-1, which is in the order of the
diffusion of the polymers we used in our work, is shown
by the dark gray color (plane C). The RMS displacement
of the bulk fluid is calculated from eq 5. To visualize more
easily the different diffusion regions, a plot of the resultant
RMS displacement (which can be regarded as equivalent
(21) Van de Ven, Theo C. M. Colloidal hydrodynamics; Academic
Press Limited: London, 1989.
10302 Langmuir, Vol. 18, No. 26, 2002
Figure 5. Signal attenuations (I/I0) versus the magnitude of
the field gradient g for 10 wt % 1000 cS PDMS emulsion for a
diffusion time of ∆ ) 0.1 (open circles), 1.0 (open squares), and
2.0 s (open triangles). The duration of the field gradient was
δ ) 2 ms. The full lines are fits to the model for Brownian
particle diffusion. The small filled circles represent the fit of
the data with diffusion time of ∆ ) 0.1 s (open circles) using
the model for restricted diffusion.
to displacement measured by NMR) is plotted as a mesh
surface in Figure 4. It has been calculated using eq 15 as
follows. 〈rmesh2〉 ) 〈rbulk,B2〉 + 〈rBrownian,A2〉 when plane B is
lower or close to the plane C (bulk or partially restricted
diffusion case) and 〈rmesh2〉 ) 〈rrestricted,C2〉 + 〈rBrownian,A2〉
when plane B is higher than plane C (restricted diffusion
case). For very large droplets, the mesh surface approaches
the plane for bulk diffusion (plane C), and the bulk
diffusion of the fluid dominates. In the region close to the
intercept between planes for bulk and restricted diffusion
of the fluid (plane C and B), partially restricted diffusion
dominates. Close to the intercept between the restricted
and Brownian diffusion planes (B and A, the critical range
from ∼100 nm to 5 µm), the restricted diffusion of the
polymer inside the droplet is of the same order of
magnitude as the Brownian diffusion. For very small
particles, the mesh surface approaches the plane for the
Brownian diffusion (plane A).
Experimental Results
The results for the 10 wt % 1000 cS PDMS emulsion
using the STE-PFG sequence22 with the duration of the
field gradient, δ ) 2 ms and diffusion times ∆ ) 0.1, 1.0,
and 2.0 s are given in Figure 5 (300 MHz Bruker data).
The diffusion time has been increased in order to enhance
the contribution of the Brownian diffusion of the droplets
and diminish the contribution of the partially restricted
diffusion of the fluid (as described above). From nonlinear
fits to the Brownian model, the resultant mean radius,
R0, and polydispersity, σ, of the log-normal distribu-
tion function were obtained. The results are shown in
Table 1.
The fit to the Brownian model for a data obtained with
a very short diffusion time of 0.1 s (Figure 5) gives a mean
radius for the droplets, R0 ) 0.14 ((0.01) µm and
polydispersity σ ) 0.54 ((0.03). These data may also be
fitted to the model for restricted diffusion, which indicates
that for such a short diffusion time the distance travelled
by the droplets is of the same order of magnitude as the
size of the particles.
A fit to the restricted diffusion model gives a value of
R0 ) 0.41 ((0.008) µm and polydispersity σ ) 0.40 ((0.06).
These values are larger than from the Brownian model
as the polymer has travelled not only the distance allowed
(22) Gibbs, S. J.; Johnson, C. S., Jr. J. Magn. Reson. 1991, 93, 395.
Garasanin et al.
Table 1. Results of the STE-PFG NMR Volume Weighted
Size Distribution Determination for 10 wt % 1000 cS
PDMS Oil-in-Water Emulsion as a Function of the
Diffusion Time ∆ and Volume Weighted Size Distribution
Results from the PCS Measurement of the Same
Emulsion but Dilute to 0.1 wt %
NMR ∆/s R0/µm ( 2%a σ ( 5%a
0.1 0.14 0.54
1.0 0.18 0.51
2.0 0.18 0.50
PCS R0/µm σ
0.15 0.35
a Errors estimated from data fitting.
Figure 6. Relative frequency N(R) versus radius of the particles
R for the 1000 cS PDMS emulsion. Black circles correspond to
a ∆ ) 0.1 s, white circles correspond to a ∆ ) 1.0 s, and black
triangles correspond to a ∆ ) 2.0 s, NMR data. The empty
squares are the PCS results. Data normalized using eq 14.
by the size of the droplet but also from the diffusion of the
droplet itself and corresponds to the mesh region around
the interception between all three planes A, B, and C in
Figure 4. However, for such a small diffusion time, the
diffusion of the polymer inside the largest droplets is only
partially restricted which increases the effective diffusion
coefficient. For the longer diffusion times of 1.0 and 2.0
s, the restricted model gives unfeasible results which
indicates that the distance travelled by the droplets is
now much larger than their size. This corresponds to the
mesh region close to the plane A under the dark gray
triangle in Figure 4 (D). The contribution from the
Brownian diffusion of the droplets is significantly larger
than the diffusion of the polymer itself, which allows us
to obtain a correct value for the particle size using the
Brownian model.
In Figure 6 and Table 1, the resultant particle size
distributions from NMR and PCS are compared. For a
short diffusion time of 0.1 s (black circles), the fit to the
Brownian model gives a mean radius R0 ) 0.14 ((0.003)
µm and polydispersity σ ) 0.54 ((0.03). When the diffusion
time is substantially increased to 1.0 and 2.0 s (white
circles and black triangles, respectively), the resultant
radius distributions are very close giving the same value
(Table 1). As the diffusion time is increased, the Brownian
diffusion of the emulsion particles becomes the dominant
diffusion process, and diffusion of the internal fluid
becomes negligible, giving larger values for the particle
size. The results from PCS measurements for the 0.1 wt
% 1000 cS emulsion (white squares) give a mean radius
for the emulsions particles of 0.15 µm and polydispersity
of 0.35. The main difference between the NMR results
(with enough large diffusion times 1.0 and 2.0 s) and PCS
is that a higher proportion of larger particles are found.
Studies on PDMS Oil-in-Water Emulsion
Figure 7. Signal attenuation (I/I0) versus the duration of the
field gradient δ for the concentrated 50 wt % commercial 350
cS PDMS emulsion (circles) and for a bulk 350 cS PDMS fluid
(triangles). The magnitude of the field gradient g ) 0.81 T/m.
Black circles and triangles are the data with a diffusion time
of ∆ ) 2.04 s, and the white circles and triangles are the data
with a shorter diffusion time ∆ ) 0.74 s. The solid lines are the
fit to the model for restricted diffusion of the emulsion data,
and the dotted lines are a Gaussian diffusion model for the
bulk PDMS data.
Figure 8. Relative frequency N(R) versus radius of the particles
R for the commercial 50 wt % 350 cS PDMS emulsion. Black
circles correspond to a diffusion time of ∆ ) 2.04 s for which
the diffusion is restricted, and white circles correspond to ∆ )
0.74 s for which the diffusion is partially restricted. Data
normalized using eq 14.
From Figure 6, it can be seen that the most probable radius
is the same for both techniques. A full comparison of the
NMR method with other sizing methods will be published
separately.
In Figure 7, we present data (100 MHz JEOL NMR)
from a concentrated 50 wt % 350 cS PDMS commercial
emulsion (circles) and bulk 350 cS PDMS (MW/MN ) 2.05)
fluid (triangles) using the STE-PFG sequence22 for two
different values of the diffusion time. The value of the
field gradient was g ) 0.81 ((0.01) T/m and diffusion times
0.74 and 2.04 s (white and black symbols, respectively).
Considering that the PDMS fluid is very polydisperse (MW/
MN ) 2.05), its average diffusion coefficient has been
estimated using the Gaussian1 model and is 2.64 ((0.2)
× 10-12 m2 s-1. The signal attenuation from the emulsion
(circles) becomes smaller as the diffusion time increases
as opposed to the attenuation of the bulk fluid (triangles),
which gets larger. This shows that, by increasing the
diffusion time, the diffusion of the PDMS inside the droplet
becomes more restricted. The resultant size distributions
obtained using the model for restricted diffusion are given
in Figure 8. It can be seen that as the diffusion time is
increased the distribution has moved toward smaller
effective radii and the polydispersity is decreased. This is
Langmuir, Vol. 18, No. 26, 2002 10303
Figure 9. Signal attenuation (I/I0) versus the duration of the
field gradient δ for the 1000 cS PDMS emulsion. The diffusion
time was ∆ ) 2.04 s, and the magnitude of the field gradient
was g ) 0.81 T/m. Filled circles are the data from the
concentrated 60 wt % 1000 cS emulsion. Empty circles are the
data from the dilute 10 wt % 1000 cS emulsion.
Figure 10. Relative frequency N(R) versus radius of the
particles R for the 1000 cS PDMS emulsion. White circles
correspond to a 60 wt %, and black circles correspond to a 10
wt %. Data normalized using eq 14.
because the droplet restricts PDMS fluid diffusion. For a
diffusion time of ∆ ) 0.74 s (white circles), a mean radius
for the emulsion droplets R0 ) 0.25 ((0.01) µm and
polydispersity σ ) 0.64 ((0.09) are obtained. For the longer
diffusion time of ∆ ) 2.04 s (black circles), a mean radius
for the emulsion droplets R0 ) 0.14 ((0.01) µm and
polydispersity σ ) 0.57 ((0.08) are obtained and are the
best estimates as full restriction is measured.
In Figure 9, the results (100 MHz data) from a 10 (white
circles) and 60 wt % (black circles) 1000 cS emulsion are
presented. The diffusion time used was 2.04 s, sufficiently
long for Brownian diffusion to be dominant in the dilute
system and for the diffusion of the PDMS molecules in the
concentrated emulsion system to become restricted. The
value of the field gradient was g ) 0.81 ((0.01) T/m. The
resultant size distributions are shown in Figure 10. The
fit to the Brownian diffusion model of the 10 wt % emulsion
data gives a mean radius Rout 0 ) 0.18 ((0.01) µm and
polydispersity σ ) 0.42 ((0.02) (black circles). The fit to
the model for restricted diffusion of the 60 wt % emulsion
gives Rinn0 ) 0.16 ((0.01) µm and polydispersity of σ )
0.62 ((0.09) (white circles). The result from the dilute
system gives a larger value as the outer radius of the
droplets (hydrodynamic radius) is measured. The differ-
ence between resulting outer and inner mean radius of
the droplets (Rout0 - Rinn
0 ) is related to the interfacial
thickness though this number exceeds the extended length
of the surfactants considerably, and other factors such as
10304 Langmuir, Vol. 18, No. 26, 2002
a stationary PDMS layer near the interface and hydro-
dynamic effects need to be investigated.
Conclusions
The contributions from all of the diffusion processes in
an emulsion system need to be considered if a correct size
distribution is to be determined. For a concentrated
emulsion system, an increase of the diffusion time enables
the condition for restricted diffusion to be satisfied. This
allows us to apply the Murday and Cotts4 model for
diffusion inside the spherical particle to determine the
Garasanin et al.
size distribution. By increasing the diffusion time in very
dilute systems, the Brownian diffusion of the particles
dominates so that the Stokes-Einstein18 model can be
applied for size distribution analysis.
Acknowledgment. T.G. would like to acknowledge
BRUKER for measurements with their 300 MHz NMR,
Dow Corning Ltd., Barry and the University of Bristol for
funding.
LA026109X