Research Article

Received: 14 January 2009 Revised: 19 February 2009 Accepted: 17 April 2009 Published online in Wiley Interscience: 14 August 2009

(www.interscience.wiley.com) DOI 10.1002/pi.2656

On-line acoustic attenuation spectroscopy

of emulsions stabilized by vinyl alcohol–vinyl
acetate copolymers: a model system
for the suspension polymerization of vinyl
chloride
Virginie Boscher,a Renaud Helleboid,b Thierry Lasuye,b Bernard Stasikb
and Gérard Riessa∗

Abstract
BACKGROUND: Poly[(vinyl alcohol)-co-(vinyl acetate)] (PVA) copolymers obtained by partial hydrolysis of poly(vinyl acetate)
are currently used as industrial stabilizers in the suspension polymerization of vinyl chloride monomer (VCM). Their molecular
characteristics, mainly the average degree of hydrolysis (DH) and average degree of polymerization (DPw), have a major
influence on the monomer droplet size and the properties of the final poly(vinyl chloride) resin.

RESULTS: The average droplet size and size distribution of chlorobutane/water emulsions, as a model system for VCM/water
emulsions, were studied using acoustic attenuation spectroscopy on-line with an agitated laboratory reactor. The emulsions
were stabilized by PVA with DH values between 73 and 88 mol% and DPw values between 450 and 2500. The effects of
agitation speed, stirring time and concentration of the PVA copolymers were investigated. An attempt was made to correlate
the interfacial tension and the droplet size.

CONCLUSION: On-line acoustic spectroscopy appears to be a suitable technique for the real-time control of the droplet size
of monomer suspensions. The advantages and limitations of the technique are outlined. The validity and the application
limits of the commonly cited correlation between the droplet size and the Weber number are established for polymeric
surfactant-stabilized emulsions.

c 2009 Society of Chemical Industry

Keywords: PVA; emulsion stabilization; on-line acoustic attenuation spectroscopy; droplet size distribution; interfacial properties

INTRODUCTION For the optimization of the polymerization process, it is of

The industrial production of poly(vinyl chloride) (PVC) is essentially interest to identify the factors governing the stabilization of
achieved by suspension polymerization, a process where droplets the initial VCM/water emulsions, keeping in mind that these
of vinyl chloride monomer (VCM), dispersed in an aqueous emulsions, in a size range of 20–100 µm, are unstable in the
absence of agitation and have a tendency to coalesce. Therefore it
phase, are polymerized by free radical initiation. In this type
is important to determine, especially for on-line control, not only
of turbulence-stabilized process, the coalescence of VCM droplets
the average size, but also the size distribution of the emulsion
is considerably reduced by the combined action of agitation and of droplets.
so-called ‘protective colloids’ such as vinyl alcohol–vinyl acetate In the past, these characteristics for VCM or corresponding
copolymers (PVAs) obtained by the partial hydrolysis of poly(vinyl model systems were studied by withdrawal of samples from
acetate) (PVAc). Owing to their hydrophilic poly(vinyl alcohol)
and hydrophobic PVAc segments in the polymer chain, these
copolymers adsorb at the VCM/water interface and provide steric ∗ Correspondence to: Gérard Riess, Laboratoire COBM, Ecole Nationale
stabilization of the VCM droplets.1 The molecular characteristics Supérieure de Chimie de Mulhouse, 3 rue Alfred Werner, 68093 Mulhouse
Cedex, France. E-mail: [email protected]
of PVAs, such as average degree of hydrolysis (DH), average
degree of polymerization (DPw), sequence distribution of acetate a University of Haute Alsace, ENSCMu, Laboratoire de Chimie Macromoléculaire
and alcohol segments and unsaturations, have a very important COBM, 3 rue Alfred Werner, 68093 Mulhouse Cedex, France
influence on the stability of VCM/water emulsions and on the
1209

b LVM-Tessenderlo Group, Département Qualité-Innovation, BP 49, 62160 Bully-

properties of the final PVC resin.1,2 les-Mines, France

Polym Int 2009; 58: 1209–1216 www.soci.org

c 2009 Society of Chemical Industry
 www.soci.org V Boscher et al.

liquid–liquid dispersions prepared in agitated reactors. The An emulsion drop should break up at a critical Weber number
droplet characteristics were then determined off-line using either (We) which is a function of different parameters, such as stirring
microphotographic3,4 or laser diffraction5 techniques. The main speed, volumic mass, interfacial tension, etc. For defined reactor
limitation of these techniques is that in general the emulsions have characteristics, the maximum stable drop size Dmax above which
to be diluted and coalescence cannot always be avoided under the drop will definitively break up is given by Dmax = KWe−0.6 ,
static conditions, especially for relatively unstable emulsions of where K is a constant for a given system.16
large droplet size. Moreover, sampling techniques can often be Within this context, the emulsifying efficiency of the PVA
non-representative and should be treated with care. polymeric surfactants was evaluated and an attempt was made to
For these reasons, the on-line monitoring of droplet size has correlate the interfacial tension of the ClBu/water system with the
appeared to be the most appropriate technique for suspension average droplet size.
polymerization studies. Chung and Wasan,6 by using a specially
designed in situ microphotographic technique, monitored the
evolution of the droplet size distribution for a VCM model emulsion. EXPERIMENTAL
Recently, in situ real-time measurements of droplet sizes and size Materials
distributions of various liquid dispersions have been made possible Chlorobutane (>99%), supplied by Acros, was selected as organic
by the development of new particle size analysers based on in situ phase model for VCM. This model has been used quite extensively
video microscopy,7,8 on laser backscattering9,10 and on ultrasonic by different authors,5,19,20 as its characteristics such as density,
spectroscopy.11 – 13 This latter technique, also called acoustic refractive index and solubility are close to those of VCM, with the
attenuation spectroscopy, is based on the frequency-dependent advantages of easier handling and less toxicity.
extinction of ultrasonic waves arising from particles. It offers The continuous phase of emulsions consisted of demineralized
the unique opportunity for the characterization of concentrated water with the required amount of PVA. The PVAs examined in
dispersions and emulsions in their original states, without any this study were commercial samples supplied by Kuraray, Nippon
further dilution that may change the dynamic equilibrium of the Gohsei and Synthomer. Hereafter, these samples, used without
system.11 It is with the commercial development of ultrasonic further purification, are identified by their DH and DPw values; for
spectrometers that this technique has become available for the instance, PVA-73-700. These PVA main characteristics, determined
on-line or in-line particle size analysis in industrial environments, using 1 H NMR and SEC, are summarized in Table 1.
mainly for process control involving solid dispersions.11,14 The value of DH with a precision of ±1 mol% was determined
Ultrasonic spectroscopy has been applied for the characteri- using 1 H NMR (Bruker AC-400F operating at 400 MHz) in
zation of bromobenzene, silicone oil or vegetable oil aqueous dimethylsulfoxide-d6 at 70 ◦ C according to Van der Velden
emulsions, but, to the best of our knowledge, on-line acoustic and Beulen.21 The SEC measurements were carried out with
spectroscopy has not been studied so far for the suspension a Shimadzu LC-20AD liquid chromatograph equipped with
polymerization of VCM, which in its initial step consists of dispers- two Varian PL gel 5 µm MIXED-C columns (column, injection
ing liquid VCM in an aqueous medium containing the required and refractometer temperature: 30 ◦ C; injection volume: 100 µL;
amount, generally below 0.15 wt% with respect to VCM, of PVA solvent: tetrahydrofuran at 1 mL min−1 ) and a refractive index
stabilizer. The starting point of the process corresponds therefore detector (Shimadzu RID-10A). The PVA samples were previously
to VCM dispersions with a droplet size of 20–100 µm. Thus, the reacetylated as recommended by Bugada and Rudin22 and the
aim of the study reported here was at first to determine the ‘universal calibration technique’23 with polystyrene standards was
conditions under which attenuation acoustic spectroscopy can applied for the calculation of Mn and Mw .
be adapted to the characterization of chlorobutane (ClBu)/water
emulsions of large droplet size, as a model for the initial VCM/water Interfacial tension
emulsions. The average droplet size and size distribution of these The dynamic and equilibrium interfacial tension between ClBu
turbulence-stabilized emulsions were examined, on the one hand, and aqueous PVA solutions was measured at 20 ◦ C using a
as a function of the reactor parameters, such as agitation speed pendent drop tensiometer (Krüss DSA 100). The PVA solutions
and time, and, on the other hand, as a function of the PVA charac- were separately prepared by solubilization of the polymer in water
teristics, in particular their DH, DPw and concentration. A second at 70 ◦ C. The evolution of the interfacial tension as a function of
objective of practical interest, as suggested by Lazrak et al.,15 was time was calculated from the change of the aqueous drop shape,
to determine to what extent the interfacial tension generated by formed in a ClBu continuous medium.24 The average accuracy of
polymeric surfactants such as PVAs would allow one to predict the the measurements was typically 0.1 mN m−1 .
droplet size of emulsions. Hence, according to Hinze,16 Shinnar17
and more recently Jahanzad et al.,18 it was demonstrated that Emulsion preparation
in liquid–liquid dispersions the size of the drops is determined ClBu/water emulsions were prepared at 20 ◦ C in a standard
by the balance between the rates of breaking and coalescence. conformation 1 L vessel of 0.082 m internal diameter (Fig. 1). This

Table 1. Molecular characteristics of the PVA samples

PVA-73-700 PVA-78-720 PVA-80-2100 PVA-88-450 PVA-88-2500

DH (mol%) 73 78 80 88 88
Mw (g mol−1 ) 37700 39000 109000 22050 121300
Mn (g mol−1 ) 18400 11150 50450 9200 57800
DPw 700 720 2100 450 2500
1210

www.interscience.wiley.com/journal/pi

c 2009 Society of Chemical Industry Polym Int 2009; 58: 1209–1216
Acoustic attenuation spectroscopy of stabilized emulsions www.soci.org

82 mm

42 mm

Figure 2. Typical attenuation spectra of water and ClBu/water emulsion

(
= 2 wt%; [PVA-73-700] = 20 ppm/water; stirring time = 60 min) as a
function of ultrasound frequency.

advertised as to enable fast and sensitive on-line measurements of

concentrated emulsions for droplet size distributions from 1 µm
to 3 mm.25 The emulsion was pumped through the OPUS sensor,
82 mm

which comprised an emitter and a receiver of ultrasound. The

ultrasound attenuation A by the emulsion droplets was determined
8 mm
in a typical ultrasonic frequency range of 1–200 MHz.
28 mm

For the present study, the following settings were used:

31 predetermined frequencies within the range 1–120 MHz
corresponding to 31 droplet size classes; a gap width of 2 mm;
and a droplet size measuring range of 0.5–500 µm. A background
8 mm 2 mm
measurement of the ultrasound attenuation by pure water was
performed at the start of the experiment and the attenuation
Figure 1. Experimental scheme of the reactor and impeller.
by the ClBu/water emulsion was measured every minute. For the
interpretation of the attenuation spectra, the densities, viscosities
and, in particular, the speed of sound in water and ClBu, 1483 and
round-bottomed glass vessel (Büchi BEP 280) was fitted with two
1140 m s−1 , respectively, have to be known.
equally spaced stainless steel baffles and mechanically agitated
Typical attenuation spectra as a function of the ultrasound
with a stainless steel three-curved-blade turbine of diameter
frequency are shown in Fig. 2, in which A is the ultrasounds
d = 0.042 m.
attenuation difference between pure water and the ClBu/water
The PVA solutions were prepared as mentioned previously by
emulsion. Owing to the working limits of the OPUS equipment,
dissolving PVA samples in water at 70 ◦ C. In a typical experiment,
the ultrasound attenuation A has to be kept below 145 dB. For
the reactor is first filled with 862.4 g of demineralized water
the ClBu/water emulsions, A is only significant for the high-
containing the required amount of PVA aqueous solutions. ClBu,
frequency range. Therefore, the frequency range of interest for
at a weight percentage
of 2 wt% (corresponding to 2.27 vol%),
the droplet size characterization of the ClBu/water emulsions
was then added to the vessel and the agitation speed was adjusted
is typically 50–120 MHz. A is then transformed into the
at 700 rpm. The PVA concentrations were between 5 and 610 ppm
corresponding droplet size distribution by mean of Lambert–Beer-
with respect to water, which corresponded therefore to 0.025 and
type equations.26 In this way, one has access to the droplet size
3 wt% with respect to ClBu.
distribution, the volume mean diameter Dx (meaning that x vol%
Due to its overall dimensions, the ultrasonic granulometer had
of the droplets have a mean diameter lower than Dx ) and the
to be connected in parallel to the reactor. The circulation of the
volume fraction of dispersed phase Cvol .
emulsion was performed from the bottom of the reactor to the
measuring cell of the granulometer, through Viton tubing, by
means of a low-shear peristaltic pump operating at a flow rate
of 1.26 L min−1 . In these conditions, it has been calculated that
RESULTS AND DISCUSSION
Interfacial tension
the emulsion arrives from the bottom of the reactor into the
granulometer measuring cell within 7 s. By stopping the agitation, It is now well established that the interfacial tension γ , in static
but by maintaining the circulation, it could be checked that the as well as in dynamic conditions, is an important parameter for
emulsion characteristics do not change within that time interval. emulsions, as already shown by different authors for ClBu/aqueous
emulsions stabilized by PVAs.19,27 In particular, He et al.27 in a recent
study examined the dynamic interfacial tension and the adsorption
Acoustic attenuation spectroscopy behaviour of PVA at the ClBu/water interface, for different PVAs,
In this study, the diameters of emulsion droplets were determined mainly in the concentration range 50–1000 ppm/water. In order to
using an OPUS ultrasonic granulometer and its WINDOX 5.2 complete this study in view of industrial applications, it is therefore
1211

software. This equipment, commercialized by Sympatec, is of interest to examine a larger concentration range, even below

Polym Int 2009; 58: 1209–1216

c 2009 Society of Chemical Industry www.interscience.wiley.com/journal/pi
 www.soci.org V Boscher et al.

40 30

Interfacial tension (mN m-1)

Interfacial tension (mN m-1)

35 25

30 20

25 15

20 10

15 5

10 0
0 100 200 300 400 0 50 100 150 200 250
Time (s) PVA concentration (ppm/water)
PVA-88-2500 PVA-88-450 PVA-80-2100
PVA-88-2500 PVA-88-450 PVA-80-2100
PVA-78-720 PVA-73-700
PVA-78-720 PVA-73-700
Figure 4. ClBu/water interfacial tension at 400 s (γ 400 ) as a function of
Figure 3. ClBu/water interfacial tension γ as a function of time for the concentration for the various PVAs.
various PVAs at a concentration of 20 ppm/water.

Emulsion characteristics
50 ppm, with PVAs of different DH and DPw, this latter parameter The turbulence emulsification processes in agitated reactors have
having not directly been taken into account by He et al.27 been quite extensively studied in order to correlate the drop size
An example showing the decrease with time of the interfacial distribution of the dispersion to the physical characteristics of
tension for ClBu/water for the various PVA samples at a concen- the liquids and to the mixing conditions.33,34 From these studies,
tration of 20 ppm/water is shown in Fig. 3. This concentration of carried out for aqueous dispersions of organic solvents, mostly in
20 ppm/water corresponds for the present emulsion formulation the absence of surfactants, it turns out that for isotropic turbulence
to a concentration of 0.1 wt%/ClBu, which is a typical PVA concen- conditions in the reactor, and if the droplet break-up with respect
tration for industrial VCM polymerizations. It can be noticed that to droplet coalescence is the predominant effect, the maximum
in the time interval 0–400s and whatever the PVA type, the inter- stable drop Dmax is defined according to Hinze16 as follows:
facial tension decreases with time without reaching a complete
−0.6
equilibrium. This observation is in agreement with the results of Dmax N2 d 3 ρ
= KWe−0.6 = K
Lankveld and Lyklema,28 as well as He et al.,27 who have shown d γ
that the real equilibrium is only obtained under static conditions
after 20–30 h. = KN−1.2 d−1.8 ρ −0.6 γ 0.6 (1)
The initial rapid decrease of the interfacial tension with time
corresponds to a diffusion-controlled adsorption of the copolymer where We is the Weber number, N (rpm) the agitation speed, d
at the liquid–liquid interface. This is followed by a decrease of (m) the impeller diameter, ρ (kg m−3 ) the volumic mass, γ (N m−1 )
minor importance generally attributed to conformation changes the interfacial tension and K a constant.
of the copolymer located at the interface.27 – 29 These trends, typical Different authors3,6,15,35 have shown that the droplet size of
for polymeric amphiphiles, were also observed by other authors polymeric surfactant-stabilized emulsions can be predicted by
for, among others, modified dextrane,30 cellulosic derivatives31 Eqn (1). However, only a few of these studies were related to
and modified chitosan.32 In fact, according to these studies, the dispersions of VCM, or VCM ‘model liquids’, stabilized in aqueous
adsorption of PVA at the ClBu/water interface has to be considered medium by PVAs3,6 and no systematic correlation was established
as a three-step process involving diffusion, surface adsorption and between the interfacial characteristics of the PVAs and the
conformation rearrangement of the polymers at the interface. droplet sizes of the dispersed phase. Our objective was therefore
As the interfacial tension at 400 s (γ 400 ) corresponds primarily to conduct a more detailed experimental study with PVAs of
to the adsorption step of PVA at the interface, and also for practical different DH and DPw by using an on-line acoustic attenuation
industrial reasons, γ 400 will be considered as a standard value in spectroscopic technique.
this study.
From Fig. 3, it appears furthermore that PVA-73-700 is the most Influence of stirring time
surfactive efficient, as it leads rapidly to the lowest interfacial A typical example of the droplet size distribution as a function
tension, followed by the other PVA samples of higher DH. This is a of the stirring time is shown in Fig. 5, for a standard emulsion
first evidence that the efficiency of PVA-type surfactants depends formulation of 2 wt% ClBu and 20 ppm/water of PVA-73-700.
on DH and DPw. From an initial broad size distribution and an average droplet
In order to rank the surfactive efficiency of the different PVAs, size of about 70 µm, these emulsion characteristics decrease
γ 400 was determined as a function of their concentration (Fig. 4). progressively with the agitation time. In fact, both size distribution
As expected, γ 400 decreases and then levels off with increasing and average size level off with stirring time and reach almost
copolymer concentration. Furthermore, in agreement with He constant values after 50–60 min of stirring, which correspond to
et al.,27 it can be noticed that the plateau value of γ 400 increases the dynamic equilibrium between the break-up rate of the droplets
with the DH value of the copolymer. Moreover, it appears that the and their coalescence rate. This evolution towards a steady state
surface activity not only increases with decreasing DH values, but can furthermore be explained by the time required for the PVA
1212

also with decreasing DPw values. molecules to be adsorbed onto the droplet surface, to rearrange

www.interscience.wiley.com/journal/pi

c 2009 Society of Chemical Industry Polym Int 2009; 58: 1209–1216
Acoustic attenuation spectroscopy of stabilized emulsions www.soci.org

30
Table 2. Droplet characteristics as a function of the weight and
volume ClBu concentration
([PVA-73-700] = 20 ppm/water; stirring
Differential distribution

25
time = 60 min)
20

(wt%)
(vol%) Cvol (vol%) D50 (µm) PI (µm) Amax (dB)
15
2 2.27 2.20 49.7 36.9 122
10 3 3.40 3.33 49.7 37.3 134
4 4.52 4.53 48.8 36.4 145
5
5 5.64 (9.30) (20.2) (34.2) 156
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Droplet diameter (µm)
5 min 20 min 45 min 60 min
From these experiments it appears that reproducible and
significant results concerning the size distributions can only be
Figure 5. Evolution of the droplet size distribution with stirring time (
= 2 obtained for ClBu volume fractions
up to 4 vol%, which is
wt%; [PVA-73-700] = 20 ppm/water; N = 700 rpm). a serious limitation of this technique. Nevertheless there is the
advantage that at low droplet concentration, nonlinear effects due
to multiple scattering of the ultrasonic wave become negligible.11
110
Moreover, as pointed out in Table 2, this technique also gives
100
Mean diameters (µm)

access to the volume fraction of dispersed phase Cvol that is in

90 perfect agreement with that defined in the formulation. Emulsions
80 with 2 wt% of ClBu will be examined in detail in the following.
70
60 Influence of stirring rate
50 The influence of the agitation speed N in the range 500–800 rpm,
40 and as a function of time, was examined with the standard
30 emulsion formulation of 2 wt%ClBu and 20 ppm/water of PVA-
0 10 20 30 40 50 60 73-700. The stirring range of 500–800 rpm was chosen for the
Stirring time (min) study because, below 500 rpm, the agitation speed is not efficient
D90 D50 D10 enough to disperse all the ClBu, and above 800 rpm a vortex with
air bubble incorporation is created. With these stirring conditions,
Figure 6. Evolution of the average diameters D10 , D50 and D90 with stirring
time (
= 2 wt%; [PVA-73-700] = 20 ppm/water; N = 700 rpm). the Reynolds number Re is in the range 14 000–25 000. These
values, being above 104 , are typical for locally isotropic turbulent
agitation conditions.4
themselves into an equilibrium conformation at the ClBu/water The decrease of the average droplet size D50 and the
interface and thus to protect the droplets from coalescence.4 The polydispersity index PI versus stirring time for various agitation
distribution curves give access to the average diameters, such speeds are shown in Figs 7 and 8. The evolution of the droplet
as D10 and D50 that are of general use in industrial practice and characteristics is typical for turbulent agitated systems as already
D90 that is for experimental reasons more precise and thus the reported in the literature.4 At first, after a rapid decrease of D50 and
most representative of Dmax . The evolution of these droplet size PI, a steady-state regime is reached within 50 min, for agitation
characteristics with stirring time is shown in Fig. 6. speeds of 700–800 rpm. Therefore, 700 rpm is defined as standard
In addition, from these values, the polydispersity index PI agitation speed.
= D90 − D10 becomes available as a function of time. In this According to Eqn (1), it is also verified that, in the steady-
particular case, shown in Fig. 6, it can be noticed that after an state conditions (stirring time = 60 min), D50 as well as D90 vary
initial rapid decrease, PI levels off to an almost constant value after with the agitation speed as D50 ≈ N−1.11 and D90 ≈ N−1.17 .
10–15 min of stirring time. These exponents of −1.11 and −1.17, respectively, are in very
At this point, it is worth mentioning that the reproducibility good agreement with the theoretical value of −1.2, which is
of the experiments is quite satisfactory. For 35 runs, carried characteristic of turbulent agitated systems controlled by the drop
out with the standard formulation (
= 2.27 vol%; [PVA-73- breakage process.
700] = 20 ppm/water) and agitation conditions (N = 700 rpm;
stirring time = 60 min), the following average size characteristics
Influence of PVA type and concentration
are obtained: D10 = 32.8 ± 0.7 µm; D50 = 49.4 ± 0.9 µm;
The influence of the PVA type and concentration on the ClBu
D90 = 69.9 ± 1.1 µm and Cvol = 2.21 ± 0.08 vol%.
droplet size was studied in the concentration range of 5 to
200 ppm/water, which corresponds to 0.025 to 1 wt%/ClBu.
Influence of the concentration of dispersed phase Figure 9 shows the decrease of the mean droplet diameter D50
As previously indicated, the maximum ultrasound attenuation with time for various PVAs at a concentration of 20 ppm/water.
(Amax ) measured by the OPUS granulometer is limited at 145 dB. It From Fig. 9, it appears clearly that the rate of droplet size reduction
was therefore necessary to check the maximum volume fraction of and the steady-state D50 are related to the DH and DPw values
dispersed phase that meets this requirement. These data are given of the different PVAs. PVA-73-700 is the most efficient surfactant,
in Table 2 for ClBu/water emulsions stabilized with 20 ppm/water for both the initial size reduction and the final droplet size, which
1213

of PVA-73-700. reaches a minimum value in this case. An interesting comparison is

Polym Int 2009; 58: 1209–1216

c 2009 Society of Chemical Industry www.interscience.wiley.com/journal/pi
 www.soci.org V Boscher et al.

130 120
120 110
Diameter D50 (µm)

110

Diameter D50 (µm)

100 100
90 90
80 80
70
70
60
50 60
40 50
30 40
0 10 20 30 40 50 60 0 20 40 60 80 100
Stirring time (min) PVA concentration (ppm/water)
500 rpm 600 rpm 700 rpm 800 rpm PVA-88-2500 PVA-80-2100 PVA-88-450
PVA-78-720 PVA-73-700
Figure 7. Mean droplet diameter D50 as a function of stirring time for
various agitation speeds (
= 2 wt%; [PVA-73-700] = 20 ppm/water). Figure 10. Mean droplet diameter D50 as a function of concentration for
the various PVAs (
= 2 wt%; stirring time = 60 min; N = 700 rpm).

100
90 70
PI = D90-D10 (µm)

80 65

PI = D90-D10 (µm)
70 60
60 55
50 50
40 45
30 40
20 35
0 10 20 30 40 50 60
30
Stirring time (min) 0 20 40 60 80 100
500 rpm 600 rpm 700 rpm 800 rpm PVA concentration (ppm/water)
PVA-88-2500 PVA-80-2100 PVA-88-450
Figure 8. Polydispersity index PI = D90 − D10 as a function of stirring time PVA-78-720 PVA-73-700
for various agitation speeds (
= 2 wt%; [PVA-73-700] = 20 ppm/water).
Figure 11. Polydispersity index PI = D90 − D10 as a function of con-
centration for the various PVAs (
= 2 wt%; stirring time = 60 min;
170 N = 700 rpm).

150
Diameter D50 (µm)

130 quantity of PVA, but rather by the kinetics of the emulsification

process. It may therefore be assumed that PVA of low DPw, having
110
a higher diffusion coefficient, migrates more readily to the droplet
90 surface and, as a consequence, promotes the stabilization of the
70 smaller droplets, formed in the agitation shear field.
The same conclusions can be drawn from Figs 10 and 11, which
50
show the evolution of steady-state D50 and PI, respectively,
30 versus the PVA concentration. From Figs 10 and 11, it can be
0 10 20 30 40 50 60 noticed that, for a given PVA, D50 and PI decrease at first
Stirring time (min) rapidly and eventually reach almost constant values for PVA
PVA-88-2500 PVA-80-2100 PVA-88-450
concentrations above 20 ppm/water. The emulsifying efficiency
PVA-78-720 PVA-73-700
ranking as determined by attenuation acoustic spectroscopy for
Figure 9. Mean droplet diameter D50 as a function of stirring time for the the different PVAs appears therefore to be as follows:
various PVAs (
= 2 wt%; [PVA-73-700] = 20 ppm/water; N = 700 rpm).
PVA-73-700 ≥ PVA-78-720 > PVA-80-2100
that of PVA-88-450 and PVA-88-2500, having the same DH values > PVA-88-450 > PVA-88-2500
but different DPw values. The PVA of low DPw is definitively
more efficient as an emulsifier than the PVA of higher DPw. The emulsifying efficiency corresponds to the smallest droplet size
Lankveld and Lyklema36 came to similar conclusions, by showing obtained at a given stabilizer concentration. This means that, in
that for paraffin/water emulsions stabilized with different PVAs this series, PVA-73-700 leads to the lowest droplet size. It clearly
of DH = 88 mol% and different DPw, the specific surface of the appears that the most efficient stabilizers are those of low DH,
emulsion droplets increases and thus the droplet size decreases whereas at constant DH a higher efficiency is reached with PVAs
with decreasing DPw of PVA. According to those authors, the of low DPw. The same tendencies have already been outlined by
1214

characteristics of emulsions are not determined by the adsorbed Zerfa and Brooks with laser diffraction measurements.4

www.interscience.wiley.com/journal/pi

c 2009 Society of Chemical Industry Polym Int 2009; 58: 1209–1216
Acoustic attenuation spectroscopy of stabilized emulsions www.soci.org

Correlation between interfacial tension and mean droplet size to around 0.40 with increasing values of DH. These exponents are
For oil/water emulsions prepared in the absence of surfactants systematically below the theoretical one of 0.6 given in Eqn (1).
or to some extent with low molecular weight surfactants, the In agreement with Walstra39 and Williams et al.,40 this is a first
equilibrium droplet size can be estimated by taking into account indication that one has to consider, on a purely phenomenological
the Laplace pressure P defined by the interfacial tension γ and basis, an ‘effective value’ of γ , which is systematically higher than
the drop diameter D:P = 4γ /D. However, as demonstrated by that determined by classical static or dynamic interfacial tension
Lucassen-Reynders and Kuijpers37 and Jones and Middelberg,38 measurements. In fact, as pointed out by different authors,37,40,41
this approach is not directly applicable to emulsification processes the emulsifier adsorbed on the droplet increases the emulsion
and to systems where the surfactant, in particular a ‘polymeric stability and the interfacial viscosity, which makes the break-up
surfactant’, develops a viscoelastic interface around the droplets. more difficult, and thus the drop size is larger than expected from
Another approach is given by Eqn (1), where Dmax , and thus D50 , equilibrium interfacial tension values. This effect can be noticed
varies as γ 0.6 . At this point, it is therefore of interest to check quite clearly in Fig. 12, where, at a given value of γ 400 , different
for PVA-stabilized emulsions the correlation between γ and D, D50 are obtained for the various PVAs, under constant agitation
keeping in mind that γ itself is directly correlated to the surfactant conditions. The comparison of the emulsions obtained with PVA-
concentration.
88-450 and PVA-88-2500 is illustrative in this respect. These two
Figure 12 shows a graphical representation of ln D50 versus ln
PVAs, of same DH and similar n exponent, differ only in their
γ 400 , according to Eqn (1) with γ 400 as the only variable. For a
DPw. At a given γ 400 (= 17.5 mN m−1 ), the droplet diameters D50
given PVA, Fig. 12 shows a straightforward correlation between ln
obtained with PVA-88-450 (at a concentration of 50 ppm/water)
γ and ln D50 over a wide concentration range, typically between
and PVA-88-2500 (at a concentration of 100 ppm/water) are 52.1
10 and 200 ppm/water, such as D50 ≈ γ n , with n the exponent
of γ for this relationship. The same type of correlation could be and 66.3 µm, respectively. This is evidence that the larger and
established between D90 and γ . A deviation from linearity appears more viscous interphase resulting from PVA-88-2500 limits the
only at very low PVA concentration, typically below 10 ppm/water break-up of the droplets and thus leads to larger droplets.
(not shown) and thus at γ values above 25 mN m−1 , where the Another comparison of interest can be made by considering
dispersions are quite unstable. The corresponding values of the the samples PVA-73-700 and PVA-88-450 at their crossover point,
exponents n for D50 and D90 as well as the correlation coefficients where the two samples lead practically to the same droplet size
R2 are given in Table 3 for the various PVAs. (D50 = 52 µm) and interfacial tension (γ 400 = 17.5 mN m−1 ), with
From Fig. 12 and Table 3, it is worth noticing that each PVA is the only difference being their respective concentrations. It turns
defined by a specific exponent n of γ , which increases from 0.12 out that in order to reach similar viscoelastic surface properties,
one needs to increase the concentration of PVA-88-450 by a factor
of 3.33 with respect to PVA-73-700. Due to the higher amount of
-9.2 ‘hydrophobic’ acetate groups of PVA-73-700, it can be assumed,
-9.3 as suggested by Lankveld and Lyklema,36 that this PVA is more
-9.4 tightly fixed to the interface and thus is in a more flattened
-9.5
ln D50 (m)

conformation than PVA-88-450, which would have a tendency to

-9.6 form larger loops. Loop formation with conformational changes,
-9.7 such as lateral loop packing of the chains, would lead to an
-9.8 increase of adsorbed PVA chains and thus to a thickening of the
-9.9 interface.
-10 The smallest droplet sizes and thus the most efficient dispersing
-10.1 properties are obtained with PVAs of low DH and low DPw,
-5.5 -5 -4.5 -4 -3.5
ln interfacial tension (N m ) -1 in particular PVA-73-700 and PVA-78-720. One can also notice
PVA-88-2500 PVA-80-2100 PVA-88-450 that there is only a slight decrease in droplet size for these
PVA-78-720 PVA-73-700 two PVAs with decreasing values of γ , which seems to be an
Figure 12. Correlation between the steady-state mean diameter D50 and indication that the surface coverage is almost completed even
the interfacial tension γ 400 . (For each curve and from right to left, each at low PVA concentrations. In contrast, especially for PVAs with
experimental point corresponds to the following PVA concentrations: 10, higher DH, such as PVA-80-2100, PVA-88-450 and PVA-88-2500,
15, 20, 50, 100, 150, 200 ppm/water). a more pronounced decrease in droplet size with decreasing
γ values can be observed, as a consequence of a larger loop
conformation.
Table 3. Exponents n and correlation coefficients R2 for D50 ≈ γ n The conclusion of this comparative study is that the dynamic
and for D90 ≈ γ n
interfacial tension γ has a definite influence on the droplet
D50 D90 diameters D50 and D90 . However, γ , by itself, is not sufficient
to predict the emulsifying efficiency of the different PVAs, even
n R2 n R2 if one takes into account the copolymer concentration that is
PVA-73-700 0.12 0.954 0.14 0.966 needed to generate that given value of γ . In fact, as shown by
PVA-78-720 0.18 0.976 0.20 0.968 Lankveld and Lyklema36 for paraffin emulsions stabilized by PVAs,
PVA-80-2100 0.28 0.977 0.28 0.980 it is not only γ and the amount of adsorbed PVA that determine the
PVA-88-450 0.44 0.996 0.34 0.976 droplet size under given agitation conditions, but more likely also
PVA-88-2500 0.38 0.892 0.48 (0.732) the thickness of the interfacial PVA layer, the interfacial gradient
1215

and its viscoelastic properties.

Polym Int 2009; 58: 1209–1216

c 2009 Society of Chemical Industry www.interscience.wiley.com/journal/pi
 www.soci.org V Boscher et al.

CONCLUSIONS REFERENCES
As a contribution to the problem of VCM suspension polymeriza- 1 Saeki Y and Emura T, Prog Polym Sci 27:2055–2131 (2002).
2 Endo K, Prog Polym Sci 27:2021–2054 (2002).
tion, a comparative study was carried out on the emulsification
3 Zerfa M and Brooks BW, Chem Eng Sci 51:3223–3233 (1996).
of ClBu, a model liquid for VCM, in an aqueous phase contain- 4 Zerfa M and Brooks BW, Chem Eng Sci 51:3591–3611 (1996).
ing PVAs of different DH and DPw, in the concentration range 5 Chatzi EG and Kiparissides C, Chem Eng Sci 49:5039–5052 (1994).
5–200 ppm/water. 6 Chung SI and Wasan DT, Colloids Surf 29:323–336 (1988).
The determination of the dynamic interfacial tension γ , 7 Alban FB, Sajjadi S and Yianneskis M, Trans IChemE 82:1054–1060
(2004).
using the pendent drop technique, showed, in agreement with 8 O’Rourke AM and MacLoughlin PF, Chem Eng Proc 44:885–894 (2005).
the literature, that the evolution of γ is a time-dependent 9 Dowding PJ, Goodwin JW and Vincent B, Colloids Surf A 192:5–13
phenomenon, involving at first a rapid diffusion of PVA to the (2001).
interface, followed by conformational changes of the adsorbed 10 Lovick J, Mouza AA, Paras SV, Lye GJ and Angeli P, J Chem Technol
Biotechnol 80:545–552 (2005).
surfactant layers. PVAs of lower DH and DPw appeared to be the
11 Dukhin AS and Goetz PJ, Adv Colloid Interface Sci 92:73–132 (2001).
most efficient in reducing the interfacial tension to a minimum, 12 Richter A, Voigt T and Ripperger S, J Colloid Interface Sci 315:482–492
even at low concentrations and within a short time interval. (2007).
For the droplet size distribution characterization of the 13 Challis RE, Povey MJW, Mather ML and Holmes AK, Rep Prog Phys
ClBu/water emulsions, it was demonstrated that on-line droplet 68:1541–1637 (2005).
14 Neumann AM and Kramer HJM, Part Part Syst Charact 19:17–27 (2002).
size monitoring using acoustic attenuation spectroscopy with the 15 Lazrak N, Le Bolay N and Ricard A, Eur Polym J 34:1637–1647 (1998).
OPUS device developed by Sympatec is only applicable for volume 16 Hinze JO, AIChE J 1:289–295 (1955).
fractions of dispersed phase below 4 vol%. Under these conditions, 17 Shinnar R, J Fluid Mech 10:259–275 (1961).
it has however the advantage to give access in real time to the 18 Jahanzad F, Sajjadi S and Brooks BW, Ind Eng Chem Res 44:4112–4119
(2005).
droplet characteristics and to the instantaneous volume fraction
19 Padovan D and Woods DR, AIChE Symp Ser 82:91–99 (1986).
of dispersed phase with a precision of about 2%. 20 Hong S, Albu R, Labbe C, Lasuye T, Stasik B and Riess G, Polym Int
From the systematic study, with the aim of preparing ClBu/water 55:1426–1434 (2006).
emulsions in the size range 20–100 µm, it was confirmed in 21 Van der Velden G and Beulen J, Macromolecules 15:1071–1075 (1982).
agreement with Hinze’s equation that an increase in agitation 22 Bugada DC and Rudin A, J Appl Polym Sci 30:4137–4140 (1985).
23 Netopilik M and Kratochvil P, Polymer 44:3431–3433 (2003).
speed leads to a significant reduction in drop size and a 24 Lin SY, Chen LJ, Xyu JW and Wang WJ, Langmuir 11:4159–4166 (1995).
narrowing of the size distribution. The correlation between the 25 Geers H and Witt W, Particulate Systems Analysis Conference 2003,
mean diameters and the stirring rate, with an exponent of Harrogate, UK (2003).
−1.2, was verified and is an indication that, for the present 26 Riebel U and Löffler F, Part Part Syst Charact 6:135–143 (1989).
experimental procedure, drop break-up is predominant over 27 He Y, Howes T and Litster JD, J Chem Eng Jpn 37:181–186 (2004).
28 Lankveld JMG and Lyklema J, J Colloid Interface Sci 41:454–465 (1972).
droplet coalescence. 29 Nilsson H, Silvegren C and Törnell B, J Vinyl Technol 7:112–118 (1985).
In agreement with and as a completion to previously reported 30 Rotureau E, Leonard M, Dellacherie E and Durand A, J Colloid Interface
results, it was shown that the characteristics of the PVA stabilizer, Sci 279:68–77 (2004).
such as the DH value, have a great influence on the droplet 31 Nahringbauer I, J Colloid Interface Sci 176:318–328 (1995).
32 Babak VG, Desbrières J and Tikhonov VE, Colloids Surf A 255:119–130
size. In this respect, copolymer PVA-73-700 appeared to be the (2005).
most efficient. It was further demonstrated that DPw also has a 33 Angle CW and Hamza HA, Chem Eng Sci 61:7325–7335 (2006).
non-negligible effect on the droplets size distribution. 34 Kraume M, Gäbler A and Schulze K, Chem Eng Technol 27:330–334
Finally, an original aspect of this study was the attempt to (2004).
correlate the droplet size characteristics and the interfacial tension 35 Chatzi EG, Boutris CJ and Kiparissides C, Ind Eng Chem Res 30:536–543
(1991).
γ , which in turn is dominated by the viscoelastic properties of the 36 Lankveld JMG and Lyklema J, J Colloid Interface Sci 41:475–483 (1972).
PVA adsorbed at ClBu/water interface. Further work is in progress 37 Lucassen-Reynders EH and Kuijpers KA, Colloids Surf 65:175–184
to examine these aspects. (1992).
38 Jones DB and Middelberg APJ, AIChE J 49:1533–1541 (2003).
39 Walstra P, Formation of emulsions, in Encyclopedia of Emulsion
Technology, vol. 1, ed. by Becher P. Marcel Dekker, New York,
ACKNOWLEDGEMENTS pp. 57–127 (1983).
The authors express their gratitude to Tessenderlo Group and the 40 Williams A, Janssen JJM and Prins A, Colloids Surf A 125:189–200
ANRT for financial support of this work (CIFRE/ANRT contract no. (1997).
41 Tadros TF and Vincent B, Emulsion stability, in Encyclopedia of Emulsion
295/2005). They also thank Drs D Van Deynse, C Colemonts and Technology, vol. 1, ed. by Becher P. Marcel Dekker, New York,
J Lionet for their interest and advice, Mr. C Behrens and G Mischel pp. 129–285 (1983).
(Sympatec) for their technical assistance, as well as the various
PVA suppliers.
1216

www.interscience.wiley.com/journal/pi

c 2009 Society of Chemical Industry Polym Int 2009; 58: 1209–1216