Professional Documents
Culture Documents
KONTOPbubbledissol
KONTOPbubbledissol
KONTOPbubbledissol
net/publication/229765474
CITATIONS READS
95 1,348
2 authors, including:
John Vlachopoulos
McMaster University
188 PUBLICATIONS 4,147 CITATIONS
SEE PROFILE
All content following this page was uploaded by John Vlachopoulos on 13 November 2018.
The presence of bubbles is inherent in the rotational molding process and can
often cause poor quality in rotomolded parts. In an effort to gain a fundamental un-
derstanding of bubble removal in polymer melts, a model that includes diffusion,
surface tension, and viscosity effects has been implemented. This work has been
complemented with experiments performed in a heated chamber, which enabled
the visualization of the bubble formation and removal process. It was concluded
that bubble removal is mainly dBusion controlled and is not influenced sic-
cantly by melt viscosity when the latter lies within ranges typical for rotational
molding resins. Air concentration in the bulk of the p o w e r melt and initial bubble
size were found to be of great importance. The application of pressure after the
polymer has melted accelerates the dissolution rate, owing to an increased driving
force for diffusion.
meric solutions of relatively low viscosities and at rela- Crawford and Scott (20).in a detailed study of the
tively high Reynolds numbers. In these cases, bubble phenomena occurring during the bubble formation
dynamics are mainly controlled by pressure differ- and removal, confirmed Kelly's suggestion that bub-
ences and fluid velocities. The bubble growth or col- bles remain virtually stationary in the melt, because
lapse mechanism is modeled by solving the coupled of the high viscosity of the melt. They described the
continuity and momentum equations, without taking melting and collapse of the powder as the combina-
diffusion into account. tion of two bulk movements: first the progression of
Bubble formation and removal is not a situation en- the melt front through the powder mass to the free
countered frequently in common polymer processing surface and then the collapsing of powder particles
methods, so there is very limited literature covering into the melt. Air is trapped during the latter process.
the aspect of bubble removal from polymer melts. A Spence and Crawford (23-25) presented an exten-
great deal of work has been done, though, on the for- sive study on the processing variables affecting the
mation of bubbles in polymer melts during the foam- formation and removal of bubbles. They studied the
ing process. Han and Yo0 (15) and Arefmanesh and effects of viscosity, powder characteristics, processing
Advani (16, 17) in their work on bubble growth in poly- conditions, moisture, mold material, and mold release
meric foams showed that the important phenomena agents. They also found that mold pressurization dur-
occurring during bubble growth are mass, momen- ing the rotomolding cycle is an efRcient method to re-
tum, and energy transfer between the bubble and the move bubbles from rotomolded parts.
fluid surrounding it. The transfer of mass in foam Crawford and his co-workers (24. 26) proposed a
processing is due to the diffusion of dissolved gas model for the removal of bubbles from rotomolded
from the fluid into the bubble. Ramesh et al. (18)veri- parts, which is based on the concept that the mecha-
fied Arefmanesh and Advani's theoretical predictions, nism for bubble removal is governed by the diffusion
by comparing them with experimental data. of gas through the polymer melt. The approach to-
Bubble# in R o t a t i d Molding wards the development of the W s i o n based model
follows the approach by Greene and Gaffney (4).
In an initial attempt to describe the bubble forma- Although these models seem to describe qualitatively
tion problem relevant to rotational molding, Rao and the removal of bubbles from polyethyelme melts, they
Throne (19) drew extensively from the theories of rely on the estimation of constants by curve fitting.
metal sintering and densfication. According to them
the formation of a homogeneous melt from powder THEORY
particles is a two-step process. During the first stage,
Dyaamiaa of Bubble M..drrtion
which is denoted as sintering,the powder particles
stick, or fuse, together at their points of contact. Although in reality the polymer melt during rota-
Interfaces and bridges develop between adjacent par- tional molding contains many bubbles, a simplified
ticles, until the mass becomes a porous three-dimen- approach will be implemented, which treats the disso-
sional network. During this stage there is little change lution of a single spherical gas bubble in an inAnite
in the density of the mass.The next step is the densi- polymer melt (Flg. 1). The problem is assumed to be
j b f b n ,where the network begins to collapse into the isothermal.
void spaces, which become bubbles. The bubbles are The kinematics of the problem are satisfied by the
either pushed ahead of the coalescing front to the free continuity equation, which for spherical geometry,
surface, if the melting is sufficiently slow, or become uni-directional flow in the r direction, constant den-
encapsulated and remain in the polymer after it is sity, p and using the boundary condition that the
quenched, which is the usual case for rotomolding. bubble wall velocity, at r = R is U,= dR/dt, is simpli-
Other researchers (20,211 distinguished three dis- fied to (27):
crete mechanisms: a sintering type of mechanism, R2R
bubble formation. and removal and surface leveling. u, = y-r
Progelhof et aL (21) observed two phenomena: forma-
tion of voids in the molded part and surface leveling. The dynamics of the system are governed by the con-
They reported a slow movement of the voids towards servation of momentum in the radial direction
the f k e surface. Kelly (22)suggested that bubbles re-
ap
main stationary and that the oxygen in the air bubble
begins to physically dissolve in the polymer, as the
p
(-+ ur-
ar
aur)
= - - +ar @..Ir
temperature of the melt increases. The oxygen is the
first gas to dissolve, because it has twice the solubility
of nitrogen in polyethylene. The depletion of oxygen
results in the reduction of the bubble diameter, which Since the fluid is highly viscous, the inertia terms can
in turn results in the increase of the pressure inside be neglected and Equation 2 becomes
the bubble, as dictated by the laws of surface tension.
The increase in pressure forces nitrogen to dissolve in ap + -
-- a (r2Tm) - Tee +
1 - =
(3)
the polymer, further reducing the bubble diameter. ar P ar r
iim POLYMER ENGINEERING AND SCIENCE, JULY 1990, Vol. 39, No. 7
Bubble Dissolution in Molten Polymers
Taking into account spherical symmetry ( T =~ T,J, an The components of the rate of deformation tensor for
integration from the bubble wall to infinity allows us uniaxial elongational flow are:
to relate the stresses within the fluid to the gas pres-
sure inside the bubble and the ambient pressure (14):
POLYMER ENGINEERING AND SCIENCE, JULY 1990, Vol. 39, No. 7 1191
M. Kontopoulou and J . Vlachopoulos
Dstermlnrtion of the Remsure Inaide the Bubhle Table 1. Numerical Values of Physical Properties at 1W'C.
In order to determine the rate of dissolution of the Physical Property Value Reference
~~
bubble from Eq 12,we must have a relation for the Diffusion coefficient, m2/s 8.57 X lWg (30,31)
gas pressure inside the bubble. This can be found by Henry's law constant, mol/(m3Pa) 3.75 x lW5 (30, 31)
applying the conservation of mass principle for gas in- Surface tension, N/m 0.026 (32)
side the bubble (28).If the buildup of adsorbed gas on
~~
The concentration of gas dissolved initially in the bulk Equations 12 and 14 are solved simultaneously,
of the polymer melt can be determined from Henry's to give the new values of Pg and R at the end of
law, if the melt is saturated with air: the time increment.
Using the most recent value of Pg,the new air
concentration, c,can be determined from J3q 15.
In reality it is impossible to know the degree of satura- By using this ,c and the new bubble radius, the
tion of the polymer melt, so when comparing model concentration gradient W a r can be determined
predictions to experimental data, it is necessary to as- for the second time increment Eq 16 and the pro-
sume that c, will be a fraction x of (
4:, cedure is repeated.
In order to implement the model the initial bubble
diameter, ambient pressure and physical constants of
1192 POLYMER ENGINEERING AND SCIENCE, JULY 1990, Vol. 39, No. 7
Bubble Dissolution in Molten Polymers
0.30 1.036
--Bubble Diameter
0.25
1.031
-5
E
E 0.20 -I
0)
P
-f 0.15 1.026 $
0 X
0
nm
2 0.10
m' 1.021
0.05
1.016
0 2 4 6 8 10 12 0 2 4 6 8 10 12
Time (min) Time (min)
Flg. 2. Diameter and air pressure inside the bubble versus Flg. 4. Relative magnitude of surface tension and n o d
time for the dissolution of a typical gas bubble [d,= 0.25 stress terms as afunction of time in a PE melt [d,= 0.25 nun.
nunJ in a PE melt [k = 3000 Pas,c, = 0.995 at 190°C. fi = 3000 Pas, 19O"C,c, = 0.995 [c,),).
0.30
__ p= loo00 P a s
RESULTS-MODEL PREDICTIONS
0.25 - p = so00 Pas
......... = 500 Pas Bubble Diudution Mechadmm
-EE
L
0.20
The application of the proposed model gives an in-
E
5 0.15
s e t to the mechanism of bubble dissolution. As the
.-
0 bubble forms, the air pressure, Pg.inside the bubble
;
a
0.10 is slightly kugher than the ambient pressure, because
of the action of surface tension (Equation 6). Through
m
0.05 the application of Henry's law it can be seen that
higher air pressure inside the bubble results in a
0.00 1 higher concentration of air, c,, in the polymer/air in-
0 2 4 6 8 10 12 terface (bubble walls) compared with the concentra-
Time (min)
tion ciin the rest of the polymer melt, which is ex-
Flg. 3. Effectof melt viscosity on bubble dissolution curues in posed to ambient pressure (see also Rg. I).
a PE melt [d,= 0.25 mm, 190°C.c, = 0.995 [ c , ) ~ ) .
The concentration difference gives rise to a dnving
force, which results in the diffusion of air from the bub-
ble to the surroundings. As gas is removed from the
bubble, the diameter decreases and the air pressure in-
side the bubble inu-eases further. Flgwe 2 shows the
change in Pg as the bubble diameter decreases, for the
the system, such as viscosity, surface tension, diffusion dissolution of a typical gas bubble of initial diameter
coefficient, and Henry's law constant are required. The equal to 0.25mm in a PE melt, which is close to satu-
concentration of air in the polymer melt is determined rated conditions (q = 0.995( ~ 3 , ~ )and has a viscosity
by curve fitting, since it is impossible to determine of 3000 Pa% at 190°C.At long times, Pg increases
either theoretically or experimentally, as explained in dramatically as R --f 0 (Equation6), and this causes
the Theory section (Equations17aand 1 7b). an acceleration in the bubble dissolution rate. This
happens because the increase in Pg inside the bubble
Emtimation of Physical Conmtants results to an increase in q,,, as predicted by Henry's
law and subsequently to an even larger concentration
The diffusion coefficients and Henry's law constants difference between the bubble/& interface and the
for air into polyethylene melt were estimated by us- polymer melt, thus acceleratmg diffusion.
ing data and charts published by Griskey (30).The
Lennard-Jones parameters for air, needed for using Effect of Vi.cOrity
Griskey's charts, were obtained from Bird et aL (29).
The temperature dependence of these parameters can It has been reported that melt viscosity affects bub-
be estimated by applying an Arrhenius type equation ble dissolution rate, because as melt viscosity de-
(31).Surface tension values for PE melts were ob- creases, polymer chains become shorter and diffusion
tained from Wu (32). of air through the melt is facilitated (23).However, it
A summary of the physical constants used in this has been shown that both in the case of the diffusion
work is shown in Table 1. of gases in solid polymers (33)and in molten polymers
POLYMER ENGINEERING AND SCIENCE, JULY 1999, Vol. 39, No. 7 1193
M. Kontopdou and J. Vlachopoulos
0.30
-0.S98 Caat rn o Experimental [24]
- Simulation
0.00 0.15 9
0 5 10 15 20 0 10 20 30 40 50
Time (min) Time (min)
m.5. mxt of bulk concentrationof air on bubble dissolu- m. 7. Comparison of o d e 1 predictions with
tion curves (d,= 0.25 nun,+ = 3000P a s , 190°C). datafrom Spence (24).
(30)diffusMty does not depend on the viscosity of the term is around 150 times larger than the normal
polymer but rather on the size of the gas molecule stress term (which depends on viscosity). This means
and on the monomer type and structure. that at the early stages the magnitude of the normal
The effect of viscosity in the bubble dissolution stresses is insignificant and does not affect the result.
process can be understood better by inspecting the At much longer times. as the dissolution of bubble
force balance at the bubble-liquid interface (Equatdon proceeds, the surface tension term becomes only
6).As the bubble shrinks. a normal force is generated around three to four times larger, meaning that now
in the polymer melt, which acts on the opposite direc- the normal stress term (and thus the viscosity contri-
tion, that is, it opposes the movement of the bubble bution) becomes significant. As the bubble diameter
boundary. For a Newtonian liquid. the normal stress tends towards zero. both terms increase abruptly.
increases when viscosity increases (EQuatlon 8). Thus
for higher viscosity values, the normal force opposing E!feat of BuIk Coaaemtration o# Air ln the
the movement of the bubble boundary is larger. Polymer Melt
Bubble dissolution rates, for viscosity values rang-
ing from 500 to 10,OOO Pa-s at 190°C. are shown in Another very important parameter is the bulk con-
Ng. 3.This range of viscosities was chosen because it centration of air in the polymer melt. Polyolefin melts,
is typical for PE rotomolding grade resins. From Rg. 3 in equilibrium with the atmospheric air, will always
it can be seen that the effect of viscosity is marginal contain dissolved air. The solubility of air depends on
and is more prominent at longer times, when bubble pressure and the amount of dissolved air when the
diameter becomes small. This can be understood by polymer melt is saturated can be determined from
examining the relative magnitude of contribution of Henry’s law (30, 31). The exact amount of air con-
the terms in Equation 6. The relative magnitudes of tained in the polymer melt in typical rotomolding con-
the surface tension and normal stress terms for the ditions cannot be determined exactly, but our results
typical bubble dissolution case shown in Q. 2, are show that the melt is close to the saturation condi-
plotted in Ng. 4. In the beginning the surface tension tions. However, even when the melt is slightly below
the saturation point, the dissolution rate is signifi-
cantly accelerated, because of an increased concen-
tration gradient: this is demonstrated in Ng.5.
1194 POLYMER ENGINEERING AND SCIENCE, JULY I-, Yo/. 39, No. 7
Bubble &solution in Molten Polymers
Material MFR (@I0 min) Zero Shear Viscosity (Pa-8) Density (kg/m3) Melting Point ("C)
0.30 0.30
- model predktions Bubble interactions 1
\ C, = 0.974 (CJ.,
0.05 1
"."" r -
0 2 4 6 8 10
0 0.5 1 1.5 2 2.5 3
Time (min) Time (min)
R g . 1 1 . llJpical bubble dissolution curves for D P E with Rg. 13. Regbns obserued during the dissolution of a typical
MFI=5at190°C. air bubble in LLDPEfir high bubble concentmtions.
would become saturated with air upon melting, so ness of the powder bed inside the cup was around
there would be no increased gradient to facilitate the 5 mm. The process of powder melting and air entrap-
bubble dissolution. ment inside the melt was observed through an Olym-
pus optical microscope and recorded with a Pana-
EXPERXHENTAL sonic video camera. Several bubbles were observed as
they subsequently dissolved and image analysis was
Bubble dissolution experiments were conducted by performed to determine the dissolution rate. All the
using a heated chamber, which consisted of a cylin- experiments were done isothermally at 190"C,under
drical oven with a diameter of 16 cm and a height of atmospheric pressure and the duration of the experi-
11 cm.A sample cup holder, made of glass (D = 2.5 ments was 30 min. Table 2 shows a listing of the
cm. H = 1 cm) was loaded with a predetermined resins used and their properties.
amount of powder and placed in the center of the Thermo-Gravimetric analysis OGA) was performed
heating chamber, which was kept at a constant tem- in the polymer samples using a Netzsch STA-409
perature with the aid of a temperature controller. The Simultaneous Thermal Analyzer, at 190°C under at-
temperature of the polymer melt was constantly moni- mospheric air. The purpose of this test was to exam-
tored,through a K-typethermocouple. The initial thick- ine the possibility of degradation, existence of residual
0.30
0 low bubble content
0.25
l o A A A A A MFh5
.-O--LLDPE
0.20 -.
0
A
0.15 -~
0
A
0.10 --
0.05 ~-
1196 POLYMER ENGINEERING AND SCIEWCE, JULY I-, Vol. 39, No. 7
Bubble Dissolufion in Molten Polymers
monomer vapors in the polymers or gas generation This result seems to be somewhat contradictory to
when the polymer is exposed to conditions similar to the general belief that increased viscosity results in
the experimental conditions. These tests showed that slower bubble dissolution. It has been observed in our
there was no degradation or generation of vapors of experiments, though, that when the melt viscosity is
any kind for the time period applicable to the bubble higher, more and larger bubbles form initially upon
removal experiments. melting and densification of the powder. This may be
the cause for the longer times needed for complete
Rem~ltaand Cornparimon With Model predictlop. dissolution of bubbles observed in practice.
Initial bubble size is very important. Small bubbles
From the experimental results it was seen that there
tend to dissolve quickly, while big ones may not dis-
is a great variability in the behavior of the bubbles, de-
solve at all during the experiment. This effect can be
pending on their size, their proximity to other bubbles
clearly seen in Rg. 11, and it is consistent with model
and the overall bubble content in the polymer melt.
predictions.
Some typical dissolution curves for an LLDPE with
MFI = 5, along with the model predictions, are shown
CONCLUSIONS
in Rg. 11. In Rg. 11, time zero is considered as the
time when the bubbles have formed and have as- A model that includes diffusion, surface tension,
sumed a spherical shape. The bubbles shown have and viscosity effects has been applied, in an effort to
been carefully chosen so that they are relatively iso- describe bubble dissolution in polymer melts. It was
lated inside the polymer melt, to eliminate' any inter- concluded that the process is controlled by the W-
action effects. The model predictions, obtained by sion of air from the bubbles to the melt. Bubble disso-
using ci = O.95(cJMt.are in reasonably good agree- lution depends sgnifhntly on initial bubble size,sur-
ment with the experimental data. face tension, and air concentration in the polymer
Bubble dissolution is considerably slower in regions melt, but it is not influenced considerably by melt vis-
of high bubble content. Figure 12 compares the be- cosity when the latter lies within ranges typical for ro-
havior of one of the isolated bubbles shown in Rg. 1I, tational molding resins.
with that of a bubble in a region of high bubble con- The application of pressure after the polymer has
tent. It appears that in the latter case, after the bubble melted leads to a sigdicant acceleration of the disso-
forms, there is an induction period where there is very lution rate, owing to an increased driving force for dif-
little dissolution, followed by a period during which fusion.
considerable dissolution takes place. The possibility Bubble removal experiments confirmed model pre-
that this behavior might be due to degradation oc- dictions for relatively isolated bubbles. In the case of
curring during the course of the experiments was high initial bubble contents in the polymer melt, dis-
excluded by performing TGA analysis in the polymers. solution was delayed considerably and two distinct re-
The existence of two periods is illustrated in Rg. 13, gions were observed: an initial period, where dissolu-
for the same bubble shown in Rg. 12. It is speculated tion is very slow as a result of interactions between
that the initial period is associated with interaction bubbles, followed by a fast dissolution period, associ-
between neighboring bubbles, which can result to the ated with the disappearance of surrounding bubbles.
transfer of air from one bubble to another. Owing to These results reveal that one of the most important
the presence of many bubbles diffusing simultane- factors governing the bubble dissolution process is
ously, the melt becomes almost saturated, thus reduc- the bubble content initially present in the polymer
ing the concentration difference and the drivhg force melt.
for dissolution. From an inspection of the dissolution
process of all the bubbles, it was seen that the change ACKNOWLEDGMENTS
in slope observed in F@. 13 was associated with the
This work has been supported financially by the
disappearance of the smaller surroundmg bubbles, so
N a t u d Sciences and Engineering Research Council
that the bubble under consideration remained even-
of Canada (NSERC), the Ontario Government, and
tually relatively isolated. I n this case, the model can
Imperial Oil Canada. The authors would also like to
be applied successfully only for the second region,
thank Imperial Oil Canada (Exxon Chemical Canada)
which coincides with the disappearance of the sur-
and Mr. Blair Graham for their help and for supplying
rounding bubbles. A value q = 0.974 (c,),~ is used in
the resins used in this work and Ms. E. T&cs for
this case.
performing the resin characterization.
From the above it becomes clear that the higher the
overall bubble content in the melt, the slower the bub-
NOHENCLATURE
ble dissolution.
Polymer viscosity does not seem to have a s i w c a n t Concentration (moi/m3)
effect on the rate of bubble removal. This confirms the Concentration of gas inside bubble (m01/m3)
model predictions and was observed by using three Concentration of gas in bulk of polymer
LLDPE resins with Merent MFIs (Table2). Some typi- melt (moi/m3)
cal dissolution curves of relalively isolated bubbles for Saturation concentration of gas in bulk of
the three LLDPE resins are shown in Rg. 14. polymer melt (m0l/m3)
1198 POLYMER ENGINEERING AND SCIENCE, JULY 1990, Vol. 39, No. 7