Polymer Plasticization and Devolitilization Studied by Inverse Gas Chromatography

1
Polymer Plasticization and Devolitilization
Studied by Inverse Gas Chromatography
By Adam T. Jones and Marcos Perez-Blanco
Introduction
Similar plots were generated for methyl
acetate and toluene. For methyl acetate all of
the partition coefficient data falls along a
trendline, however there is more deviation from
the trend in this case than with methanol. In
this case as well, the reference datasets are
very similar to the experimental data. Although
linear behavior can be observed for diffusivity
as well, deviation from the trend can be observed
also. The experimental data is slightly
different from the literature values. In
the case of toluene the partition coefficient
data falls closely along a trendline, and there
is relatively little deviation. Here the
reference datasets are similar to the
experimental data, but there are notable
differences at some temperatures. Linear behavior
can be observed for diffusivity as well, but
deviation from the trend can also be observed.
Some of the literature data is significantly
different from the experimental data, however
overall the results are consistent with the
literature. Finally, the figure below shows
the effect of column pressure on the values
calculated for the partition coefficient and
diffusivity. The carrier gas pressure has the
effect of slightly reducing the value measured
for diffusivity. It has a negligible effect on
the partition coefficient. This could be the
result of some interaction of the carrier gas
with the polymer. However since the carrier gas,
helium, is inert, it is highly improbable that it
interacts with the polymer. Another explanation
could be that the effect of pressure actually
depends on the diffusivity of the sample in the
carrier gas. The Taylor dispersion of the sample
in the carrier gas stream 7 depends on the
velocity profile of the carrier gas and the gas
diffusivity of the sample. To generate a fit to
the narrower peak, the regression method produces
a lower value for the diffusivity of the solvent
in the polymer. The decrease in diffusivity is
not actually a physical change in the polymer,
but is only an inaccuracy in the calculation of
the value which is the result of interactions in
the gas phase which are not accounted for in the
regression model.
The solubility and diffusivity of solvents
and small gas molecules in polymers plays a
critical role in a number of significant
commercial processes. For example, a time
release drug delivery system has been designed
that involves injection of biodegradable polymer
particles containing the drug. All traces of the
solvent (which is toxic) must be removed from
these particles prior to use. A possible process
is to swell the particles (plasticize the
polymer) with an inert gas and increase the rate
of diffusion of the solvent. The objective
of this research is to develop a better
understanding of the solubility and diffusivity
of solvents and plasticizing gasses in different
polymers, using Capillary Column Inverse Gas
Chromatography (CCIGC), which has been shown by
previous research to be an accurate and
relatively efficient method. The inside of the
chromatography column is coated with the polymer,
and the plasticizing gas is run through the
column, as with traditional chromatography.
Since it is the polymer (stationary phase) which
is of interest instead of the fluid phase, it is
given the term inverse gas chromatography.
The mathematical model of the behavior of the
solvent in the capillary column solves the
differential equations for the diffusion of
solvent in a capillary column in the Laplace
Domain, giving

A computer program using a Fourier domain
fitting procedure was used to generate K and Dp
values from experimental peak data, as well as a
theoretical peak created using those parameters.
By comparing the theoretical fit peak to the
actual data, the accuracy of the fit parameters
can be evaluated. The value of Dg (the gas
diffusivity) needs to be known. This was
calculated for each sample IGC run as a function
of the column pressure, the oven temperature, the
molecular masses of the solvent and carrier, and
the functional group contributions of the
solvent and carrier gas molecules, using the
AIChE DIPPR Data Prediction Manual Program.
To reproduce values that are given in the
literature 2,5,6, experiments were run using
three different solvents, methanol, methyl
acetate, and toluene, on the polyvinyl acetate
polymer GC column. This comparison to literature
values verifies that the CCIGC machine is
functioning properly and provides a measure of
the accuracy of the results. Most of the
experiments were run with a column pressure
between 230 and 260 PSIG. Some experiments were
also done for methanol at higher pressures of
400, 600, 800, and 1000 PSIG.
Pressure effect for MeOH
Where C is the outlet concentration of solvent,
Co is the concentration at injection, u is the
carrier gas velocity, L and r are the length and
radius of the column, tau is the thickness of the
polymer film, Dg is the diffusivity of the
solvent in the carrier gas, and S is the Laplace
variable. K is the solvent partition coefficient
(solubility in the polymer), and Dp is the
diffusivity of the solvent in the polymer. 1
A solvent is injected into a stream of the
plasticizing gas, and the peak produced as the
solvent emerges from the column is analyzed using
computer regression procedures to determine the
diffusivity and solubility of the solvent in the
polymer. literature values were reproduced
for systems that have already been studied, to
make sure the machine was functioning properly,
and to refine the experimental procedures. The
gas flowrate in the column needs be set within a
certain range to produce useful results. If it
is too high the sample doesnt have time to
interact with the polymer and the data is
useless. If the flowrate is set too low the data
will be fine but it has to be collected over a
long period of time (several hours or more)
Results Discussion
Some examples of the solvent peaks are shown
below, on the same graph with the regressions.
Corresponding K and Dp values are also shown.
Note the sharp peak for the methyl acetate and
the comparatively symmetric peak for methanol.
Several factors have been observed to
influence the accuracy of the IGC setup and the
regression procedure. When a solvent peak with a
sharp front and long tail is generated, the
regression results tend to be inaccurate, even
though the fit appears to match the data.
According to the mathematical model of the
capillary column, this happens if the parameter
value (beta2) is too high, which is caused when
the carrier flow rate is too high compared to the
diffusivity of the solvent in the polymer. Note
that the diffusivity of methyl acetate in poly
vinyl acetate is particularly low compared to the
other two solvents. This could explain why
methyl acetate had a tendency to give these kinds
of peaks. High flowrates should be avoided when
using polymer solvent systems that have low
diffusivities of the solvent in the polymer.
The tailing may have been exacerbated by
injector malfunction. For some of the
experiments it is suspected that a clog caused it
to bleed the sample into the carrier gas stream
instead of introducing it as a single pulse.
Another factor that could have an effect on the
reliability of the data produced by the IGC is
the oven temperature. Since diffusivity
increases with temperature 2, at low
temperatures the peaks will have longer tails.
The effect of column pressure is at most
trivial for the partition coefficient. There is
however a slight downward trend of diffusivity
with pressure.
Experimental
Diagram of the CCIGC. Emergence of sample from
column recorded with TCD. Pressure monitored
with Transducer at head of column. Pressure is
controlled with a pressure regulator, flowrate is
controlled with backpressure control valve (BPCV).
Conclusions Future work
Capillary Column Inverse Gas
Chromatography has been shown to be an accurate
and efficient method of determining the partition
coefficient and diffusivity of polymer solvent
systems such as methanol, methyl acetate, and
toluene on poly vinyl acetate. Literature values
from 3 different sources are in agreement, and
they have been reproduced experimentally with the
CCIGC apparatus. This work gives experimental
verification for our IGC setup, and it can be
considered a reliable instrument to be used in
further research. Some problems were
noticed during the research. One was that the
diffusivity of some systems was too low compared
to the carrier flowrate. Another was that
injector seemed to be malfunctioning with the
result being an artificially generated tail in
the solvent peaks. Either way the results were
inaccurate when these problems occurred. The
solution to the low diffusivity is to reduce the
carrier flowrate, or to raise the temperature to
increase the diffusivity. A new injector
assembly will have to be acquired to replace the
malfunctioning one. Further research
will consist of beginning research with non inert
gasses such as carbon dioxide and ethylene,
instead of helium. The effect of the
plasticization on the partition and diffusivity
coefficients at different pressures and
temperatures will be studied. Over the
long term, some new mathematical models for
determining polymer-solvent interaction
properties may be developed. For example, a new
model would be needed for using supercritical
carbon dioxide as the carrier, or one could be
developed to account for Taylor dispersion of the
sample in the carrier gas.7
A plot of the log of the partition
coefficient versus inverse temperature, for
methanol, is shown below to the left. Clearly
all of the partition coefficient data falls
closely along a trendline. Some literature
values are shown for comparison. For methanol,
the trendline for the experimental data is almost
identical to the reference datasets. The plot for
methanol of log of diffusivity versus inverse
temperature is shown below to the right.
Although linear behavior can be observed here as
well, the experimental data clearly seems to fall
along two separate trends, one above and with a
greater slope than the other. Most of the
literature values for diffusivity of methanol are
closer to the lower trendline, and there is a
good correlation between the experimental and
literature data. However, notice that even if
only the lower set of experimental values and the
literature values are considered, there is a
noticeable difference between the different
literature datasets as well as the experimental
data. One explanation for the two distinct
trends in the methanol diffusivity data is that
the tailing of some of the peaks causes the
diffusivity value calculated by the regression to
be lower.
Acknowledgements
A sample of solvent is injected into a
stream of the carrier gas. Because the
interactions are being studied at high pressures,
the IGC must be operated at high pressure, up to
1000 PSIG. A traditional GC Injection port
cannot be used because it would be destroyed by
these pressures. Instead a mechanical
sample injector must be used to introduce the
sample into the carrier gas. The mathematical
model uses the assumption that the solvent is at
infinite dilution. A sample loop with a very
small volume (0.5 L) must be used for these IGC
experiments so that the assumption of infinite
dilution of solvent in the column is valid.
Diagram of the Injector is shown in the following
figure.
Thanks to Dr. J. L. Duda and Dr. R. P. Danner,
Adam T. Jones, and all the other members of the
Centre for the Study of Polymer Solvent Systems.
References
R. P. Danner, F. Tihminlioglu, R. K. Surana, J.
L. Duda, Inverse Gas Chromatography Applications
in Polymer-Solvent Systems, Fluid Phase
Equilibria 148 (1998) 171-188 Arnould, D.D.,
Capillary Column Inverse Gas Chromatography for
the Study of Diffusion in Polymer-Solvent
Systems, Ph.D. Thesis, University of
Massachusetts, Amherst (1989). A. Jones, High
Pressure Inverse Gas Chromatography, Center for
the Study of Polymer Solvent Systems,
2005 http//www.rheodyne.com/pdfs/product_bulletin
_111.pdf Zielinski, J.M. Fry, R. Kimak, M.F.,
Probing Multicomponent Thermodynamic Effects by
Low- and High-Pressure Capillary Column Inverse
Gas Chromatography, Macromolecules, 37,10134
(2004). Surana, R.K., Advances in Diffusion and
Partition Measurements in Polymer-Solvent Systems
using Inverse Gas Chromatography, Ph.D. Thesis,
The Pennsylvania State University (1997). E.
Hamdan, J.F. Milthorpe, J.C.S. Lai, New approach
for evaluation of capillary column inverse gas
chromatography, Journal of Chromatography A, 1078
(2005) 144151.