Application of Compressed Carbon Dioxide in the

Incorporation of Additives into Polymers

ALAN R. BERENS,* GARY S. HUVARD,’ R I C H A R D W. KORSMEYER,* and F. W. K U N l G

The BFGoodrich Company, Research & Development Center, Brecksville, Ohio 441 41

SYNOPSIS

It has been found that carbon dioxide remarkably accelerates the absorption of many low
molecular weight additives into a number of glassy polymers. This effect is due to the high
diffusivity,solubility, and plasticizing action of compressedCOPin polymers. The transport
of CO, and the effects of CO, pressure on the transport of other low molecular weight
compounds in polymers have been studied by a simple gravimetric method Polymer film
samples were contacted in a pressure vessel with compressed CO,, or with COPplus various
organic liquids or solids, and the sample weight was followed with a fast-responseelectronic
balance during subsequent desorption at atmospheric pressure. Upon release of the pressure,
absorbed CO, rapidly diffuses from the polymer, while the other compounds desorb much
more slowly. The amount of additive absorbed can be determined from the plateau weight
of the sample after most of the CO, has escaped. Extensive kinetic and equilibrium data
are reported for the model system poly (vinyl chloride)/dimethyl phthalate/CO,, and a
number of other examples of C0,-assisted additive absorption are given. This “infusion”
process, in effect, amounts to the partitioning of the additive between the COz- and polymer-
rich phases; consequently, the relative solubility of the additive in CO, and in the polymer
is a major factor governing the amount of additive absorbed. Data reported here illustrate
the generality and potentially broad applicability of C0,-assisted polymer impregnation.

INTRODUCTION common glassy polymers from the 80-100°C range

t o below room temperature. Although these attri-
Recent studies in this laboratory’ and elsewhere butes of C02/polymer systems are typical of organic
have demonstrated t h a t carbon dioxide under high liquid/polymer systems, COz has the high diffusivity
pressure shows a n unusual combination of effects in polymers t h a t is characteristic of gases of similar
in its interactions with high polymers: At temper- molecular size.
atures and pressures approaching its critical con- Other recent work^^*'^ have reviewed the effects
ditions, T,= 31°C a n d Pc = 73 atm, the solubility of the molecular size of the diffusing species and the
of C 0 2in many polymers is as high a s t h a t of typical state of the polymer on the kinetics of transport of
organic liquid swelling agents, ranging from around small molecules in polymers. To illustrate the major
10 t o more than 30%by weight. Carbon dioxide also trends, diffusivities, D, of a number of gases, vapors,
has a strong plasticizing effect; concentrations of 8- and liquids in both glassy (unplasticized) and rub-
10 wt % depress the glass transition temperature of bery (plasticized) poly (vinyl chloride) are plotted
vs. molecular diameter in Figure 1.” In the glassy
state, D is a n extremely strong function of the mo-
~ ~~

lecular size of the penetrant, decreasing by roughly

Presented at the AIChE Annual Meeting, Washington, DC,
November 28-December 2,1988. 10 orders of magnitude from the small fixed gases,
* Present address: RD#2, Box 3510, Middlebury, V T 05753. He and Hz, t o solvent molecules such a s n-C6H14
To whom correspondence should be addressed at Huvard and C C 4 . In the rubbery state, the dependence of
Research and Consulting, Inc., Richmond, VA 23231.
*
Present address: Pfizer Central Research, Groton, CT 06340. D on molecular size is more moderate. Consequently,
Journal of Applied Polymer Science, Vol. 46, 231-242 (1992)
the difference in D between the glassy and rubbery
0 1992 John Wiley & Sons, Inc. CCC 0021-8995/92/0Z0231-12$04.00 states increases sharply with increasing size of the
231
232 BERENS ET AL.

-4

-6

0 0
-8
0

N
.
0,
v)
-10

5
p^ -12
m
-
0
At 0 Plasticized
-14 4% A Undasticized

A
-16
A
A
-18
2 4 6 8 10
4 **- * *
Gases Solvents Plasticizers, etc.

Molecule Diameter, A
Figure 1 Diffusivities of penetrants in rigid and plasticized PVC vs. molecular diameter.

diffusing species. Thus, for the small gases, D may of the process are also illustrated and explanations
increase by about one order of magnitude upon plas- are suggested.
ticization, whereas for common solvents, the in- In the course of this work, a patent on a very
crease may be by a factor of lo6 or loR, and for similar impregnation process was published.‘* The
plasticizers and other polymer additives, it may be present paper shows that the acceleration of ab-
still greater. sorption is much more pronounced in glassy poly-
Consideration of the trends illustrated in Figure mers than in the rubbery and semicrystalline poly-
1, in combination with the high solubility, diffusivity, olefins and olefin copolymers cited in this patent’s
and plasticizing action of C O z ,led us to suggest that examples.
compressed COZ could be useful as a temporary
plasticizer to facilitate the absorption of additives
into glassy polymers.” The concept was to expose EXPERIMENTAL
a polymer sample simultaneously to an additive
substance and to COz under high pressure; the COP The gravimetric procedure used here involves sorp-
would rapidly diffuse into the polymer, plasticizing tion of COP,or COBplus an additive, into polymer
it and accelerating the absorption of the additive. film or sheet samples in a simple pressure vessel,
When the COzpressure was released, the gas would followed by rapid venting and transfer of the samples
rapidly diffuse out of the polymer, deplasticizing it to a balance for recording weight changes during
and sharply reducing the additive diffusivity; the desorption at atmospheric pressure. All experiments
additive would then be left in the polymer to desorb were carried out at 25”C, and pressures ranged up
a t the slow rate governed by its diffusivity in the to the saturated vapor pressure of C02: 6.53 MPa
COz-freepolymer. We have applied the terms “COP- (950 psia) . Samples included a wide range of poly-
assisted impregnation” or simply “infusion” for mer types, which were compression-molded to
brevity in referring to this process. thicknesses from 0.1 to 1 mm.
In this paper, we describe a simple gravimetric
procedure that provides transport kinetics and sol-
Polymer/CO, Binary Systems
ubility equilibrium data for both polymer /COz bi-
nary systems and polymer / COP/additive ternary The procedure for COz sorption experiments is il-
systems, and we present results demonstrating the lustrated schematically in Figure 2 . A preweighed
efficacy and scope of the C02-assisted impregnation polymer sample (20-200 mg) was placed in a 100
process. Some of the complexities and limitations mL pressure vessel fitted with a pressure gauge,
 APPLICATION OF COMPRESSED COP INTO POLYMERS 233

Time -
Figure 2 Schematic curves for pressure and sample weight vs. time during sorption/
desorption experiments on polymer/C02binary systems: 0 load film sample and evacuate;
@ apply C 0 2 pressure; @ polymer absorbs COz; @ rapidly release pressure; @ transfer
sample to balance; 8 record weight during desorption.

valve, and a screw closure that could be opened obtained only during the desorption process at at-
quickly. The vessel was evacuated, then filled to the mospheric pressure.
desired sorption pressure from a cylinder of liquid
CO, and left at this pressure for an appropriate
sorption period. Experiments were run a t various Polymer/CO,/ Additive Ternary Systems
pressures, sorption times, and sample thicknesses
for each polymer studied. At the end of the sorption Experiments on ternary systems closely paralleled
period, the C 0 2 pressure was rapidly vented to at- those on the binaries with only these changes in
mospheric, the vessel opened, and the sample quickly procedure: The liquid or solid additive was charged
placed on the pan of a fast-response electronic digital to the pressure vessel, along with the polymer sam-
balance readable to 0.00001 g (Mettler AE163). Us- ple, before closing, evacuating, and pressurizing with
ing a computerized data acquisition system, sample C02, and the polymer and additive were in direct
weights during desorption were recorded a t intervals contact to minimize any mass-transfer barrier. The
as short as 5 s beginning within 10-20 s after venting amount of the additive, in most cases, was in excess
the vessel. of its expected uptake by the polymer. When the
Plots of sample weight vs. the square root of de- vessel was opened, after the period of absorption
sorption time were initially linear, in accord with under CO, pressure, the unabsorbed excess additive
Fickian diffusion kinetics. The diffusivity of C 0 2was was wiped or blotted from the surface of the polymer
readily estimated from the slope of these plots and sample. This step added only a few seconds to the
the initial sample thickness. Linear extrapolation time between opening the vessel and placing the
to zero desorption time gave the weight of CO, in sample on the balance to record weight during de-
the sample at the end of the sorption period. Re- sorption.
peated experiments at a series of sorption times then The sorption/desorption sequence for the ternary
defined the absorption kinetics, and the attainment systems is shown schematically in Figure 3. While
of sorption equilibrium was evidenced by reaching under CO, pressure, the sample absorbs both COz
a constant level of C 0 2 uptake. Experiments at var- and, presumably, some of the additive, at an unde-
ied pressures provided the sorption isotherms, i.e., termined rate. When the pressure is released, COz
the equilibrium solubility of C 0 2 in the polymer as is rapidly desorbed, as in the binary case. But now,
a function of C 0 2 pressure. Thus, a complete de- extrapolation of the initially linear weight vs. t ' / ,
scription of both the transport kinetics and the plot for desorption gives the sample weight gain due
equilibrium solubility of high-pressure C 0 2 in a to sorption of both components. Because of the much
polymer could be developed from gravimetric data higher diffusivity and volatility of CO, compared to
234 BERENS ET AL.

Time -
Figure 3 Schematic curves for pressure and sample weight vs. time during sorption/
desorption experiments on polymer/CO,/additive ternary systems: 0 load polymer film
sample and additive and evacuate; 0 apply COP pressure; @ polymer absorbs CO, and
additive; @ rapidly release pressure; 6)transfer sample to balance to record weight; 8 COP
is rapidly desorbed 0 additive desorbs very slowly.

the typical additives, the desorption of C 0 2 is much four common glassy polymers: poly (vinyl chloride)
faster than that of the additive; the time scale for ( PVC ) , polycarbonate ( PC ) , poly ( methyl meth-
C 0 2 desorption can be estimated from binary ex- acrylate) ( PMMA) ,and poly (vinyl acetate) (PVA).
periments on samples of the same polymer at similar The solubility of C02 in all these polymers is sub-
thickness. When the CO, has largely escaped, the stantial, though it varies markedly with the polymer
weight loss becomes much slower or virtually stops; structure. The downward curvature of the PVC and
the remaining weight gain of the polymer, above its PC isotherms illustrates the “dual-mode’’ form typ-
initial weight, is then a close approximation of the ical for gases absorbed in glassy polymers. The re-
amount of additive present in the sample at the end verse curvature for PMMA and PVA is character-
of the absorption period. Just as in the CO,/polymer istic for swelling agents in rubbery polymers and
binary systems, the absorption kinetics and equilib- suggests that these two polymers are plasticized into
ria of the additive in the C0,-swollen polymer at the rubbery state by C 0 2 at sufficiently high pres-
various C 0 2 pressures can be defined by a series of sure, i.e., their glass transition temperature is de-
experiments using varied absorption times. pressed below 25°C by the absorbed C02.
In Figure 5, D d (mean diffusivities for desorption
over the concentration interval of the experiment)
of CO, in PVC, PMMA, cellulose acetate (CA) , and
RESULTS AND DISCUSSION
polystyrene (PS) at 25°C are plotted vs. the initial
C02concentration. At low concentrations, there are
Polymer/ C 0 2 Systems substantial differences in COP diffusivity between
these polymers; the order of Dd, Ps % Pvc
Results of our gravimetric desorption experiments > PMMA, is the same as previously found for or-
on the interactions of compressed C 0 2 with a num- ganic vapors in these glassy polymers’ and presum-
ber of polymers have been reported elsewhere.’ Both ably reflects differences in packing or stiffness of
equilibrium solubility and kinetic data provide evi- the polymer chains. The diffusivity of CO, increases
dence of the plasticization of most glassy polymers with concentration in all four polymers, approaching
by CO,; illustrative data are repeated here for ready a range of to lop7 cm2/s, typical for C 0 2 in
reference. rubbery p01ymers.I~ The transport kinetics thus
Figure 4 shows the 25OC sorption isotheims, at provide further evidence of the plasticizing action
pressures up to 6.53 MPa (950 psia, liquid CO,) ,for of C 0 2 in glassy polymers.
 APPLICATION OF COMPRESSED C02 INTO POLYMERS 235

40 I PVA

0
0 200 400 600 800 1000
P, psia

Figure 4 Sorption isotherms for CO, in four glassy polymers a t 25°C.

The System PVC/CO,/ Dimethyl Phthalate glassy polymer. DMP is well known to be a plasti-
cizer for PVC, i.e., it has a substantial solubility in
We have used the combination of PVC with di- the polymer. From published values of the polymer/
methyl phthalate ( D M P ) as a model polymer/ad- solvent interaction parameter for this pair,14 it is
ditive system for exploring the effects of COz on the estimated that PVC in equilibrium with pure liquid
transport kinetics and equilibria of an additive in a DMP would be swollen at least 300%. The rate of

-7

-10 1 I I I
0 10 20 30
CO,, g l l 0 0 g polymer

Figure 5 Diffusivity of CO, vs. concentration in four glassy polymers.

236 BERENS ET AL.

50

40
0
>
n.
m
8
.v-
m
30

fi
I
L
20
0
N
0
u
10

A I I I J
0 v
0 100 200 300 400 500
Time, hrs
Figure 6 Sorption/desorption in C02/DMP/PVC system, 0.6 mm film.

absorption of DMP by PVC at room temperature, the normal behavior for a simple PVC/plasticizer
however, is extremely slow: Immersion of PVC film system.
samples in excess DMP for times up to 64 h resulted The effects of PVC film thickness and CO, pres-
in absorption of no more than 1wt % of DMP. Thus, sure on the absorption of DMP are illustrated in
the absorption of DMP by PVC is thermodynami- Figure 7. Amounts of DMP absorbed by PVC of four
cally favored, but kinetically limited at room tem- different thicknesses, in the presence of COPat 6.53
perature. DMP has limited miscibility with liquid MPa (950 psia), are plotted vs. the square root of
CO,; Francis l5 reports the compositions of the two sorption time in Figure 7 (A);the equilibrium uptake
phases of the C02/DMP system a t equilibrium at is about 45 wt % for all the films, but the time re-
25°C to be 94/6 and 43/57 wt % C02/DMP, respec- quired to approach sorption equilibrium increases
tively. with film thickness. Figure 7 ( B ) shows results of
Results of a typical series of sorption/desorption similar experiments under 3.78 MPa (550 psia) C 0 2
experiments on the ternary PVC/DMP/C02 system pressure; here the approach to equilibrium is much
are presented in Figure 6. Here, samples of 0.6 mm slower, but the equilibrium uptake is substantially
PVC film were exposed to an excess of DMP under higher, compared to results a t the higher CO, pres-
6.53 MPa (950 psia) COPpressure (liquid COP)for sure. The shifts in sorption kinetics from apparent
times up to 64 h. Amounts of COP plus DMP ab- Fickian to non-Fickian swelling observed with
sorbed were determined from the sample weights at changes in C02 activity and with changing film
zero desorption time (extrapolated). The amount thickness are reminiscent of similar behaviors ob-
of DMP absorbed was estimated from the weight served for organic solvent / PVC systems described
after 24 h desorption, the time for about 95% COB and discussed in detail in a recent publication from
desorption in comparable binary runs. The sorption our laboratory.'6 Such behaviors seem typical of
of DMP appeared to reach an equilibrium level near glassy polymer/penetrant systems when large vari-
45% of the PVC weight by 64 h. After the rapid ations in penetrant activities and sample thickness
desorption of the CO,, the data show a continued are studied.
very slow weight loss attributable to desorption and The effect of COzpressure on the kinetics of DMP
evaporation of the slightly volatile DMP. Even after absorption into PVC is further illustrated in Figure
more than 1000 h, however, more than 80% of the 8,where M,/M,, the fraction of equilibrium sorp-
absorbed DMP remained within the PVC. Clearly, tion, is plotted vs. t'I2/Z, with 1 the film thickness.
the absorption of DMP was strongly accelerated by Such plots are initially linear for sorption following
the presence of COP, while its slow desorption is Fickian diffusion kinetics, with slopes proportional
 APPLICATION OF COMPRESSED CO, INTO POLYMERS 237

- A) 6.53 MPa (950psia)

20

u
>
n o
ul 0 2 4 6 8
0

.
0
v-
n
B) 3.78 MPa (550 psia)
s J

6ol
0
n
P)

40

0 2 4 6 8
$12, hrl/2

Figure 7 DMP sorption by PVC under COPpressure: weight absorbed vs. G,varied
film thickness.

to D ' I 2 . The initial uptake data are sufficiently lin- in PVC has not been measured, but the correlation
ear to allow estimation of apparent Fickian diffu- with molecular size (Fig. 1 ) indicates that D for
sivities for DMP in PVC a t varied COz pressures, DMP in unplasticized PVC would be -4 cm2/
with the results shown in Table I. D for DMP alone s. It therefore appears that the diffusivity of DMP

1.o

0.8

h
& 0.6
z
n
v

-"
d
5 0.4

0.2

0.0
0 10 20 30 40 50

6ttl.t.(\'sec/cm) x 0.001

Figure 8 DMP sorption by PVC: fractional uptake, M J M , vs. G / l , at varied C02

pressure.
238 BERENS ET AL.

i
\
\
\
80
loo \

KINETICALLY
LIMITED
20

0 . l , , , I , I , I , I , I

0 1 2 3 4 5 6 7
(1018.1 psia)
CO, pressure, MPa

Figure 9 DMP uptake by PVC vs. COPpressure; 0.1 mm film, 17 h exposure.

in C02-swollenPVC may be larger than that in pure A reinforcing factor may be the increasing concen-
PVC by a factor of a t least lo6. tration of C02 in the PVC; given the low solubility
Figure 9 shows the effect of C02pressure on the of DMP in COP,it is reasonable to expect increased
amount of DMP absorbed by 0.13 mm PVC films C 0 2in the polymer to reduce the solubility of DMP
in 17 h at 25OC. The uptakes as functions of sorption in that phase.
time showed that equilibrium was reached in 17 h
at pressures of 2.06 MPa (300 psia) and higher. At
Other PVC/COz/Additive Systems
lower COPpressure, however, the DMP sorption had
not approached equilibrium in 17 h; hence, the ob- In an effort to define, at least qualitatively, the scope,
served uptake was kinetically limited. The clear limitations, and governing parameters of the C02-
trend of the equilibrium data is a decrease in the assisted polymer impregnation process, exploratory
amount of DMP absorbed with increasing C02pres- experiments have been carried out with a variety of
sure. The probable explanation of this effect is that compounds as “additives.” These have included
increasing COz pressure increases the concentration other plasticizers, model compounds of varied
of C 0 2 dissolved in the DMP, thus reducing the chemical type, bioactive compounds for potential
thermodynamic activity of the DMP in the DMP/ slow-release applications, and others. With most of
C02phase. The decrease of equilibrium concentra- the candidate additives, experiments were less ex-
tion of DMP in PVC with increasing CO, pressure tensive and systematic than with DMP, as the ob-
is thus the normal effect of decreasing DMP activity. jective was mainly to learn what types of compounds
could be infused into PVC with the help of com-
Table I Apparent Diffusivity of DMP in PVC pressed C02.
at Varied Pressures of COz In one series of experiments, PVC samples from
0.1 to 1.2 mm thick were exposed for 71 h, in the
COz Pressure Apparent D presence of liquid C02, to each of the “additives”
(MPa) (cm2/s) DMP, naphthalene, n -butanol, and octane. The
amounts of each additive absorbed, determined from
0 ( 4 0 - l ~ ) the sample weights after most of the COz had de-
1.4 1.0 x 10-1’ sorbed, are listed in Table I1 as percent of the PVC
2.3 4.0 X lo-’’ sample weight. Octane and butanol are miscible with
3.8 1.3 x 10-9
liquid COP,l7 but are nonsolvents for PVC; the ab-
6.5 8.0 x 1 0 - ~
sorption of these compounds was negligible.
 APPLICATION OF COMPRESSED COz INTO POLYMERS 239

Naphthalene and DMP, on the other hand, have Table I11 Absorption of 1-Hexanol by Four
limited solubility in liquid C02,l5 but are quite sol- Polymers (0.1 mm Films, 1 H Exposure)
uble in PVC. The results of these experiments, and
W t % Hexanol Absorbed
analogous results with other additive compounds,
seem to show a direct correlation between additive Polymer At 6.53 MPa COP Without COP
solubility in the PVC and amounts infused into the
polymer; infusion levels do not directly correlate PVC 0.5 -0
with the solubility of the additive in C02. PS 1.2 <1
PMMA 24 <1
PC 0.9 -0
Other Polymer/Additive/CO, Systems
A number of polymers other than PVC have also
been used in exploratory C02-assisted impregnation
trials. The variables of sorption time, C02pressure, Films of poly( caprolactone) (PCL),ethyl cellulose
etc., were not explored in sufficient detail to define ( E C ) , and cellulose acetate (CA), 0.25 mm thick,
the kinetics or equilibrium solubilities for each sys- were contacted with their own weight of either in-
tem, but it was established that the presence of C 0 2 domethacin or aspirin in the presence of liquid C02
a t high pressure could greatly accelerate the ab- for 64 h a t room temperature. The amounts of ad-
sorption of additives in many different polymers. A ditives infused, as percent of the polymer weight,
few examples are cited here to illustrate the scope are listed in Table IV. As in the previous examples,
and limitations of the infusion process. the amounts infused vary with both the additive and
One type of experiment compared the absorption the polymer, again indicating the importance of
of a particular additive in several different polymers specific additive/ polymer solubility.
under similar conditions. For example, 0.1 mm films Nearly 100 polymer / additive combinations, in-
of PVC, PS, PMMA, and PC were contacted with volving about 20 different polymers and over 40 ad-
an excess of 1-hexanol and CO, at 6.53 MPa (950 ditives, have now been tested in this study of C02-
psia) pressure for 1 h at room temperature. The assisted impregnation. Rather than cite further spe-
amounts of hexanol absorbed, as a percent of the cific examples here, we may simply say that most
sample weight, were determined from the sample cases seem to follow the general, qualitative rule that
weight after essentially all the COz had desorbed. COPwill markedly accelerate the absorption of ad-
For comparison, similar films were exposed to hex- ditives that are soluble in the particular polymer
anol for 1h with no C02present. Table I11 gives the used. Additives incompatible with the polymer gen-
results. A substantial amount of infusion occurred erally are not absorbed to a significant extent even
only in PMMA, the one polymer in the group in in the presence of high-pressure C02.
which hexanol is appreciably soluble. It is presum-
ably the very low diffusivity of hexanol in PMMA Desorption of Additives
that accounts for its low uptake in the absence of
C02. This experiment again shows a direct corre- After impregnation, the rapid desorption and con-
lation between the solubility of the additive in the sequent deswelling of the polymer matrix generally
polymer and the concentration resulting from the results in deplasticization of the polymer before sig-
infusion process. nificant loss of the additive can occur. From their
The following examples illustrate the infusion of molecular size, typical additives would be expected
pharmaceutical compounds into glassy polymers: to have extremely low diffusivities in glassy polymers

Table I1 Additive Infusion into PVC (0.1-1 mm Table IV Absorption of Pharmaceuticals by

Films, 71 H in Liquid COz) Glassy Polymers (0.25 mm Films, 64 H Exposure
in Liquid COz)
Additive Wt ?6 Infused
Polymer Wt % Indomethacin Wt % Aspirin
Octane <2
n-Butanol <1 PCL <1 8
DMP 8-37 EC 4 12
Naphthalene 13-15 CA 2.4 3
240 BERENS ET AL.

I I I I I
0.01 0.1 1.o 10 100 1000
Time, hr

Figure 10 Release of infused naphthalene from a PVC suspension resin (Geon 103EP)
into 50/50 v / v ethanol/water at room temperature.

(cf. Fig. 1) and, consequently, very long desorption additive left adhered to the surface of the polymer
times. Indeed, the release rates are sufficiently slow after infusion.
that such systems have potential for controlled-re-
lease materials and devices.
Mechanism
To illustrate, Figure 10 shows the release of
naphthalene (still a useful insect repellant) from 1 There seem to be two possible interpretations of the
g of a typical PVC suspension resin powder (Geon mechanism by which COPpromotes the impregna-
103EP, BFGoodrich Co.) into 100 mL of a 50/50 v / tion of polymers with additives: One is that the ad-
v mixture of ethanol and water. The wash mixture ditive dissolves in the COz and the polymer is then
was changed periodically to minimize the buildup of swollen by the COPsolution of the additive.' If this
the external naphthalene concentration. The naph- were the case, one would expect that any additive
thalene concentrations in the wash solutions were soluble in COz would be infused into any polymer
determined using a calibrated UV spectrophotom- that COz will swell, and the concentration of the
eter. Although the liquid COP-assisted infusion of additive in the COP would be the major factor af-
naphthalene into the resin was accomplished in only fecting the amount of additive impregnated into the
1 h at 25"C, the release of the additive from the polymer. This concept does not seem consistent with
resin continued after more than 300 h. our results; however, for example, COz-misciblebu-
Progesterone was infused into a 0.014 cm (5.5 tanol and octane are not absorbed by COP-swollen
mil) film of polycarbonate ( M39F, Mobay Chemical PVC, while the slightly COP-solublenaphthalene is
Co.) to a level of 0.5 wt % after exposure of the film highly absorbed.

-
to the additive in the presence of COPat 5.95 MPa
( 865 psia) for 66 h at 25°C. After C 0 2desorption,
a 1 g sample of the infused film was washed with
An alternative interpretation considers that the
COP-assistedimpregnation process actually involves
the approach toward an equilibrium distribution of
100 mL of a 75/25 v/v mixture of ethanol and water three components (polymer, additive, and COP)
with periodic solution changes and measurements among the phases present. The phase relations in
as above. The release of progesterone from the film three-component systems involving COPare highly
is graphed in Figure 11 and shows nearly linear re- varied and complex, l5 but a somewhat simplified
lease of the additive over a period approaching 400 view seems appropriate for the systems and condi-
h. In both the naphthalene and progesterone ex- tions considered here. The polymers we have used
periments, a rapid initial release of the additive was have negligible solubility in COPand are swollen only
observed and is believed due to a small quantity of to the extent of 30% or less by COP, and the volume
 APPLICATION OF COMPRESSED C 0 2 INTO POLYMERS 241

0
0 100 200 300 400
Time, hr

Figure 11 Release of infused progesterone from 0.014 cm-thick polycarbonate film into
75 / 25 v / v ethanol/water solution at room temperature.

of our pressure vessel is large compared to the vol- polymers tested, additives that exhibit good infu-
ume of swollen polymer. Thus, in the absence of sionability generally have solubility parameters, 6,
additive, our systems consist of a C02-swollen poly- fairly close to the reported 6 values of the polymers."
mer-rich phase and essentially pure C02,present as An example of this correlation is shown in Figure
vapor only, or vapor and liquid, depending on the 12, where additive uptakes in liquid-COP-assisted
pressure and temperature. It then seems to be a use- infusion into PVC are plotted against the solubility
ful and valid approximation to consider the impreg- parameter of the additive. Although the data show
nation process as the partitioning of the additive considerable scatter, it appears that the probability
between these phases, followed by rapid removal of of good infusionability is greatest for additives whose
the C02. In this light, the concentration of additive 6 values lie in a band from about 2 MPa ' I 2below to
absorbed by the polymer will depend upon the rel- 5 MPa'12 above that of PVC. The asymmetry of the
ative solubility of the additive in the polymer-rich correlation seems reasonable, since 6 values farther
and COPphases. Preferential partitioning of the ad- below that of the polymer would suggest greater sol-
ditive toward the polymer phase, i.e., a high degree ubility in liquid COP( 6 = 12.3 MPa'12) '' and con-
of impregnation, will be favored if the additive is sequent partitioning of the additive toward the C02
relatively more soluble in the polymer than in C02. rather than the polymer phase during the infusion
This seems to be the case, for example, in the PVC/ process.
DMP/C02 system. The partitioning concept, in-
deed, does seem qualitatively consistent with most
of our results. The major effect of the C02, then, is CONCLUSIONS
to accelerate the kinetics of additive absorption
when this is thermodynamically favored by the Useful equilibrium solubility and transport kinetic
polymer/additive interaction. In principle, then, it data on both C02/polymer binary and C02/poly-
should be possible to predict infusion results from mer /additive ternary systems are readily obtained
knowledge of the relative solubilities of additive through a simple gravimetric procedure involving
compounds in COPand in polymers. Unfortunately, rapid periodic weighing of samples during desorp-
such solubility data are not plentiful. tion after exposure to high-pressure C 0 2 .
In the absence of data allowing quantitative pre- Carbon dioxide under high pressure is rapidly ab-
diction of infusion results, the use of solubility pa- sorbed by many glassy polymers, plasticizing them
rameters seems to be of some value in estimating into the rubbery state at room temperature; upon
the suitability of untested polymer/ additive com- the release of pressure, C02 is rapidly desorbed
binations for COP-assisted infusion. For all of the and the plasticizing effect is reversed.
242 BERENS ET AL.

>50 e m

50

X
6 40
c
m
n
3
g 30
.-.-U
I

20
' e
0

10 I- -

0 I. I
12 15 20 25 30
-
1
6 , MPa2

Figure 12 Additive absorption in liquid-C0,-assisted infusion in PVC vs. additive sol-

ubility parameter.

The impregnation of glassy polymers with many 4. T. Hirose, K. Mizoguchi, and Y. Kamiya, J. Polym.
additive compounds is dramatically accelerated by Sci. Polym. Phys., 24, 2107 (1986).
the presence of COPa t sufficiently high pressure. 5. R. G. Wissinger and M. E. Paulitis, J. Polym. Sci.
Polym. Phys., 25, 2497 (1987).
The COP-assistedimpregnation process amounts,
6. W. V. Wang, E. J. Kramer, and W. H. Sachse, J.
in effect, to the partitioning of the additive be-
Polym. Sci. Polym. Phys., 20, 1371 (1982).
tween the COPand polymer phases; therefore, the 7. J. S. Chiou, J. W. Barlow, and D. R. Paul, J. Appl.
major factor in producing a substantial concen- Polym. Sci., 30,2633 ( 1985).
tration of absorbed additive is solubility of the 8. M. D. Sefcik, J. Polym. Sci. Polym. Phys., 24, 935,
additive in the polymer. 957 ( 1986).
Upon the release of pressure on a C02/polymer/ 9. A. R. Berens and H. B. Hopfenberg, J. Membrane
additive system, COPis rapidly desorbed and the Sci., 10, 283 (1982).
degree of plasticization is sharply reduced, leaving 10. A. R. Berens, Makromol. Chem. Macromol. Symp., 29,
95 (1989).
the absorbed additive trapped in the polymer to
11. A. R. Berens, G. S. Huvard, and R. W. Korsmeyer,
diffuse at the normal rate for the polymer/additive U.S. Pat. 4,820,752 (1989) (to BFGoodrich).
binary composition. 12. M. L. Sand, U.S. Pat. 4,598,006 (1986) (to Hercules).
13. J. Crank and G. S. Park, Diffusionin Polymers, Ac-
ademic Press, London, 1968.
14. P. Doty and H. Zable, J. Polym. Sci., 1, 90 (1946).
REFERENCES
15. A. W. Francis, J. Phys. Chem., 58, 1099 (1954).
16. A. R. Berens, J. Appl. Polym. Sci., 37,901 (1989).
1. A. R. Berens and G. S. Huvard, in Supercritical Fluid 17. D. K. Dandge, J. P. Heller, and K. V. Wilson, Ind.
Science and Technology, ACS Symposium Series 406, Eng. Chem. Prod. Res. Deu., 24, 162 (1985).
K. P. Johnston and J. M. L. Penninger, Eds., Amer- 18. E. A. Grulke, in Polymer Handbook, 3rd ed., J. Bran-
ican Chemical Society, Washington, DC, 1989, Chap. drup and E. H. Immergut, Eds., Wiley, New York,
14. 1989.
2. G. K. Fleming and W. J. Koros, Macromolecules, 19, 19. A. F. M. Barton, CRC Handbook of Solubility Param-
2285 (1986). eters and Other CohesionParameters, CRC Press, Boca
3. I. S. Liau and M. A. McHugh, Supercritical Fluid Raton, FL, 1983, Table 1,p. 219.
Technology, J. M. L. Penninger, M. Radosz, M. A.
McHugh, and V. J. Krukonis, Eds., Elsevier, New Received October 15, 1991
York, 1985, p. 415. Accepted November 8,1991