Chem. Mater. 1995, 7, 707-715
707
Kinetic Control of Pore Formation in Macroporous Polymers. Formation of “Molded”Porous Materials with High Flow Characteristics for Separations or Catalysis Frantisek Svec and Jean M. J. Frechet” Department of Chemistry, Baker Laboratory, Cornell University, Ithaca, New York 14853-1301 Received October 27, 1994. Revised Manuscript Received February 2, 1995@ The preparation of large macroporous polymer objects with controlled macroporous structures can be carried out in a n unstirred mold through careful control of the polymerization kinetics. The polymerization is carried out in a mold using a mixture of monomers, porogenic solvent and free-radical initiator under conditions t h a t afford macroporous objects with extremely large channels that provide for the high flow characteristics required for applications in separation or catalysis. In contrast, bead polymers prepared from identical polymerization mixtures but in a suspension polymerization process do not exhibit the same type of macroporous structure with large flow-through channels. The main differences between the two processes are the lack of interfacial tension between aqueous and organic phases and the absence of dynamic forces resulting from stirring in the case of the polymerization in a n unstirred mold. Control of the kinetics of the overall process through changes in reaction time, temperature, and overall composition allows the fine tuning of the macroporous structure and provides a n understanding of the mechanism of large pore formation within the unstirred mold. For example, a decrease in the reaction temperature t h a t slows down the rate of polymerization and the use of shorter reaction times than required for complete monomer conversion lead to porous objects with larger flow through channels.
Introduction Macroporous polymers are characterized by their rigid porous matrix that persists even in the dry state. These polymers are typically produced as spherical beads by a suspension polymerization process using a polymerization mixture that contains both a cross-linking monomer and an inert diluent, the porogen. Porogens can be solvating or nonsolvating solvents for the polymer that is formed, or soluble non-cross-linked polymers or mixtures of polymers and solvents. The mechanism of pore formation as well as the properties of macroporous polymers and their applications have been reviewed several The size distribution of pores within a porous polymer may cover a broad range from a few nanometers to several hundred nanometers. Pores with a diameter of less than 2 nm are classified as micropores, pores ranging from 2 to 50 nm are mesopores, while pores over 50 nm are macropores. The larger the pores, the smaller the surface area. Therefore, porous polymers with very large pores have relatively low specific surface areas, typically much less than 10 m2/g. The morphology of macroporous polymers is rather c o m p l e ~ . l , ~ They , ~ - ~ consist of interconnected microAbstract published in Advance ACS Abstracts, March 1, 1995. (1) Seidl, J.; Malinsky, J.; Dusek, K.; Heitz, W. Adu. Polym. Sci. 1967,5,113. (2) Guyot, A,; Bartholin, M. Prog. Polym. Sci. 1982,8, 277. (3) Hodge, P.; Sherrington, D. C., Eds. Syntheses a n d Separations Using Functional Polymers; Wilev: New York, 1989. (45 Kun, K. A,; Kinin, R. J . Pblym. Sci., A1 1968,6, 2689. (5) Pelzbauer, 2.; Lukas, J.; Svec, F.; Kalal, J. J . Chromatogr. 1979,
spheres (globules) that are partly aggregated in larger clusters that form the body of the beads. The size of the spherical globules that form the bulk of the macroporous polymer ranges from 10 to 50 nm. The pores in the macroporous polymer actually consist of the irregular voids located between clusters of the globules (macropores), or between the globules of a given cluster (mesopores), or even within the globules themselves (micropores). The pore size distribution reflects the internal organization of both the globules and their clusters within the macroporous polymer and largely depends on the composition of the polymerization mixture and the reaction conditions. The most effective variables that control pore size distribution are the percentage of cross-linking monomer, the type and amount of porogen, the concentration of the free-radical initiator in the polymerization mixture, and the reaction temperature.2 In analogy to conventional sieving processes, the use of polymers with large pores is advantageous in promoting rapid mass transfer through a porous polymer. In chromatography this may be beneficia17-11for a variety of preparative as well as analytical applications. In catalysis, convection through a catalyst that has very large pores increases the catalyst effectiveness factor,12 and large pore supports are therefore used in numerous
@
171,101. (6) Revillon, A.;Guyot, A.; Yuan, Q.; daPrato, P. React. Polym. 1989, 10, 11.
0897-4756/95/2807-0707$09.00/0
(7) Af’eyan, N. B.; Gordon, N. F.; Maszaroff, I.; Varady, L.; Fulton, S. P.; Yang, Y. B.; Regnier, F. E. J . Chromatogr. 1990,519, 1. ( 8 ) Tennikova, T. B.; Bleha, M., Svec, F.; Almazova, T. V.; Belenkii, B. G. J. Chromatogr. 1991,555,97. (9) Svec, F.; Frechet, J. M. J. Anal. Chem. 1992,64, 820. (10) Wang, Q. C.; Svec, F.; Frechet, J. M. J. Anal. Chem. 1993,65, 2243. (11) Frechet, J. M. J. Makromol. Chem., Makromol. Symp. 70171, 289, 1993. (12) Nir, A; Pismen, L. Chem. Eng. Sci. 1977,32, 35.
0 1995 American Chemical Society
708 Chem. Mater., Vol. 7, No. 4, 1995
catalytic processes.13 Other areas of application of very large pore materials include supports for the growth of mammalian cell cultures14 and the production of biomass.l5 Two approaches are most frequently used for the preparation of porous polymers with very large pores: (i) Polymerization of a mixture containing a large volume fraction of a non-solvating diluent.16 (ii) Polymerization in the presence of a linear polymer porogene4 J 7,18 Most of the macroporous polymers prepared to date have been almost exclusively produced in a shape of spherical beads that are used as ion-exchange resins, chromatographic separation media, adsorbents, etc. Therefore, studies of the mechanism of formation of macroporous structures have focused exclusively on materials prepared by suspension p~lymerization.l-~ For example, an extensive studylg of the effects of different variables on the properties of macroporous poly(glycidy1methacrylate-methylene dimethacrylate) beads prepared by suspension polymerization has appeared. The average pore size of the copolymers prepared in this study that involved the use of cyclohexano1 and dodecanol as porogens ranged from 20 to 150 nm.19 In our search for enhanced and simpler chromatographic separation media, we polymerized mixtures containing monomers and porogenic solvents directly within a chromatographic column used as a m ~ l d . ~ - l l The macroporous material that is obtained contains two very different families of pores:1° large channels and more conventional diffusive pores. Examination of the unusual pore size distribution curve of a typical poly(styrene-co-divinylbenzene) rod shows the existence of a sharp peak at about 1000 nm and another small peak in a size range corresponding to small mesopores.1°Rods prepared from poly(glycidy1 methacrylate-co-ethylene dimethacrylate) also contains similar bimodal pore size distribution including very large pores.g Because the rod columns are essentially a single "molded" polymer monolith traversed by large channels and permeated by small pores, their hydrodynamic properties are excellent and even high flow rates can be used. They are unlike any of the existing porous materials that are typically used in packed beds because flow through the rod column does not involve any interstitial space but results entirely from the existence of the large flow-through channels that are built into the porous polymer monolith. The continuous polymer rod media afford excellent resolution in the chromatographic separation of proteins, peptides, and small molec~les.~f',~~ Recently, our approach has been used for the preparation of continuous rods of molecularly imprinted polymers capable of (13) Rodrigues, A. E.; Lopez, J . C.; Lu, Z. P.; Loureiro, J . M.; Dias, M. M.J . Chromatogr. 1992,590,93. (14) Vournakis, J.; Ronstadler, P. BiolTechnology 1989,7 , 143. (15) Brettenbucher, K.; Siegel, K.; Knupper, A,; Radke, M. Water Sci. Technol. 1987,5,835. (16) Chung, D. Y.; Bartholin, M.; Guyot, A. Angew. Mukromol. Chem. 1982,103,109. (17) Hilgen, J.; deJong, G. J.; Sederel, W. L. J . Appl. Polym. Sei. 1975,19,2647. (18) Guyot, A.; Revillon, A.; Yuan, Q. Polym. Bull. 1989,21,577. (19) Horak, D.; Svec, F.; Ilavsky, M.; Bleha, M.; Baldrian, J.;Kalal, J . Angew. Mukromol. Chem. 1981,95,117. (20) Wang, Q. C.; Svec, F.; FrBchet, J . M. J . J . Chromatogr. 1994, 669,230.
Svec and Frdchet
molecular recognition of positional isomers and enantiomem21 All of these rods were prepared from polymerization mixtures essentially identical to those that are used for the preparation of macroporous beads by suspension polymerization, yet beads prepared in parallel experiments by suspension polymerization do not contain any of the very large micrometer-size pores found in the molded continuous media.lg This indicates that somewhat different mechanisms of pore formation must operate during the preparation of macroporous rods by our approach and of beads by the standard suspension polymerization technique. This report explores the effects of polymerization conditions on the porous properties of rods prepared by polymerization of a mixture containing glycidyl methacrylate and ethylene dimethacrylate in a steel tube and provides an explanation for the formation of much larger pores during the polymerization in a tube as compared to the porous beads resulting from a suspension polymerization.
Experimental Section Preparation of Polymers. Polymerization Mixture. Azobisisobutyronitrile (1% of the weight of monomers, Kodak) was dissolved in 4 vol parts of a mixture consisting of 60% glycidyl methacrylate (2-methyl-2-propenoic acid oxiranylmethyl ester, CAS reg. no. 106-91-2,Aldrich) and 40% ethylene dimethacrylate (2-methyl-2-propenoic acid 1,2-ethanediyl ester, CAS reg. no. 97-90-5, Sartomer). Cyclohexanol (Aldrich) was admixed slowly to the monomers followed by the addition of dodecanol (Aldrich); the total volume of the alcohols was 6 parts. The mixture was purged with nitrogen for 15 min. The stock polymerization mixture was stored in a closed flask in a refrigerator a t a temperature of 5 "C and consumed within 7 days. Typically, polymerizations were repeated with two different fresh mixtures and with duplicate experiments done for each polymerization mixture. Suspension Polymerization. The polymerization mixture (4 parts) was added to a 1%aqueous solution of poly(vinylpyrrolidone) (Aldrich) MW 360 000 (6 parts) and deaerated. The polymerization was carried out in a 250 mL glass reactor (Buchi BEP 280) equipped with an anchor stirrer and a heating jacket. The beads were washed with water, extracted in a Soxhlet apparatus with methanol for 24 h and dried at 60 "C. Polymerization in Bulk Solution. A stainless steel tube (50-mm x 8-mm i.d., Labio) was charged with 2.5 mL of the polymerization mixture then sealed with rubber nut plugs. The polymerization was allowed to proceed in a water thermostat. The tubes either stood vertically in the bath or the contents were subjected to an end-over-end rotation while immersed. After the chosen polymerization time elapsed, the rubber plugs were replaced at one end by the column end fitting and the rod was forced out of the steel tube by applying a pressure of THF using a chromatographic pump. The length of the rod was measured using a ruler. The soluble compounds were removed from the rod by extraction in a Soxhlet apparatus with methanol for 24 h and the rod was dried a t 60 "C. The conversion was calculated from the weight of the extracted dry rod. In a modified procedure, the polymerization mixture was placed in a 5 mL polypropylene syringe barrel, the piston was left in the upper position, and the syringe was submerged in a water bath. Once the polymerization was completed, the end of the barrel was cut off and the rod was pushed out of the plastic tube using the syringe piston. (21) Matsui, J.; Kato, T.; Takeuchi, T.; Suzuki, M.; Yokoyama, K.; Tamiya, E.; Karube, I. Anal. Chem. 1993,65, 2223.
Chem. Muter., Vol. 7, No. 4, 1995 709
Pore Formation in Macroporous Polymers
Table 1. Polymerization Conditions and Properties of the Macroporous Poly(glycidy1methacrylate-co-ethylene dimethacrylateP
i o
t 1
100 Pore dlameter, nm 10
1000
Figure 1. Differential pore size distribution curves of the poly(glycidy1 methacrylate-co-ethylene dimethacrylate) beads (W) and rod ( 0 ) prepared at a temperature of 70 "C. For conditions see Table 1. Porous Properties. Following washing or solvent extraction, the porous properties of the beads or rods were determined by mercury intrusion porosimetry and the specific surface areas calculated from nitrogen adsorptioddesorption isotherms using a custom made combined BET-sorptometer and mercury porosimeter (Porous Materials, Inc., Ithaca, NY). Prior to the measurements, the rods were cut to small pieces with a razor blade. Gas Chromatography. Gas chromatographic determinations were carried out in a H P capillary column (crosslinked methylsilane gum, 0.d. 0.32 mm, length 25 m, i.d. 0.17 mm, temperature gradient from 100 to 280 "C in 15 min) using a Hewlett-Packard 5890 chromatograph equipped with a HP76739 automatic autosampler and TCD detector and helium as a carrier gas. The data were collected by a HP-3393 integrator.
expt polymerization 1 suspension 2 suspension 3 PPtube 4 steel tube 5 PPtube 6 steel tube 7 steel tube 8 suspension 9 PPtube 10 steel tube 11 PPtube 12 steel tube 13 steel tube
dodecanol;
temp,
%
"C
0 6 6 6 6 6 6 12 12 12 12 12 12
70 70 70 70 55 55 50-70f 70 70 70 55 55 50-7Of
V,,: mug 1.23 1.29 1.40 1.33 1.33 1.10
1.33 1.39 1.58 1.46 1.24 1.18 1.50
S,,d m2/g 96.0 173.6 128.4 137.2 62.8 65.6 103.1 102.8 94.2 102.7 38.7 81.2 172.3
Dp,max,@ nm 53 63 91 93 809 935 214 85 283 315 1530 1527 1690
Reaction conditions: polymerization mixture: glycidyl methacrylate 24%, ethylene dimethacrylate 1696, porogenic solvent (cyclohexanol dodecanol) 60% Percentage of dodecanol in the polymerization mixture. e Pore volume. Specific surface area. e Median pore diameter. f Temperature was raised from 50 to 70 "C in steps by 5 "C lasting 1 h each and kept a t 70 "C for another 4 h. a
+
Suspension polymerization is generally treated in the literature as a variant of bulk polymerization in which each droplet of the dispersed phase containing monomer is an individual bulk reactor.22 Therefore, one might have anticipated that the properties of the products of both suspension and bulk polymerizations would be nearly identical. As indicated above, this is not the case, and properties such as pore size distribution are actually entirely different. Figure 1 shows the considerable discrepancy that exists between the pore size distributions of macroporous glycidyl methacrylate-ethylene dimethacrylate copolymers prepared by both suspension and the bulklike rod polymerization at 70 "C from an identical polymerization mixture containing 12%dodecanol, 48% cyclohexanol, 24% glycidyl methacrylate, and 16% ethylene dimethacrylate. The median pore size diameter for the beads is 85 nm while for the rod it is 315 nm. In contrast to the median pore diameter, the specific surface areas and the pore volumes do not exhibit such marked differences (Table 1). Since the reaction conditions in both polymerizations were comparable, this unprecedented difference in median pore size diameter has to result from the polymerization technique itself. While suspension polymerization has already been analyzed in the literature several little is
known on how to control the properties of macroporous polymers obtained by a bulk polymerization within a mold. Therefore, we have studied this type of polymerization more thoroughly and investigated the effect of reaction variables such as composition of the porogenic solvent, reaction time, and reaction temperature on the porous properties of the molded rods. We did not take into consideration two other variables, the concentration of cross-linking agent and the monomers to porogenic solvent ratio. On the basis of our experience, we chose a standard composition of monomer mixture including 40% ethylene dimethacrylate and 60% glycidyl methacrylate for all experiments. This composition is deemed ideal because any lower concentration of the cross-linking agent could impair the mechanical properties of the final polymer rods, while a higher one would decrease the content of reactive epoxide groups that are needed for the subsequent functionalization of the rods. The 4:6 monomers/porogenic solvent ratio has already proven to be the most advantageous for the preparation of macroporous poly(glycidy1 methacrylate-eo-ethylene dimethacrylate) materialseZ3 Effect of Polymerization Time. The influence of reaction time on conversion is well demonstrated for all reactions. Since the polymerization at 70 "C proceeds too fast to be monitored readily, we chose a temperature of 55 "C at which the rate of polymerization is low enough t o be readily monitored (Figure 2). Although the conversion of monomers to polymer is close to quantitative after about 10 h, some additional structural changes still occur within the rod if the system is kept longer at the polymerization temperature. However, no significant changes are observed at reaction times exceeding 22 h. The length of the completely polymerized rod prepared under the conditions specified in Table 2 in a tubular mold charged with 2 mL of the polymerization mixture is 35 mm. It would be expected that the length of the rod itself would not depend on the polymerization
(22) Yuan, H. G.; Kalfas, G.; Ray, W. H., J. Macromol. Sci., Reu. Macromol. Chem. Phys. 1991, C31, 215.
(23) Horak, D.; Pelzbauer, Z.; Svec, F.; Kalal, J. J. Appl. Polym. Sci. 1981, 26, 3205.
Results and Discussion
710 Chem. Mater., Vol. 7, No. 4, 1995
Svec and Frkchet
Table 2. Porous Properties of Poly(glycidy1methacrylate-co-ethylenedimethacrylate) Rods Prepared Using Different Polymerization Timesa
BET
mercury porosimetry
v,:
min
60 75 100 130 150 200 300 600 1320 1800
3.759 3.453 2.926 2.532 2.347 1.673 1.312 1.257 1.093 1.108
217.5 149.2 136.0 127.8 123.8 125.0 73.0 79.2 65.6 66.0
702 870 966 1124 1090 974 966 934 935 940
S,,d m2/g
mug 0.688 0.360 0.335 0.296 0.287 0.239 0.165 0.153 0.139 0.140
618 811
996 1201 1150 1099 1128 1125 1154 1148
523.9 317.3 283.5 255.7 249.1 239.8 149.7 138.0 120.1 119.8
D P , dnm 6.33 6.53 6.37 6.71 6.82 6.32 6.22 6.18 6.78 6.49
Dp,surf,h nm 3.44 3.28 3.40 3.29 3.22 3.28 3.28 3.33 3.33 3.29
Reaction conditions: polymerizationmixture: glycidyl methacrylate 24%,ethylene dimethacrylate 16%, cyclohexanol54%, dodecanol 6%; polymerization temperature 55 "C. Polymerizationtime. Pore volume. Specific surface area. e Median pore diameter. f Pore diameter at the maximum of the distribution curve. g Median pore diameter based on pore volume. Median pore diameter based on surface area.
100
.
j
400
I
I
300
$>
s C
200
5 0 i t
1
1.
100
I
I.
0
0
1000
2000
60
20
Polymerization time, min
Figure 2. Kinetics of the polymerization of glycidyl methacrylate and ethylene dimethacrylate at a temperature of 55 "C. For conditions see Table 2. time as the polymerization ought to take place throughout the entire volume of the mixture in the tube. However, this is not the case, and if tubular molds were held vertically during the polymerization reaction, the rods obtained after 60 and 75 min of polymerization were significantly shorter (21 and 25 mm, respectively) and occupied only the bottom part of the mold. The liquid remaining on the top of the rods under these conditions was removed with a syringe and analyzed by gas chromatography. Even after 150 min of polymerization a small amount of the liquid was still found, but this is most likely due to the oxygen inhibition of the polymerization at the surface of the rod because the tube was not completely filled with the polymerization mixture and the residual space can contain some air. The composition of all the liquids collected was generally close to that of the original polymerization mixture. The liquid did not contain any polymeric components as confirmed by the lack of precipitation during dilution of the samples with methanol for GC analysis. Table 2 summarizes the porous properties of the rods. During the early stage of the polymerization, the pore volume is very high reaching almost 4 m u g . This represents a porosity of about 82% at the low conversion of 15.7%. The pore volume decreases as the polymerization progresses, eventually reaching a value slightly above 1 m u g that represents a porosity of about 60%. This porosity is directly related to the volume of the porogenic solvents used for the polymerization and confirms that all of the monomers have been consumed.
100
Converrion, %
Figure 3. Effect of conversion on the specific surface area determined by BET method (U) and mercury intrusion porosimetry (A)during the polymerization at a temperature of 55 "C. For conditions see Table 2.
e ,
I
- 1
'n
I
s 0
" 3
5 4
0 1
10
100
1000
10000
Pore diameter, nm
Figure 4. Differential pore size distribution curves of the poly(glycidy1 methacrylate-co-ethylenedimethacrylate)rods after 1 h (W) and 14 h ( 0 )of polymerization at a temperature of 55 "C. For conditions see Table 2. The specific surface area, calculated from both mercury porosimetry and BET measurements, also decreases with the polymerization time. Figure 3 documents that the specific surface area decreases linearly within the range of conversions from 20 to 100%. Figure 4 shows the pore size distribution curves for rods formed after 60 and 1320 min, respectively. Although the maximum of curve l (corresponding to the most porous rod formed within 1 h) is located a t 618
Chem. Mater., Vol. 7, No. 4, 1995 711
Pore Formation i n Macroporous Polymers
E
10000
I8
1
1
cn
z
0
n
d
0
5
i5
10
70
40
100
1
10
100
1000
10000
Pore dlemeter, nm
Converslon, 96
Figure 5. Diameter of the largest detectable pores in the poly(glycidyl methacrylate-co-ethylenedimethacrylate) rods prepared a t a temperature of 55 "C as a function of conversion. For conditions see Table 2
Figure 7. Differential pore size distribution curves of the poly(glycidy1 methacrylate-co-ethylene dimethacrylate) rods prepared from mixtures containing 6% (A,a) and 12% dodecanol(M, 0 ) by a polymerization at a temperature of 55 "C (closed points) and 70 "C (open points). For general conditions see Experimental Section.
100
n
I
..
0
E.
P
,
I
-100
0
E.
n
m
-200
-300 0
I
I
10
20
time, h
Figure 6. Difference between the calculated median pore diameter Dp,med and the pore diameter corresponding to the maximum of the distribution curve Dp,mar for the poly(glycidy1 methacrylate-eo-ethylene dimethacrylate) rods prepared a t a temperature of 55 "C as a function of polymerization time. For conditions see Table 2.
nm, the rod also contains a substantial amount of very large pores with diameters up to 10 mm. In contrast, the peak for curve 2 is located at 1154 nm, but it is narrower and without pores over 2 mm in diameter. The almost 4-fold difference in the pore volumes of the two molded rods obtained after 1 and 14 h, respectively, is also reflected in the much larger area beneath curve 1, particularly in the range of large pores. The pore size distribution narrows as the polymerization progresses because the largest pores disappear. Figure 5 shows the size of the largest pores detected by mercury porosimetry at different stages of the polymerization of the rod and documents that their size decrease is a function of the conversion. Mercury porosimetry measurements provide two kinds of pore diameters: the calculated median pore diameter Dp,med and the pore diameter that corresponds to the maximum read from the distribution curve Dp,max. Figure 6 shows that the difference Dp,med - D p , m a x decreases smoothly within the whole range of conversions. At low conversion, the median size exceeds the peak value. However, this already changes after about 90 min of polymerization as the median size decreases and the difference becomes negative. The data also
document that the very large pores that are characteristic of rods in the early stages of the polymerization and which contribute considerably to the median pore size, disappear as the polymerization progresses while the influence of the small pores on the average diameter becomes increasingly important. Table 2 shows that the size of the pore diameter D p , m a x reaches a plateau after about 2 h of polymerization. In contrast, the calculated median pore diameter D p , m e d initially increases, then reaches a maximum also after about 2 h. and then decreases again continuously. It should be emphasized that any direct comparison of the BET and mercury porosimetry data would not be appropriate as each method covers a different range of pores. This can be confirmed by the comparison of the pore diameter data summarized in Table 2. While the mercury porosimetry monitors efficiently the significant changes affecting the pore diameters, the BET data do not show any change in the median pore diameters calculated from both pore volumes and surface areas during the polymerization. On the other hand, the surface areas measured by BET involve also the pores smaller than those detected by the mercury intrusion method. Therefore, the BET specific surface areas are about twice as large as those calculated from the mercury intrusion porosimetry. However, Figure 3 shows that the trends in changes of specific surface areas are similar for both the BET and the mercury porosimetry measurements. Effect of the Porogenic Solvent. It was observed earlier that the addition of dodecanol to cyclohexanol used as the porogenic solvent results in the formation of larger pores in poly(glycidy1methacrylate-co-ethylene dimethacrylate) beads.lg This is also confirmed in this study. The median pore diameter for beads prepared at 70 "C in the presence of 0 , 6 , and 12% of dodecanol is 53, 63, and 85 nm, respectively (Table 1). Figure 7 shows the shift induced by dodecanol in the maxima of the pore size distribution curves for molded rods prepared a t two different temperatures. For example, the median pore size of rods prepared with 6 and 12% dodecanol a t 70 "C increases from 91 to 283 nm, respectively.
712 Chem. Mater., Vol. 7, No. 4, 1995
Svec a n d Frkchet
1
I
a
5 0
0 1
10
100
1000
1
10000
Pore diameter, nm
Figure 8. Differential pore size distribution curves of the poly(glycidy1 methacrylate-co-ethylene dimethacrylate) rods prepared by a 22 h polymerization at a temperature of 55 "C (e),12 h a t 70 "C (W) and a t a temperature increased during the polymerization from 50 to 70 "C in steps by 5 "C lasting 1 h each and kept at 70 "C for another 4 h (0).Conditions: polymerization mixture: glycidyl methacrylate 24%, ethylene dimethacrylate 16%, cyclohexanol 54%, dodecanol 6%.
Effect of the Polymerization Temperature. An effect similar t o that of dodecanol was also observed when the temperature was changed in otherwise identical preparations. The lower the polymerization temperature, the larger the pores. Figure 7 shows the effect of temperature on the pore size distributions of continuous rods prepared at 55 and 70 "C. This temperature effect may be very useful in practice because it allows the control of the pore size distribution of the molded rods without requiring any change in the composition of the polymerization mixture. However, the lower limit of the polymerization temperature depends on the decomposition rate of the free-radical initiator used. Figure 8 shows the pore size distribution curves for rods prepared in steel tubes under different temperature conditions. In addition to rods prepared by polymerization a t fixed temperature of 55 and 70 "C, Figure 8 also includes a curve obtained for a rod prepared using a step gradient of polymerization temperature. The temperature was increased during polymerization from 50 t o 70 "C in 5 "C steps lasting 1h each, and then the rod was kept at the final temperature of 70 "C for another 4 h. This approach was chosen to reduce the possibility of a steep radial temperature gradient that could lead t o a faster polymerization in the areas close to the walls of the tube with formation of a "shell", surrounding the liquid in the center of the tube. As expected, the maximum of the distribution curve in this rod prepared in a step gradient of temperature is between the maxima for the rods prepared at 55 and 70 "C because the average polymerization temperature was between these two temperatures. Since the polymerization temperature affects the pore size distribution, it is possible to use temperature changes during the polymerization to fine-tune the porosity of the rod. For example, a rod prepared by polymerization at 55 "C for 1 h then at 70 "C for 14 h with the mold standing vertically in the bath throughout the process shows a different porosity profile at its two extremities (Figure 9). While both parts of the rod contain large pores centered near 1 2 0 0 nm in diameter, the top of the rod
100
10
1000
10000
Pore diameter, nm Figure 9. Differential pore size distribution curves of the top (W) and the bottom part (A) of the poly(glycidy1methacrylateco-ethylene dimethacrylate) rod polymerized 1 h a t a temperature of 55 "C followed by 14 h a t 70 "C. Conditions: polymerization mixture: glycidyl methacrylate 24010, ethylene dimethacrylate 16%, cyclohexanol 54%, dodecanol 6010.
I
AI TOP
A
0
- 0.5
5 Q
w v ittom
0.0
f
10
I
I
1000
100 Pore diameter, nm
p4
10
100
I 1000
Pore diameter, nm
Figure 10. Differential pore size distribution curves of (a) top (a)and bottom part ( 0 )of the poly(glycidy1 methacrylateco-ethylene dimethacrylate) rod polymerized 1 h a t a temperature of 55 "C followed by 14 h at 70 "C; (b) rod polymerized 1h a t a temperature of 55 "C (W) and rod polymerized 14 h at 70 "C ( 0 ) . Conditions: polymerization mixture: glycidyl methacrylate 24%, ethylene dimethacrylate 1696,cyclohexanol 54%, dodecanol 6%.
which was formed in later stages of the polymerization contains a very significant volume of smaller pores. Figure 10a shows a magnification of the same porosity profiles for the top and bottom portions of the rod. While the volume of pores in the range 30-200 nm is
Chem. Mater., Vol. 7, No. 4, 1995 713
Pore Formation in Macroporous Polymers
only 6.7% in the bottom part, it accounts for a remarkable 25.5% at the top with a distinct maximum centered at about 80 nm. Figure 10b shows that the same type of porosity profiles can be obtained for a rod prepared by polymerization at 55 "C for 1 h only and then processed and for a rod obtained after the standard polymerization time of 14 h at 70 "C. This observation is readily explained if one considers that much unpolymerized material remains at the top of the polymerized rod after a polymerization time of only 1 h at the low temperature of 55 "C. If the polymerization is then completed at 70 "C, the porous solid that forms on top of the previously polymerized rod has the pore profile typical of rod prepared at 70 "C. Although this technique has not yet been explored in detail, it is extremely promising for the preparation of rods with a gradient of porosity that should prove useful in electrophoresis or novel modes of separation. The axial heterogeneity observed in the preceding experiment resulted from the insufficient polymerization time that left a pool of unpolymerized material on top of the growing rod. In contrast, the polymerized rod occupies the entire volume of the mold if the polymerization is allowed to proceed for 3 h at 55 "C. Therefore, a control experiment was carried out involving polymerization for 3 h at 55 "C followed by 8 h at 70 "C. In this case, porosimetric measurements reveal no difference in porosity profiles for the top and the bottom parts of the rod. No evidence of axial heterogeneity is uncovered and both parts of the rod have porosity profiles that are very similar to that of a rod obtained only at 55 "C. The fraction of pores in the range 30200 nm is only about 8%. This is not unexpected because the conversion after 3 h of polymerization is almost 70% and only 30% of monomers remain available for the further stages of polymerization at 70 "C. This is not enough to introduce a very substantial amount of smaller pores into the rod. Table 1 documents that there is essentially no difference in porous properties for rods prepared in steel or in polypropylene tubular molds. This confirms that heat convection through the wall of the tube (slower in the case of the plastic tube) is not a factor in the preparation of the rods. While the polymerization temperature is clearly a dominant factor controlling porosity, no effect of a small radial temperature gradient within the tubes has been observed. The heat of polymerization can also produce a radial temperature gradient and has to be taken into consideration. While dissipation of the heat does not represent a problem for columns with a diameter of up to 8 mm, it could considerably change the radial temperature profile in columns with larger diameter. Therefore a somewhat different polymerization technique has to be developed for the preparation of large preparative rods. Pore Formation in Polymer Rods. "Classical" Mechanism of Pore Formation in Macroporous Polymers. The generally accepted mechanism of pore formation during a typical polymerization in the presence of a precipitant is the f~llowing:l-~ The organic phase contains both monovinyl and divinyl monomers, initiator and porogenic solvent. The free-radical initiator decomposes at a particular temperature and the initiating radicals start the polymerization process in solution. The polymers that are formed by solution polymerization precipitate after they
become insoluble in the reaction medium as a result of both their crosslinking and the choice of porogen (poor solvent for the polymer). In this process, the monomers are thermodynamically better solvating agents for the polymer than the porogen. Therefore, the precipitated insoluble gellike species (nuclei) are swollen with the monomers that are still present in the surrounding liquid. The polymerization then continues both in solution and within the swollen nuclei. Polymerization within the latter is kinetically preferred because the local concentration of the monomers is higher in the individual swollen nuclei than in the surrounding solution. Branched or even cross-linked polymer molecules that can still be formed in the solution, are captured by the growing nuclei and further increase their sizes. The cross-linked character of the nuclei prevents them from mutual penetration and from the complete loss of their individuality through coalescence. The nuclei, enlarged by the continuing polymerization, associate in clusters being held together by polymer chains that cross-link the neighboring nuclei. The clusters remain dispersed within the liquid phase that is rich in the inert solvent (porogen) and keep growing. In the later stages of the polymerization, the size of the clusters is large enough to allow contact with some of their neighbors thereby forming a scaffolding-like interconnected matrix within the polymerizing system. The interconnected matrix gets reinforced by both interglobular cross-linking and the capture of chains that still polymerize in the solution phase during the continuing polymerization leading t o the final porous polymer body. The fraction of voids within the final porous polymer (macropores) is, at the end of the polymerization, close to the volume fraction of the porogenic solvent in the initial polymerization mixture because the porogen remains trapped in the voids of the cross-linked polymer. The classical mechanism of the pore formation during the heterogeneous polymerization of monomers in the presence of thermodynamically poor solvents used as porogens does not allow a prediction of the sizes of the pores that should result. The current knowledge of factors that control the pore size in macroporous polymers is mostly empirical, being based on the widely published pore size distributions of macroporous beads. Therefore, the knowledge acquired for beads does not appear to be directly applicable to macroporous molded rods as evidenced by the fact that the latter contain very large pores or channels not commonly found in beads. This raises the question of the origin of the pore size distribution of beads obtained by suspension polymerization: could this distribution be a direct result of the polymerization technique? Effect of the Tension at the Liquid-Liquid Interface. An analysis of the suspension system comprising generally two phases (aqueous and organic) reveals that the interfacial tension plays a very important role in droplet formation upon mixing. This includes the control of both the size and the spherical shape of the droplet^.^^,^* There is, however, another largely ignored effect of the interfacial tension that is particularly important when designing the suspension polymerization process for the preparation of macro(24) Brooks, B. W. Macromol. Chem., Macromol. Symp. 1990, 351 36, 121.
714 Chem. Mater., Vol. 7, No. 4, 1995
Svec and Frkchet
porous beads. This effect is related to the shrinkage that occurs during any vinyl p~lymerization.~~ Assuming that the process is free of coalescence, the size of the original droplets decreases during the suspension polymerization due to the volume shrinkage that is u n a v ~ i d a b l e .Referring ~~ back to the mechanism of the heterogeneous polymerization that leads to macroporous polymers, we know that at the early stage of the suspension process, the nuclei are randomly dispersed within the rotating droplets of the polymerization mixture. The size of the droplets decreases because of the shrinkage induced by polymerization. Since the interfacial tension exerts a constant pressure on the droplet surface, the spherical shape is retained, but the combined effects of the interfacial tension and the shrinkage push the nuclei closer to one another. This process may even lead to the formation of a dense shell covering the outer surface of the beads.j Since the monomers are diluted with the porogen, the overall volume shrinkage is only about 6% as opposed to approximately the 15% shrinkage ratio expected for undiluted monomers.25 Therefore, the interfacial tension is likely to contribute to the denser assembly of globules found in beads, but it can hardly be the only reason for the dramatic differences observed between the pore size distributions of beads and molded rods. Pore Formation during Polymerization in a Tube. The external conditions accompanying the polymerization within a tube are quite different from those found in the suspension polymerization process. First of all, only one phase, the organic mixture, is present in the tube. Therefore, the interfacial tension between the aqueous and organic phases characteristic of the suspension process is absent. Moreover, in contrast to the droplets that revolve in the aqueous phase as a result of stirring, the contents of the tubular mold generally do not move during the polymerization. Overall, an understanding of the features of suspension polymerization and the observations made during the polymerizations in the tube can help t o suggest an explanation for the differences in the porous properties of the beads and the rods. The basic mechanism of the polymerization in the presence of a porogenic solvent outlined above, including the precipitation of nuclei and shrinkage, is general and remains independent of the polymerization technique. However, the solid nuclei or their clusters have a higher density than the polymerization mixture. Therefore, in the absence of mixing and if the overall rate of polymerization is slow, they can sediment and accumulate at the bottom of the mold. This is documented both by the shorter length of the rod and by the presence of unchanged liquid polymerization mixture on the top of the rod in the very early stage of the polymerization at 55 “C, and by visual observation of polymerization process carried out in a transparent glass mold. The unstirred nuclei and their clusters sediment t o the bottom of the tube where they form a very loose, highly porous and less organized structure early on during the polymerization. This is confirmed by the presence of very large pores as seen in the pore size distribution curve of the rod after 60 min of polymerization (Figure 4, curve 1)and as confirmed by the large 125) Takata.
T.:Endo, T. Progr. Polym. Sci.
1993, 18, 839.
pore volume (Table 2). The very high specific surface area that is measured at this stage (523.9 m2/g) documents that the nuclei retain a great deal of their individuality. However, as they come into contact with each other, some growing polymer chains are able to connect them together and fix the loose structure. Therefore, even the rod obtained after 1 h of polymerization at 55 “C retains its shape and does not fall apart upon removal from the tube. This solid polymer has 80% porosity and is therefore fragile and easy to cut with a blade. The polymerization process continues both in the mixture that is collected on the top of the rod, as well as in the mixture that is still present within the swollen very large pores of the early rod resulting in the formation of some new nuclei. The following observations made at 55 “C support the suggested mechanism: (i)The disappearance of the very large pores in the later stages of the polymerization within a mold is clearly documented in Figure 5 that shows the size of the largest pores detected by mercury porosimetry as a linear function of the conversion. (ii)The individuality of the nuclei is reduced by the polymerization occurring within the swollen nuclei and by the capture of the dissolved polymers. This results in a decrease of the surface areas as documented in both Table 2 and Figure 3. In contrast to the molded rods, the growing nuclei in the droplets of the suspension polymerization mixture dispersed in the aqueous phase do not lose their individuality that early. The droplets revolve because of the stirring and the centrifugal force lets the nuclei move randomly within the droplet. The nuclei are therefore able to keep their individuality longer, grow separately and “pack” better within the bead (droplet). As a result, the voids between the globules that constitute a single bead are smaller. Therefore, the dynamics of the system seem to be the main cause for the difference in the pore size distribution between the molded rods and classical beads. Absence of Radial Shrinkage within the Mold. The above discussion describes the mechanism of pore formation during a polymerization within a tube, but it still does not explain completely why no radial shrinkage of the polymer rod within the mold is observed. This is likely to be also the result of both the absence of interfacial tension compressing the polymer as it is formed and the lack of mixing during polymerization. Once again, it must be emphasized, that the overall volume shrinkage is only about 6 %.24 Clearly, some of this shrinkage occurs during the early nucleation and growth within the “free-flowing” swollen nuclei. This decreases the overall volume of the polymerization mixture but does not affect the rod that has not yet been formed. Any shrinkage of the size of the polymer rod would be expected to occur only in the late stages of the polymerization when the rod already exists. At that time, the matrix is already heavily cross-linked and can hardly change its overall size. The shrinkage is likely to occur within the rod itself leading to the formation of larger or more numerous pores rather than in shrinkage affecting its external dimensions. However, even if the matrix would shrink in the radial direction, some of the polymerization mixture
Chem. Mater., Vol. 7, No. 4, 1995 715
Pore Formation in Macroporous Polymers
that is still found on the top of the rod up to a conversion of almost 70% would fill the space between the rod and the tube wall. Therefore, radial shrinkage, if any, cannot be deleterious to the function of the molded rod columns as a chromatographic separation medium. Effect of the Polymerization Rate on the Pore Formation. The reaction rate for a free-radical polymerization is not a simple function of temperature as the process consists of several consecutive reactions. Typically, the initiation step has the highest activation energy. Therefore, it is the initiation rate that is most temperature dependent. For example, the half-life of azobisisobutyronitrile is 37 h at a temperature of 55 “C but only 6 h at 70 “C. Table 1 and Figures 7 and 8 clearly document the effect of the polymerization temperature on the final pore size distribution. The lower the temperature, the larger the pores. This is quite consistent with the suggested mechanism of pore formation. The decomposition rate of the initiator, the number of growing radicals, and the overall polymerization rate are higher at higher temperature. Therefore, at higher temperatures, more nuclei are formed at once and they all compete to swell with the remaining monomers. The nucleation rate is faster than the swelling and the supply of monomers becomes exhausted after a short period of time. Because the number of nuclei that grow to the globular size is large but their sizes remain relatively small, the interstitial voids between smaller globules are also smaller. Therefore, the pore size distribution is shifted toward smaller pores for both the beads and the rods. However, the dynamic effects remain unchanged by the temperature and the difference in their porosity profiles persists. If the polymerization is carried out in a mold at a lower temperature, the reaction rate is slowed down considerably. The number of nuclei is smaller and therefore they have more time to swell. As a result, their growth is perpetuated by the polymerization that proceeds within them and they are able to sediment because space is available and the overall process is slow. This results in larger size globules and clusters. Therefore, the voids between the clusters are also larger and the pore size distribution is shifted toward larger pores.
Conclusion Our experimental results show that considerable differences exist between a classical suspension polymerization and the polymerization of the same mixture in a closed unstirred mold. The “bulk” polymerization in the presence of porogen results in macroporous materials containing very large pores with sizes that even exceed 1000 nm, at least 1 order of magnitude larger than the macropores of beads prepared by suspension polymerization. The mechanism of pore formation during the polymerization in a mold seems to be affected by the absence of both the interfacial tension and the dynamic forces that are typical for the suspension polymerization process. The pore volume and the pore size distribution of the molded materials are controlled by several variables. The reaction time provides control over the pore volume and the size of the largest pores because both the pore size and the pore volume depend on the conversion. At low conversions, the molded rods are more porous and they also include very large pores. Mixing of two nonsolvent porogens, such as cyclohexanol and dodecanol, can also be used for the control of sizes of the pores. A higher content of dodecanol leads to molded rods that have larger pores. The polymerization temperature is the most convenient variable to adjust the pore size distribution without even requiring a change in the composition of the reaction mixture. The use of a temperature gradient during the polymerization process can yield molded rods with a gradient of porosity. All of these variables represent very important tools for the engineering of the continuous macroporous materials that have t o contain large pores in order to make them easily permeable. The approach leads to materials useful in various areas such as very fast HPLC separations, catalytic processes, enzyme immobilization, water treatment, and solid phase extraction.
Acknowledgment. Support of this research by a grant of the National Institute of Health (GM 4836402) is gratefully acknowledged. This work also made use of MRL Central Facilities supported by the National Science Foundation under Award No. DMR-9121654. CM940490S
