Supercritical Fluid Engineering Science - American Chemical Society

---

Chapter 21

Light Scattering Study of Polymer Network Formation in a Supercritical Diluent

Downloaded by UNIV OF CALIFORNIA SAN DIEGO on January 30, 2017 | http://pubs.acs.org Publication Date: December 17, 1992 | doi: 10.1021/bk-1992-0514.ch021

J. Richard Elliott, Jr., and H. Michael Cheung Department of Chemical Engineering, The University of Akron, Akron, OH 44325-3906

Wehave undertaken a study of the benefits of using near- c r i t i c a l and supercritical fluids in synthesizing microcellular foams. In this study, we use dynamic light scattering to study the formation of the polymer network directly in near-critical and supercritical freon-22. The polymer system studied is ethylene- glycoldimethacrylate+methylmethacrylate. We have previously obtained microcellular foams by direct polymerization in this system and this study elucidates the gelation process by tracking the correlation length of the inhomogeneities as a function of reaction time from the monomeric state to the gel transition. A relatively little-known class of materials called aerogels has recently been the focus of several studies. The distinctive feature of these materials is^ that they combine extremely small pore sizes with remarkably low density, leading to some unique properties. For example, s i l i c a aerogel can have a density of 0.05 g/cm while maintaining pore sizes small enough that the bulk material is completely transparent; the pores are too small to refract visible light. Many applications can be envisioned for such materials, including thermally insulating windows, catalyst supports, and membranes (1). However, there are some technical obstacles. Most notably, the s i l i c a aerogels are extremely fragile. This has led to attempts to develop organic aerogels (2) also known as microcellular foams (1,3). Organic chemistry offers a broad range of possible molecular designs and polymerization mechanisms which provide alternatives to the s i l i c a aerogel process. A systematic approach for developing robust materials with unique and desirable properties may be obtained by combining an understanding of the fundamental processes occurring during the formation of aerogels with the ability to design and synthesize monomers tailored to a specific function. 3

0097-6156/93/0514-0271$06.00/0 © 1993 American Chemical Society

Kiran and Brennecke; Supercritical Fluid Engineering Science ACS Symposium Series; American Chemical Society: Washington, DC, 1992.

272

SUPERCRITICAL FLUID ENGINEERING SCIENCE

In previous studies, we have explored the feasibility of forming aerogels from methacrylates and the roles of polymerization diluent on the morphology of the resulting materials (3,4). Ve were able to synthesize microcellular foams based on the methacrylate chemistry, but the smallest cell size was about 1 urn with a density of about 0.4 g/cm . In general, we found that a strong solvent system favored smaller cell sizes and lower densities, and that mixed solvents held l i t t l e advantage for this particular system. Ve also showed that Freon-22 was a sufficiently strong solvent for this system to substitute for toluene as the polymerization diluent. Since the c r i t i c a l temperature of Freon-22 is 97 C, this finding implies that the polymer network may be formed and supercritically dried directly in the same vessel without exchanging solvents. More recent studies have led to microcellular foams based on epoxy chemistry where we have obtained strong materials with densities of 0.19 g/cm and pore sizes of 0.1 urn (5). In the present work, we seek to elucidate the mechanism of formation of the microporous structure as i t is formed in situ. For instance, scanning electron micrographs (SEM) of our microcellular foams (3,4) as well as those of Pekala and Stone (2) show that they are comprised of tiny beads of polymer from 0.01 to 1.0 urn in diameter. Ve would like to know whether these beads are formed during the polymerization or during the drying phase. Ve would also like to know how these beads form and grow so that we can think about ways of making them smaller while maintaining the ability to form a macroscopic material. To study this, we have implemented a high pressure scattering cell for dynamic light scattering. Dynamic light scattering detects the presence and diffusion coefficients of disperse inhomogeneities in a bulk fluid. Typically, diffusion coefficients can be measured for dispersions ranging in size from 1 nm to 10 urn. From the SEM's, we expect our primary particles to eventually grow to about 1 urn for the methacrylate system, so this method provides a useful tool for exploring some of the early stages of the foam formation and the nature of the polymerizing solution. Experienced readers may note that the mechanism of production of these aerogels is similar to the styrenic procedure for ion exchange resins. Indeed, a comparison of the SEM's of microcellular foams for to those of Swatling et a l . (6} shows a strong similarity in the morphology. These similarities prooably extend to the mechanism of pore formation as well. A number of theoretical analyses of network formation in ion exchange beads are available (7), however, dynamic light scattering analysis of the reacting system in situ has not been reported. It is difficult to imagine how such an analysis could be performed on the actual reacting system for ion exchange bead production, since this is generally an emulsion polymerization process. If one was to develop a model system which could be used to analyze the mechanisms occuring in ion exchange bead formation, i t might look very much like the one we have performed, but the conditions would not need to be near-critical. Therefore, our findings may have some bearing on ion exchange research, but we do not seek to emphasize this relationship at this time.

Downloaded by UNIV OF CALIFORNIA SAN DIEGO on January 30, 2017 | http://pubs.acs.org Publication Date: December 17, 1992 | doi: 10.1021/bk-1992-0514.ch021

3

3

Kiran and Brennecke; Supercritical Fluid Engineering Science ACS Symposium Series; American Chemical Society: Washington, DC, 1992.

21. ELLIOTT & CHEUNG

Polymer Network Formation

273

Downloaded by UNIV OF CALIFORNIA SAN DIEGO on January 30, 2017 | http://pubs.acs.org Publication Date: December 17, 1992 | doi: 10.1021/bk-1992-0514.ch021

Materials and Apparatus Methylmethacrylate (MMA) and ethylglycoldimethacrylate (EGDMA) were used as the comonomers in this study. Freon-22 was used as the diluent. Monomers were washed of their inhibitors and dried using molecular seives by conventional procedures. Tertbutylperoxypivalate was used as the free radical initiator. To prepare a typical reaction mixture, MMA and EGDMA were added to a 40 ml high pressure "bomb" in the proportions of 12 ml of MMA and 8 ml of EGDMA (Figure 1). 20 ul of TBPP was then added and the bomb was attached to tubing with an open-shut valve. The bomb and contents were cooled in ice and connected to a supply of Freon-22 kept at room temperature. The Freon valves were opened and Freon allowed to condense into the bomb for 10-20 minutes. The valves were closed, the Freon tank disconnected, and the bomb and contents were refrigerated until ready for use. To charge the high pressure light scattering cell (Figure 1) , the bomb was raised to room temperature and the scattering cell was cooled to 5°C. The height of mercury in the c e l l was adjusted to keep reactive solution out of the cell's mercury reservoir while keeping the mercury well below the light path. The bomb was inverted and connected to the scattering cell and the monomer+diluent solution was allowed to condense into the cell for 10-20 minutes. Valves were sealed and capped and the pressure was adjusted to about 400 psig at room temperature. The temperature in the cell was raised to the designated value by heating tape connected to an Omega temperature controller. Time-resolved scattering intensities were measured at 90° angle using an Thorn-EMI photomultiplier tube and a Brookhaven Instruments amplifier/discriminator integral with the phototube housing (Figure 2) . They were analyzed using a Brookhaven Instruments Corporation BI-2030AT correlator and IBM PC/AT computer using Brookhaven Instruments Non-Negative Least Squares (NNLS) fitting software to estimate the particle size distribution as a function of reaction time. This scattering system has been described previously (8).

Results and Discussion Results of the study are shown in Table I and Figures 3 and 4. There are at least three possible mechanisms by which the monomer solution may evolve to a macroscopic bulk material. It is helpful in discussing our results to describe these in advance. First, the system may consist of steadily growing primary particles which grow until they f i l l the entire solution. Then the particle size histograms would show a broad polydisperse population of particles with the peak slowly moving to higher sizes. Second, the system could grow in stages whereby small particles are generated then flocculated into large globules that are eventually too big to grow, then a new population of small particles evolves and begins to flocculate. The particle size distributions in that case would appear as waves of particle size peaks when considered as a function of time. A third possible scenario would be that the primary particles grow to a certain size then stop growing until the concentration of particles becomes so great that the particles percolate at a gelation point and convert from disperse sols to the

Kiran and Brennecke; Supercritical Fluid Engineering Science ACS Symposium Series; American Chemical Society: Washington, DC, 1992.

SUPERCRITICAL FLUID ENGINEERING SCIENCE

Downloaded by UNIV OF CALIFORNIA SAN DIEGO on January 30, 2017 | http://pubs.acs.org Publication Date: December 17, 1992 | doi: 10.1021/bk-1992-0514.ch021

274

Figure 1. Experimental apparatus for near-critical polymerization.

Kiran and Brennecke; Supercritical Fluid Engineering Science ACS Symposium Series; American Chemical Society: Washington, DC, 1992.

ELLIOTT & CHEUNG

Polymer Network Formation

Downloaded by UNIV OF CALIFORNIA SAN DIEGO on January 30, 2017 | http://pubs.acs.org Publication Date: December 17, 1992 | doi: 10.1021/bk-1992-0514.ch021

21.

Figure 2. Schematic representation of the arrangement of the light scattering optics.

Kiran and Brennecke; Supercritical Fluid Engineering Science ACS Symposium Series; American Chemical Society: Washington, DC, 1992.

275

276

SUPERCRITICAL FLUID ENGINEERING SCIENCE

Downloaded by UNIV OF CALIFORNIA SAN DIEGO on January 30, 2017 | http://pubs.acs.org Publication Date: December 17, 1992 | doi: 10.1021/bk-1992-0514.ch021

750

t = 16 min 2000 h ν

·

ν

· . ^ ν .

%

......

1800 36000 t = 26 min 34000

32000 [

200000

180000 h

160000 h

140000 0

200

400

600

Sample delay, m i c r o s e c o n d s Figure 3. Correlograms for the 10 volX EGDMA + 40 vol7. MHA + 50 vol7o Freon-22 systems at 90° angle and conditions given in Table I.

Kiran and Brennecke; Supercritical Fluid Engineering Science ACS Symposium Series; American Chemical Society: Washington, DC, 1992.

Downloaded by UNIV OF CALIFORNIA SAN DIEGO on January 30, 2017 | http://pubs.acs.org Publication Date: December 17, 1992 | doi: 10.1021/bk-1992-0514.ch021

21. ELLIOTT & CHEUNG

277

Polymer Network Formation

/ d

I Π

1

5

58 5Θ8 Diameter nm

5888

1

5

58 588 Diameter nm

5888

Figure 4. P a r t i c l e size d i s t r i b u t i o n s from dynamic l i g h t scattering at the conditions given i n Table I .

macroscopic material i n a very short time. In that case, we would expect very monodisperse populations that do not change i n s i z e but the scattering count changes as more scatterers evolve u n t i l the solution g e l s .

Table I . Results f o r l i g h t scattering studies of 10 v o l 7. EGDMA + 40 vol7. MMA + 50 vol% Freon-22 polymerizing solution LS# Initial 00 02 06 10 12 13 15 16 17

T(°C) 4 55 73 81 89 93 95 93 95 95

Time (min) 0 14 16 20 24 26 28 30 31 32

P(psig) 500 850 1125 1250 1400 1410 1450 1460 1460 1460

Rate(MHz) 4 4 4 4+ 4+ 5 9 20 30 60

Signal/Noise 0.31 0.32 0.32 0.28 0.25 0.28 0.10 0.14 0.25 0.26

Figure 3 shows that the most p l a u s i b l e scenario f o r free r a d i c a l polymerization i s the t h i r d scenario described above. The correlagrams i n Figure 3 show the autocorrelation function vs time f o r the low EGDMA system as i t polymerizes under s u p e r c r i t i c a l conditions. The l a t e r correlagrams c l e a r l y show the c h a r a c t e r i s t i c exponential decay which indicates the presence of small p a r t i c l e s . Considering

Kiran and Brennecke; Supercritical Fluid Engineering Science ACS Symposium Series; American Chemical Society: Washington, DC, 1992.

278

SUPERCRITICAL FLUID ENGINEERING SCIENCE 5

that the sampling time was 10" seconds for each of these correlagrams, i t is clear that, qualitatively, small particles dominate the scattering at later times. Following the correlagrams back in time, the signal from the small particles becomes progressively weaker. Table I shows that the photon count rate is very low at the early stages and this is because there are no scatterers present. The correlagram at the i n i t i a l condition shows a flat response as we would expect at 10" seconds when only dust is present. The correla­ grams between 14 minutes to 24 minutes are much "noiser" than those earlier and later because the number of scatterers is just enough to give a signal but not enough to give a clear signal. Vhat can be qualitatively observed, however, is that the rate of decay is appar­ ently consistent with what we might expect from a small population of particles of the same size as those in the later correlagrams. It is possible to argue that there might be other particle distributions present, but that we do not observe them because we only observed at a single sampling time. However, previous studies were performed over a series of sampling times from 1 μsec to 5000 /*sec and the results were always consistent with the description above. Various averaging times were also considered before deciding that a one minute averaging time provided a reasonable compromise between getting good light scattering results for short times and observing the polymerization on a reasonable time scale so that we were not averaging out too much in the way of particle size or number density changes. Ve also collected extensive data through multiple runs of the same systems to verify the reproducibility of our observations. Figure 3 shows a series of correlagrams for a system composed of 10 vol7. E6DMA + 40 vol7. IMA + 50 νο17· Freon-22 where the correlagrams were collected at different sampling times. Clearly, one particle distribution can describe a l l of the observations. The correlagrams vs time for the 20 νο17· EGDMA solution were basically similar to those of the 10 7· EGDMA solution for each experimental run. To make more quantitative arguments with regard to the particle distributions i t is necessary to make some approximations about the physical properties of the solution. In particular, the viscosity of the polymerizing solution must be known and the relative refractive indicices of the solution and the primary particles must be known to infer particle sizes from what essentially amounts to data on the diffusion coefficient. These physical properties are difficult to measure or estimate with any degree of r e l i a b l i l i t y so we have set these values at a viscosity of 0.28 cp and a refractive index of 1.59:1.33 (particle: solvent). This means that the exact values of the reportea particle sizes may be challenged but the relative mag­ nitudes should be consistent throughout our analysis. There are several options that may be considered for transforming the autocor­ relation functions into particle distributions. Perhaps the best known is the cumulant expansion technique (9), but i t does not pro­ vide detailed information about the particle size distribution. For example, i t cannot identify multimodal particle size distributions. A commonly used technique which provides more detailed information is the NNLS technique (10). In Figure 4, we used the NNLS technique because we were interested in exploring as much detail a possible. The primary particles appear at about 20 and 30 nm then they grow in

Downloaded by UNIV OF CALIFORNIA SAN DIEGO on January 30, 2017 | http://pubs.acs.org Publication Date: December 17, 1992 | doi: 10.1021/bk-1992-0514.ch021

5

Kiran and Brennecke; Supercritical Fluid Engineering Science ACS Symposium Series; American Chemical Society: Washington, DC, 1992.

Downloaded by UNIV OF CALIFORNIA SAN DIEGO on January 30, 2017 | http://pubs.acs.org Publication Date: December 17, 1992 | doi: 10.1021/bk-1992-0514.ch021

21. ELLIOTT & CHEUNG

Polymer Network Formation

279

population but not size u n t i l , finally, the solution gels when a l l the particles have grown to about 30 nm. There is also some appearance of 1000 and 10000 nm particles which show up in Figure 4. It seems likely that these particles represent dust i n i t i a l l y present in the sample. These results indicate that certain particle sizes are more favored than others and that the polymerization occurs by rapidly populating these favored sizes until three-dimensional connectivity occurs (percolation). This represents a kind of combination between the second and third scenarios described above. In order to make assertions about the particle sizes, we must also neglect variations in concentration dependent terms that relate to hydrodynamic and thermodynamic interactions. These kinds of interactions can be significant in many cases (11), but we do not seem to see significant effects in our system. The significance of the interactions in an aqueous phase tends to be related to the relatively high dielectric constant of aqueous solutions. Since our continuous phase has a relatively low dielectric constant, one might expect that electrostatic interactions would be small and hard sphere interactions would dominate. Corrections for hard sphere interactions are approximately 157. at a volume fraction of 107.. Since the particles we see are small, their number density can be relatively high before these corrections become significant (11). Given that the particles seem to appear at a certain size then cease growing, i t may be of interest to speculate about the mechanism by which this occurs. First, the appearance of the primary particles at about 20-30 nm but not at smaller sizes with subsequent growth to that size is likely a consequence of the free-radical polymerization. In free-radical polymerization, individual polymer chains grow to their f u l l length a l l at once until their growth is terminated. Then growth of another chain is initiated. This would mean that relatively few species would exist as medium-sized particles at any given instant. That population would not be large enough to generate a significant scattering signal. The population which does appear is fairly polydisperse and that polydispersity could be explained by variations in the molecular weight that one would expect. But why do these polymer molecules segregrate themselves from solution to such an extent that they generate a significant scattering signal? What is the role of crosslinking? These are very difficult questions to answer at this time. Some recent thermodynamic findings suggest that interaction sites on chain molecules show a striking "cooperative attraction" for each other (12). It is possible that long chains in solution could wrap around in a way that this effect is expressed even for sites that are actually on the same chain. Crosslinking could increase the molecular weight in such a way that cooperative attraction effects would be enhanced. Thus i t is possible to imagine ways in which these inhomogeneities could be generated, but i t may require considerable effort over a number of years to develop a complete understanding of the mechanism for production of these primary particles. SEM analyses of the final dry products (4) indicate that these same particles are preserved in the macroscopic material even through the supercritical drying process. Thus the drying process seems to have l i t t l e adverse impact on the morphology of the macroscopic material. This means that research effort should be directed at

Kiran and Brennecke; Supercritical Fluid Engineering Science ACS Symposium Series; American Chemical Society: Washington, DC, 1992.

Downloaded by UNIV OF CALIFORNIA SAN DIEGO on January 30, 2017 | http://pubs.acs.org Publication Date: December 17, 1992 | doi: 10.1021/bk-1992-0514.ch021

280

SUPERCRITICAL FLUID ENGINEERING SCIENCE

understanding the causes of certain particle sizes being favored and toward changing the chemistry such that the favored particle sizes are as small and uniform as possible. Other factors of interest in future research would include developing non-spherical primary particles and manipulating the aspect ratio. Non-sphericity should have an impact on the percolation behavior. Extensions of the experimental technique would include adding more optical signal processors at multiple angles, using probe particles to obtain viscosity data for the actual solution, eliminating the need for mercury, and adding depolarized scattering capability to measure the rotational diffusion coefficient. These additions would help to provide data which could refine the accuracy of the technique and provide detailed molecular insights into the polymerization/gelation process.

Literature Cited [1] Aubert, J . H . ; Sylvester, R.P., CHEMTECH, 1991, 21,234. [2] Pekala, R.E.; Stone, V. ACS Preprints, 1988, 29(1),204. [3] E l l i o t t , J r . , J . R . ; Srinivasan, G . ; Akhaury, R. Polymer Communications, 1991, 32,10. [4] Srinivasan, G . ; E l l i o t t , Jr., J.R. Ind. Eng. Chem. Res., 1992, in press. [5] Dhanuka, M.S. Thesis, The University of Akron, January (1992). [6] Swatling, D.K.; Hanson, J . A . ; Thomas, D.A.; Sperling, L.H. J. App. Poly. Sci., 1981, 26,591. [7] Seidl, J.; Malinsky, J.; Dusek, K . ; Heitz, V. Advances in Poly. Sci., 1967, 5,113. [8] Sasthav, M. and Cheung, H.M. Langmuir 1991, 7,1378. ['9] Koppel, D.E. J. Chem. Phys., 1972, 57,4814. [10] Grabowski, E . F . ; Morrison, I.D.; in Measurement of Suspended Particles

by

Quasielastic

Light

Scattering,

B.E.

Dahneke,

Ed.;

Wiley, New York (1983). [11] Cheung, H.M.; Qutubuddin, S.; Edwards, R.V.; Mann, Jr., J . A . Langmuir, 1987, 1,333. [12] Vasudevan, V.J.; E l l i o t t , Jr., J.R. Mol. Phys., 1992,75, 443. RECEIVED June 12, 1992

Kiran and Brennecke; Supercritical Fluid Engineering Science ACS Symposium Series; American Chemical Society: Washington, DC, 1992.