Free Radical Polymerization of Methyl Methacrylate at High Temperatures PIERRE E. BAILLAGOU and DAVID S, SOONG* Department of Chemical Engineering University of California at Berkeley Berkeley, California 94720 Freecradical polymerization of methyl methacrylate in a tubular reactor has been condueted at above-T, temperatures. AA salient feature of these experiments is the very efficent control of reactor temperature by vapor-liqutd equilibrium of the polymerizing mixture via monomer evaporation. The sys- tem pressure thus provides a powerful control variable, restict- ing the temperature in the entire reactor by changing the ‘monomer evaporation rate. In the range of our experimental conditions, the temperature and pressure in the reactor follow the Antoine equation closely. High temperature runs also re luce the length requirement of the reactor. However, molee- ular weight averages ofthe produets are not impressive, unless slow-burning. initiators are used. Modeling of above-T, reae- tionshas been attempted at two-levels of sophistication. A plug- flow model gives predictions in good agreement with our ‘experimental temperatures and conversion data. The predicted ‘molecular weights are also consistent with the experimentally observed values. However, the more elaborate sheokinetic ‘model suggests that the superficial agreement between model ancl experiment is due to initiator burn-out, which limits the final conversion to within 40 percent. The liquid layer next to the reactor wall ean never be so viscous as to form a stagnant deposit, duc to this conversion limitation. The velocity profiles are thus not very much distorted, and a plug-flow model is adequate, With a slow-burning initiator and a sufficiently long reactor, skewing of velocity profile and reactor channeling will eventually emerge. Hence, the theokinetie model must be ‘evoked to model the system under such conditions INTRODUCTION physical state of the species present in polymerizing ‘onventional polymerization processes to form Coirivinethyteetharylate) (PRIMA) often rant relatively low temperatures, ¢-g., below the glass- transition temperature (T,Jof the final product. Here, we propose a new operating strategy, i... above-T, polymerization. Operating above the ass-transition temperature of PMMA drastically changes the reaction characteristics and creates a new set of rules governing the polymerization pro ess. The primary advantage of above-T, polymeri- zation is reduced viscosity of the solution, since high medium viscosities are responsible for the mixing problems encountered even at low conver- sions in low temperature experiments, However, such u change in temperature cannot be made with- out affecting the kinetics of the reaction and the “ohn creole tobe arene ae Valsamis and Biesenberger (1) reported that, for a given ntiator, high temperature polymerizations of styrene occur at a faster rate and produce lower molecular weights than at low temperatures. These observations were also predicted by a model cently developed in our laboratory (2), Increasing reaction temperature thus creates adverse conse- quences on product quality. However, an important difference between the above-T, polymerization of styrene and MMA arises from the relative values of the glass-transition temperature of the polymer and the boiling point of the monomer, T, of polystyrene (1j,= 105°C) is somewhat lower than that of PMMA (7h = 114°C), and styrene boils at 145°C whereas the boiling point of MMA is 100°C, Itis therefore possible to produce polystyrene above its glass- transition temperature, but below styrene boiling point at 1 atm. In contrast, MMA boils before the POLYMER ENGINEERING AND SCIENCE, MARCH, 1985, Vol. 25, No.4Free Radical Polymerization of Methyl Methacrylate glass-transition temperature of PMMA is reached. Ttis thus impossible to conduct above-T, polymer- ization of MMA at 1 atm, This constraint is in fact a very important consideration for above-T, polym- terization reactor design, In the case of MMA, the polymerizer must be pressurized to suppress the evaporation of the monomer. The pressure, which {was not a parameter of primary importance at low temperatures, now plays a critical role in the con- trol of above:T, polymerizations. The upper limit of the proposed above-T, Polymerization is nate rally the ceiling temperature of the reaction, which for the case of PMMA Is 164°C (3). We expect the reverse (depolymerization) reaction to become sg nificant as this temperature is approached. Hence, ‘most reactions will be operated well below this Timit, and the products analyzed to ensure agree- ment between experimental results and model pre- 30 60 30 TIME (MIN) In contrast to runs with Kenies mixers, the control of empty tubular reactors is usually not as tight, but severe problems of temperature control or chan- neling are never encountered. During Run 19 (re- producing the conditions of Run 21) a conversion fs high us 65 percent was achieved under control ‘with empty pipes. The cause for the loss of temper- ature control in Run 21 is unclear Boi The hypothesis of thermodynamic vapor-liquid equilibrium in the reactor can be confirmed by tleulating the boiling point of the reac ig mitre ‘under different pressures. The presence of polymer the liquid phase will be ignored (as its mole fraction is usually small), and the calculation as- sumes that the monomer boiling point in the pure state is approximately equal to that in the mixture, No correlations ean be found in literature for the vapor-liquid equilibrium of methyl methacrylate. However, the parameters for the Antoine equation are given for ethyl acrylate by Reid, et af. (4). Tt seems reasonable to assume that the agreement between the vapor-liguid equilibrium of the two acrylie monomers of identical molecular weight un- der reduced pressures reported by Luskin (5) also holds above atmospheric pressure, The boiling point at different pressures of methyl methaerylate ‘was calculated with the Antoine equation. The fol- Towing parameters are recommended (4): In [P(mm Hg)} = A ~ BT(°K) + C with a) A= 16.0890 B=2074.94 C= -58.15 ‘This equation is valid between 274 and 409°K. Experimental and calculated boiling points, are plotted versus the reactor pressure in Fig. 4. ‘The experimental data show perfect agreement with the theoretical curve. It is therefore reasonable to as- sume that the reactor is maintained in thermody- namie equilibrium during the polymerization reac~ 1g Point Calculations tion. OF course, the equilibrium only holds if enough monomer is present in the reactor. The rapid evaporation of monomer to ensure pressure. temperature equilibrium cannot be maintained be- yond a certain conversion. This eritical conversion Is difficult to evaluate theoretically, because it not only requires the calculation of the polymer activity coefficient, but also because it depends on the mass transfer rates in the vapor and liquid phases. The mass transfer process is strongly influenced by the configuration of the two-phase reactor and layer thickness (ininimum physical dimension) of the phases. Since the reactor temperature is influenced by: the competing rates of heat generation by the polymerization reaction and heat removal by evap- oration, the mass transfer rates need to be much higher than the reaction rate in order for thermo. dynamic equilibrium to be preserved. Comparison of Mode! Predictions with Experimental Data ‘The data obtained in our above-T, experiments are compared with the predictions of a plug flow model, which accounts for the effects of high tem. peratures (Table 2). The details of the model are tablished elsewhere (2), and only its highlights will he discussed in the second part of the paper. ‘The temperature profiles of the two steady states obtained during Run 13 with and without chemical initiation are compared with the model predictions in Fig. 5. A value of U = 15 Biu/lar-ft.F for the ‘overall heat transfer coefficient through the wall of the Kenies mixers is found to fit the experimental data optimally. The temperature profiles obtained in empty tubes (Run 16) are not quite as wel predicted by the plug flow simulation (Fig, 6) with U= 10 Btu/hr-ft-"F. This may be due to the accu: mulation of polymer around the thermocouples (Le., stagguation points), which is found to occur 160 FEED MIXER - RUN 16 30 60 30 TIME (MIN) ure, and concersion hstores during tation experiment in hollow tubes (as Hig, 8. Temperature, p igh temperature poly 16). POLYMER ENGINEERING AND SCIENCE, MARCH, 1985, Vol 25, No.4 | | {Free Radical Polymerization of Methyl Methacrylate 179 179 160 Fire ~ RUN 18 160] PIPE ~ RUN 16 © 450) © 180) aso gusg ~ 150) ~ 130) te INITIATOR wy 20] |? 26 TZATOR w 120 Eno Eno E E & 100 & 109] = oo = a0 F go] Rowe aT 10 cH FROM INLET © go] PROBE AT 55 CH FROW INLET 70 7 0 30 = 30 Ey Fa 30 TIME (MIN) TIME (MIN) 170 60 160 Pure ~ RUN 16 PIPE ~ RUN 16 © 450! 8 gee B40 5 iso £ TP aNsTrRTOR & eo W120 ys0 oo {teeteete Ene 3 oe E 00 ie & at) B go] Rope a 25 cH FRON INLET 70 o 0 x 3 % 0 % 0 30 TIME (HIN) "Tame (aN) 170 1 10 PIPE ~ RUN 16 va{ OUTLET ~ RUN 16 5 y50] Tal SRHPLES AT 85 CH FROM INLET gue z7 = 130) t INITIATOR a4 120 ge 2 ud Ze & 100! ous & oo} 2 Bool RoBe at 40 cH FROM INLET : . ceteee’ 20 9 o 0 30 0 Fy 3 30 TIME (MIN) ‘POLYMER ENGINEERING AND SCIENCE, MARCH, 1985, Vol. 25, No. 4 30 TIME (MIND aTPierre E. Baillagou and David S, Soong tt 6 Moximum steady state temperature 60} ‘observed in 150} © Empty Tubes ted fe Kenics Mixers 150} SEmoly Tubes ond © ad 3 Keenice Mixers g g i £ E aa} § 5 2 4 Fad : ‘oq & 130) > oo l20} 10} too) o 10 2 30 40 50 Pressure (psig) Pig. 4, MMA eapor-tigud equilibrium. Solid curve i caleulated fow the Antoine eyuation and the points are steady state doa, ‘Table 2. Comparison Experimental Data With Modol rediations. font Th x MM MH #P* exp cale'd exp cale’d oxp _cale'd exp cale'd 8 25 a7 1987 0.05 0.0004 250000 ra4000 30 20 46 25 187 138-7 0.04 0.0004 188000 119000 25 20 48 40 150 180.5 0.10 0.0008 142000 @6000 23 20 13 30 140 199.7 0.04 0.0004 92000 110000 22 20 45 18 190 1288 0.56040 19000 18000 21 23 49 21193 1923 083088 12000 18000 27 26 16 92145 1438 0.57 030 $000 1000 39 26 48 45 155 1645 031 020 6000 9000 108 28 48. 45 155 1848 040.025 11000 12000 79 26 21 20145 1312 098.088 11000 17000 27 24 23 20192 1312 0340.82 26000 92000 1.8 20 more frequently in empty tubes. In both cases, the temperature abruptly levels off wt the boiling point, which is aceuratel tion. Our conversion data are generally higher than predicted because of the large error due to the evaporation of monomer at the reactor outlet. Model predictions give probably a better estimate of the actual conversion, The experimental conver- sion would probably be more realistically computed by making an isenthalpic flash calculation at the outlet valve, (It is important to note that by ope: ating at high temperature and pressure, the evap- oration of the residual monomer and solvent rapidly occurs as soon as the mixture leaves the reactor; this facilitates the separation processes required for the purification of the polymer.) Much higher con- versions than predicted are obtained experimen- ly predicted by the Antoi equa: 0 8a Axo! Ostonce fem) Stay slate temperatre profiles for 4 igh temperate polimertztionexperinnt tn Kener msers (Bm 13) 120} Temperate (0) TARTS an, (hon “Ores, P2009 rr a a a) Axia! Distance (em) Pig, 6. Steauy stave temperature profiles for high temperate valmerization expernien holo pipes (li 18), tally for the thermally initiated experiments, The model probably underestimates the rate of thermal initiation, whieh may be enhanced by the presence of impurities. However, the experimental conver sions are still too low to be of industrial interest The order of magnitude of the molecular weight obtained in chemically initiated experiments is cor rectly predieted by the model, However, the model fives a higher molecular weight sensitivity to pree sure than experimentally observed. The number average molecular weight is around 10" when AIBN used to initiate the polymerization and in the neighborhood of 3 x 10* with benzoyl peroxide When no chemical initiator is employed, molecular weights as high as 2.5 X 10° ean be obtained With a pressure Tower than 40 psig, the thermo- dynamic control of the temperature ensures almost isothermal conditions throughout the reactor. This partly explains why the polydispersity index of the observed molecular weight distribution is always less than 8. At higher pressures, the thermodynamic control is probably less absolute, and the regions close to the reactor walls are presumably at higher temperatures, This would explain the experimen- tally observed polydispersity on the order of 10 for Run 18, Benzoyl peroxide initiated polymerization POLYMER ENGINEERING AND SCIENCE, MARCH, 1985, Vol 25, No.4Free Radical Polymerization of Methyl Methacrylate can provide extremely narrow molecular weight distebution with a polydispersity as low as 1.85. Quite acceptable polydispersities (from 2 to 3) are also obtained by thermal initiation. CONCLUSIONS It was shown experimentally that Kenies mixers and empty tubular reactors are suitable for the above-T,, polymerization of methyl methacrylate. The vapor-liquid equilibrium of MMA can be used to fix the system temperature via pressure regula- tion of the reactor. This provides a very efficient control of the temperature in spite of the high rates of polymerization, The problems of thermal run- away and ignition can be entirely avoided by this method. The control of the temperature by vapor- ization of monomer is not restricted to tubular reactors. Tt can be used in other polymerization reactors (stirred tanks or extruders). The same idea of using the latent heat of vaporization of the mon- omer to remove the heat of polymerization is com- monly applied in many industrial processes (6) in a slightly different manner. The heat of polymeriza- on is removed by the latent heat of vaporization in atmospheric pressure reactors equipped with condensers. In the CSTR experiments reported by Schmidt and Ray (7) quasi-isothermal conditions have been achieved by the vapor-liquid equilibrium of the polymerizing mixture. Since AIBN decomposes rapidly at above-T, tem- peratures, high conversions (40 to 60 percent) can be attained within short residence times. Unfortu- nately, the molecular weight obtained under such conditions (around 10") is much lower than com- mercially desirable, When no initiator is used, the molecular weight is quite satisfactory (10° to 2.5 X 10°), but the conversions obtained at the end of our short reactor are too small to be of industrial inter- est. Between these two extremes lie many choices of “slow-burning initiators", which do not decom- pose as fast even at high temperatures. The opera- bility and performances of such reactors at higher conversions will not be investigated experimentally because the low rates of polymerization would re- quire long vertical reactors, which are hard to main- tain at high temperatures within the present con- fines of our laboratory. The models previously de- veloped will be used to evaluate the feasibility of producing high molecular weight polymers at high Temperatures under more satisfactory operating conditions, IL._A First Approach to the Modeling of High Temperature Polymerization INTRODUCTION ‘The experiments described in Section I showed that low molecular weight polymers are produced in a controlled manner up to about 40 percent conversion in tubular reactors operating at high temperatures, Important advantages over the con- Ventional low temperature MMA. polymerization processes are attributable to the efficient tempera- Lure control by the rapid vapor-liquid equilibrium, POLYMER ENGINEERING AND SCIENCE, MARCH, 1985, Vo. 25, No. 4 and more uniform velocity profiles and lower pres- sure drops resulting from the reduced viseosty of the polymerizing mixture. The use of vapor-liquid equilibrium to control the temperature can be ap- plied to any other pressurized reactor. It is there fore important to examine the kinetics of MMA polymerization above PMMA glass-transition tem- perature, The kinetic equations wil fist be solved for a plug flow reactor, but the conclusion drawn is, also valid for an equivalent batch reactor with the same residence time. These simple calculations will help to explore the feasibility of producing com- mercial PMMA at high temperatures, The same type of study was done by Valsamis and Biesenber- ger (1) for high temperature styrene polymeriza- tions, However, as explained in Part T, styrene po- lymerization does not offer a combination of phys- ical constants (monomer boiling point and polymer glass transition temperature) as favorable as MMA for temperature control by the vapor-liquid equilib- Hium of the monomer. This control strategy was thus not investigated in their study and their reactor ‘was sufficiently pressurized to avoid any evapora- tion of styrene. The operability of a real tubular reactor at reasonable conversions (at least 20 pereent) will again be discussed on the basis of the predictions of a recently developed rheokinetic model (2) MODEL DEVELOPMENT Modeling an above-T, MMA polymerizer ean be complicated because of the presence of two phases sumptions will be made in order to apply our recent model (2) to high temperature polym- erizations with minimal modifications, Thermody- namic equilibrium will be assumed, i.e., mass trans- fer between the two phases (evaporation of mon- omer) aceurs at much faster rates than the polym- erization reaction. The description of the phase equilibrium will be simplified by assuming that the boiling point of the polymerizing mixture follows the Antoine equation (Eq. I) for pure MMA. Our previous energy balance will be used as far as the temperature is maintained below the calculated boiling point of the mixture. (The very small axial pressure drop observed experimentally does not affect the average pressure significantly in the re- actor.) Once the boiling point is reached, the tern perature will be maintained constant at this level. Note that this calculation holds only if enough mon- ‘omer remains in the reaction mixture to facilitate evaporation and maintenance of the reactor tem- perature at the MMA boiling point, It was shown experimentally in Section I that this approach is sutable as a frst analysis of the reactor behavior Itcorreetly describes the temperature profiles until a fairly high critical conversion, where the reactor temperature drifts from the MMA. vapor-liquid equllibrium temperature, Therma initiation will be incorporated in. the polymerization mechanism, since its rate becomes significant at high tempera- tures, The same assumptions are also made in the theokinetie simulation, which provides more real- 239Pierre B, Baillagou and David 8, Soong, istie radial profiles. The conservation equations re- main unchanged from the low temperature models Q). A list of model equations appeats elsewhere (2) and will thas not be reproduced here. Only modli- fications of this basic model will be presented. The radical generation terms, rederived to account for thermal initiation, are listed in Table 3, which also fives the Arrhenius temperature dependences of the different initiation rate constants. RESULTS AND DISCUSSION Above-T;, Polymerization Kin The plug flow model is used to investigate the ideal polymerizer performance above T,, evaluate {n terms of conversion and molecular weight distr bution parameters. The results obtained for AIBN initiated polymerizations are summarized in Fig. 7. ‘The curves correspond to products obtained at the outlet of a T-m long, plug flow isothermal reactor (or in a bateh reactor at the equivalent residence time) at the given pressure and initiator concentra~ tion. The horizontal axis is scaled in terms of pres- sure, as it is the most efficient control variable, The corresponding temperatures (ealeulated from the “Antoine equation) are given at the top of each plot "To understand the trends exhibited in Fig, 7, tis important to note that initiator decomposition is tntirely independent of the concentration of the other species present in the polymerizing medium ‘The initiator concentration decays exponentially, no matter what happens during an isothermal po- Tymerization reaction, Around 80°C, the initiator decomposition occurs very slowly and [I] is actually Inereasing by virtue of the volume contraction, Above the PMMA. glass-transition temperature, ‘AIBN decomposes much faster, whereas the rate of monomer consumption by propagation does: not ease as much, Therefore, even if the temper ture is held constant (no ignition or ran away 0 curs), the polymerization may be dead ended, ie. ties Temperatre (0) 120 140 150160 eT ‘Tenperatue (0) 120,140 180 169 Number Average Molestor Weight O15 25 30-4080 G0 Prastre (pig) ——1nany = ons = =a coro iy = oo r= 02 rosires ee o's rin hy MAMA vaparsliquid equilibrivn, ” fs prematurely terminated by depletion of initiator hefore complete conversion of monomer. Since ky is a function of temperature, the time seale of initiator decomposition is a degree of freedom which can be independently adjusted by the pres- sure. Maximal conversion is reached when the ii tiator is entirely decomposed just at the reactor ‘Table 3. Modifications Intoduced to the Model to Account for High Temperature Thermal nition, ‘oaiied rate af generation terms a 2th + HAM ~ Ban? tall + hal ME + KMD + (RLM + hfSTKD9 ~~ Hen ay © lk) + Ha? Bf MIDy + Re) (kl RfS]1D0 ~ Ho) ~ Kole 21.6 exp(-22 kealmoleAT(EK) (ate of MMA thermal ination, Wallg and Biggs, LACS. 68 1141, 1846) 32 x 10" exp-90.68 kcalmole/AT (°K) (ato of AIBN decomposition, Tobolsky and Baysal J. Polym. Sc, 14, 474, 1959) 101 x 10" expi(-30.0 kealfmoley (°K) (ate of Benzoy! peroxide decomposition, Tobolsky and Baysa, J. Pym. 3. 14, 47, erin) = 43 x 10" oxp\(-97.0 kealimoleyRT (°K) (ate of OTEP decomposition, Valsamis and Blesenberger,AICNE Symp Series 160,72 18,1978) For temperatures lass than or equal to MMA boing point 2074.96 T61 — WIST-7AP (pag) + 780) TO) = 4+ 88.15 ~ 278.15 {Antoine equation wih the parameters suggested by Red, Prausitz, and Sherwood, “The Propartos of Gases and Leu” Sr Ed, McGraw Hil, Now York, 1877) 240 POLYMER ENGINEERING AND SCIENCE, MARCH, 1985, Vl 25, No.4Free Radical Polymerization of Methyl Methacrylate outlet. Further inerease of pressure causes an ear~ lier exhaustion of initiator and therefore a lower final conversion, Product molecular weight decreases monotoni- cally with the reactor pressure, because the rate of initiation inereases with temperature, Even at low pressures, the molecular weight is quite low, gen- erally unaeeeptable as a commercial product. The increase of polydispersity with pressure can be un- derstood from the concept of kinetie chain length. In isothermal conditions and at low conversions (absence of gel effect), the resulting chain length can be affected only by changing either the mon- ‘omer or the initiator concentration, When the po- Iymerization dead ends, the initiator concentration varies from [Ilr (typically around 0.0129 mole/L) toa value around 10°* mole/L over the reactor Tength, Such a change in the initiator concentration significantly affects the rate of initiation and there- fore different molecular weight species are pro- duced along the reactor. The global molecular weight distribution of the product from the entire reactor is therefore broader when the reactor con- centration undergoes severe changes. This explains the sudden increase of the polydispersity just after the transition to dead ended polymerization. Fur ther increases in pressure still decrease the initiator concentration at the outlet but do not affect the polydispersity very much, The very few chains pro- ‘duced under extremely low initiator concentrations {loss than 10° mole/L) do not significantly contib- ute to the final molecular weight distribution of the product [ip ucts like a sealing factor for the initiator concentration exponential decay. The curves ob- tained by varying the pressure for different [Ilr share the same shape and essential features. The conversion curve is uniformly shifted towards higher values with increasing [I[p, but the transition to dead ending occurs at about the same pressure, The molecular weight decreases with the rate of initiation, and consequently with initial AIBN con- centration. Polydispersity variations with pressure are amplified by increasing [Z], but again the shape of the curves remains unchanged. Benzoyl pero AIBN. Under identical conditions, it yields lower conversions but higher cumulative molecular Weight averages and narrower molecular weight distribution (Fig. 8). The transition to dead ended polymerization occurs at slightly higher pressures, Benzoyl peroxide still does not yield satisfactory molecular weight distributions at high tempera- tures. ‘DIBP (di-tert-butyl-peroxide) initiated polym- cerization exhibits drastically different features. The much slower rate of initiation results in lower con- Versions but much better molecular weight distr- butions. A longer reactor was considered (6m long) in the calculation to compensate for the slower rates of polymerization. Since DIBP decomposes very Slowly even in this range of high temperatures, no dead ending is observed, The conversion monoton- ically increases with the pressure, since the rate of le decomposes more slowly than POLYMER ENGINEERING AND SCIENCE, MARCH, 1985, Vol 25, No. 4 erature (6) Temperature (20) 120,40 150 160 120" "140 150_160 PA f a sno Number Average Melcuar Weight ° O16 2086 a5 80 Go 10 2 30-40-90 G0 Presse (sigh ry can —— 105-010 = 000s reget wad, Mss. rsa i) Fhe 8 Eft of presnce on a prformane ofa rug fie (or Mh retin cue fa pmercton of a tated Iy Fey pede Ti vir temperate (oper we ey CA oer i eqn initiation is enhanced by increasing temperature. ‘As in the ease of AIBN at low temperatures, the concentration of DTBP ean be considered constant during the entire course of the polymerization re- action, The number average molecular weight eas- ily meets the commercial requirements if the pres- sure is maintained below 30 psig. A polydispersity index of 2 is rigorously maintained in the studied range of pressures and initiator concentrations (Fig 9). ‘in conclusion, a purely kinetic model showed that a judiefous choice of the initiator (DTBP) and tight control of the temperature provided by the MMA vapor-liquid equilibrium can overcome the prob- ems of low molecular weight and high polydispers- ity. ‘The Rheokinetic Predictions ‘The operability of tubular reactors for the pro- duction of commercial grade polymers under high temperature conditions suggested by the above ki- netic analysis is further tested by the more accurate theokinetic model (2). Isothermal conditions at 130°C are assumed, corresponding to a pressure of 20 psig uniformly maintained in the reactor. The calculated profiles are plotted in Figs. 20 and 11. In spite of the above-T, temperatures, severe chan- neling persists in the conversion and velocity pro- files. The viscosity curves show that pure PMMA still coats the wall. This polymer layer hasa viscosity 241Pierre E, Baillagou and David S. Soong Tamperaars C0) Tenserore te) aa fue z 9 F owot 504) 3 anol i i Sos] E awe i Sttiitn romrewe —sorery onset ROO oo orn oose®® E20! 0% § cope _— sorry» one pe erg = Sten : Setanta 1 oe Lope : g 3 er 1 L L of 0406 0810 Radial Distance (em) ig, 10, Rheokinetie model predictions for the hove-Tyiother- inal buik polymerization nf MMA iutiated by DTBP. Axial slo ‘ty, cancersion and elscosity radial profiles at cartons aia di ty comersion and cust ail fier at cart ail di. 8.0258 moe oh = 1 ais 2a Loy os| = 08+ oa oa| oof oa 06 08 10 Rodial Distance (em) a rot 3 3 st 3 o o2 04 06 08 10 Radial Distance (em) POLYMER ENGINEERING AND SCIENCE, MARGH, 1985, Vol. 25, Ho. #Free Radical Polymerization of Methyl Methacrylate 2.0%108 T T T — 200. & .5x05| g \400 3 00 2 105) g ono #800 em| a € 2 osxiost 4 L L L 1 oO 02040608 10 Rodial Distance (em) 9 T T 1 & a L L 00a 08-0810 Radial Distance (cm) Pig. 11. ltheokinetic model pretiction forthe above Tothermal bulk polymerisation of MMA initiated by DTBP. Number acerage inolecular weight end polydispersity radial profiles at various lita! distances tna tubular reactor (P= 20 psig, T= 130°C, [DTHP|p = 0.0258 mole/L, a= I gh) POLYMER ENGINEERING AND SCIENCE, MARCH, 1986, Vol 25, No. 4 of the order of 10! eP. The average molecular weight and polydispersity of the polymer (Fig. 21) are quite acceptable except within the solid layer. Nevertheless, on the basis of this model simulation, successful tubular renetor operation on an industrial scale under the present conditions again seems du- bious. ‘The experimentally observed operability of AIBN cor benzoyl peroxide initiated polymerizations at high temperatures is not due to a substantial de- crease of viscosity of the material at high conver- sion, but rather is a direct consequence of the dead ending occurring in the reactor. In this case, the very long residence times of the peripheral layers do not induce the channeling effect, since the po- lymerization has already terminated by initiator depletion before the layers become too viscous. The most logical operating strategy suggested by this work is to increase the temperature of the reactor until the viscosity of the pure polymer is low enough to allow the flow of the melt accumu- lated on the reactor walls. The melting point of the polymer used in our viscosity modeling (8) was found around 180°C, Unfortunately, the depolym- erization of PMMA, reported to occur around 160°C under reduced pressures sets an upper limit to the allowable temperature in the reactor. How- ever, depolymerization may be unimportant at high pressures (160°C corresponds to a pressure of 53 psig through MMA vapor-liquid equilibrium), and the possibility of operating above 180°C should not be definitely eliminated without experimental evi- dence. Since the distortion of the velocity profiles is mainly caused by the very long residence time of the reactive mixture near the wall, slowing down the reaction in this region would delay (although not avoid totally) the channeling phenomenon, This objective may be reached by feeding the initiator only in the core of the reactor. Subsequent polym- erization taking place in the central region would increase the viscosity of the mixture Tocally and induce the formation of a flat velocity profile in the central zone. This plug region would be surrounded almost pure monomer (since no initiator was fed at the periphery of the tube) and therefore ‘would move easily along the axis of the pipe. Even- tually, the initiator will spread over the tube radius by diffusion. The formation of a viscous (and even- tually solid) layer of polymer coating the reactor wall would at least be delayed until some initiator reaches the peripheral region by this purely diffu- sive process. This idea has been investigated com- putationally, and will be written as a short com- ‘munication in the future. CONCLUSIONS Successful experiments have been performed by operating the reactor above the PMMA glass-tran- sition temperature, This success is attributable to the disappearance of the channeling effect. Fairly ‘uniform radial profiles of the different process var ables are predicted by our model. The real reactor 23Pierre E. Baillagow and David 8. Soong exhibited plug flow characteristies, as shown by the satisfactory fit of the experimental data achieved by the simple plug flow model. However, the number average molecular weights of the polymer produced in this range of tempera- ture are quite low, varying from 10° to 2.5 x 10, respectively, with’ AIBN and benzoyl peroxide as initiators. Bath mode! predictions and experiments agree on these orders of magnitude. In an attempt to inerease the molecular weight, slow burning initiators, such as DTBP, were com- putationally investigated. On the basis of the po- Iymerization kineties alone, it was concluded that satislactory molecular weights (higher than 10°) can be ichieved by ‘DTBPanitiated. polymerization, provided the temperature is maintained below 145°C, However, the more accurate rheokinetic simulation predicts profiles similar to those pre lent at low temperature conditions. The same dis- tortion ofthe velocity profiles and channeling effect eventually emerge, seriously compromising the op- erability of an industrial tubular reactor. The decrease of the viscosity at above-T, temper- atures is not the major factor responsible for the operability and plug flow behavior of the tubul reactor observed in our high temperature expe nents. The high rates of decomposition of AIBN and benzoyl peroxide in this range of temperature ‘cause the reaction to stop prematurely by depletion of initiator. By imposing a maximum achievable conversion, this dead-ending phenomenon holds the viscosity of the mixture at a manageable level, Unfortunately, it is precisely these high rates of initiator decomposition that are responsible for the formation of low molecular weight polymers. ‘An encouraging feature of the high temperature experiments is the very efficient control of the temperature by the vapor-liquid equilibrium of the polymerizing mixture. The pressure can be viewed as a powerful control variuble restricting the tem- perature in the entire reactor. In the range of “observed pressures (less than 50 psig) and conver sions (less than 40 percent), the temperature of the reactor very closely followed the boiling point of the monomer predicted by the Antoine equation. ‘The almost unexplored field of above-T, polym. erization, is a promising direction for future re- search. Control of the reaction temperature by the monomer vapor-liquid equilibrium is not restricted to tubular polymerizers, butis applicable to differ- ent reactor configurations. Pressurizing the reactor ‘would also facilitate the separation of the polymer from the residual monomer. When exiting the out- Tet valve, the polymer-monomer mixture udergoes an isenthalpic flash. A flash drum mounted at the outlet of the reactor would serve as the first stage of the separation process to allow recovery ofthe unreacted monomer for recycling, ‘Another potential advantage of above-T, polym- ation is the, probable attenuation of the gel effect since the free volume within the polymeriz- ing medium is significantly increased in this range of temperature. This idea, mentioned by Stolin, Merzhanov, and Malkin (9) was not experimentally investigated during this study since all the high temperature polymerizations were dead ended at low conversions, before the onset of gel effect Some high conversion experiments ea be per- formed ubove-T, by using a slow burning initiator such as DTBP i a pressurized batch reactor. The conversion data gathered in such experiments would answer some important questions about the behavior and control of polymerization reactors REFERENCES 1. 1, Valsunis aud J. A Biesenberger, AICHE Symp. Seres 160, 73,18 (1976), PLE, Ballagow and D. S. Soong, Polym. igs) OJ Banda BH ners ed Panera Woo" 2nd Ed. Wiley Interseionee, New York, (1975). 4. RUC. Reid, J.°M. Praysitz, and "T. K, Sherwood, "The Properties of Gases and Liquids,” Sed Bd, MeGraw-Hil, New York, (197) 5, LS, Liskin, "Vinyl and Diene Monomers, * igh Polymer Series, 24, Wiley Interscience, New York, (1070), 6, Neher, Ine ny Clim, 63,6 (1570) 7. ALD. Schmidt and W. TL Ray, Chem. Eng Sey 36, 241 s7a). 8 PE Gon. 9. A.M. Stolin, AG. Meezhanov, and A. Ya, Malkin, Polym ‘Big, Se, 19, 1065 (1974). Sei, 95, 12 allagow snd D, 8, Soong, to appear in Clem. Bg POLYMER ENGINEERING AND SCIENCE, MARCH, 1985, Vol. 28, No.4