The Chemical Engineering

Journal, 34 (1987)

1-9

Modelling and Simulation of Styrene-Acrylonitrile Kinetics

M. OLIVERES, M. ALONSO, F. RECASENS and L. PUIGJANER* Univereidad Politknica de Catalunya, E.T.S.E.Z.B.,

Emulsion Polymerization

Diagonal, 647, 08028 Barcelona

(Spain)

(Received April 30, 1986; in final form November 3, 1985)

ABSTRACT

The emulsion polymerization kinetics of the styrene-acrylonitrile system have been studied. A dynamic time-dependent model has been developed, which successfully describes the course of the conversion of styrene as compared with that of acrylonitrile. This model avoids the difficulties encountered by previous models when applied to high acrylonitrile-con ten t polymers, and is able to predict the conversion process for both batch and continuous reactors. The model is aimed at the simulation and operation of composition-controlled semibatch and continuous processes.

1. INTRODUCTION

Emulsion polymerization constitutes a multiphase reacting system where the reactant is usually introduced as a separate phase, while the reaction occurs mainly in the polymer particles. This implies the migration of monomer from its initial phase to the polymer particles through the aqueous phase. A physical description of the system was fist given by Harkins in 1945 [ 11. This constitutes the basic framework of later models and is based on the three-stage approach represented in Fig. 1. The process starts from nucleation of polymer particles promoted by the radicals generated at the expense of some initiator. The second stage is characterized by particle growth, according to the polymerization reaction mechanism studied by several researchers [ 21, and is sustained by the transfer of the monomer to the particles through
*Author to whom correspondence addressed. 0300-9467/87/$3.50 should be

the aqueous solution. The emulsifier present in the solution stabilizes the polymer particles during their growth. Thus, the decreasing concentration of free emulsifier leads to an interruption of the nucleation process both via micelles and via precipitation [ 31. In the last stage, the particles already formed continue to grow until the total consumption of the monomer still present in the solution and dissolved in the particles. The first model equations for the system described above were presented by Smith and Ewart [ 4 1. The three-interval scheme was also modelled by Gardon [5], Ugelstad and Hansen [ 61. Although comprehensive models have been developed in order to describe continuously the emulsion polymerization process using only one set of equations [7 - 91, improved models for the three-stage system have appeared recently [lo - 121, which enable simulation in batch and semibatch operations [ 13,141. In the present work the emulsion polymerization kinetics of styrene-acrylonitrile (STAC) has been studied in order to develop a composition-control system for the process. The time-dependent model produced avoids the difficulties encountered by previous models when applied to high AC-content polymers, and it is able to predict the conversion evolution process for batch and continuous reactors. The model is aimed at the simulation and operation of compositioncontrolled semibatch and continuous processes, which will be the subject of a later publication.

2. EXPERIMENTAL

PROCEDURES

The reactor consists of a jacketed vessel of volume 1 1, equipped with an anchored @ Elsevier Sequoia/Printed in The Netherlands

Fig. 1. Physical picture of the emulsion polymerization system: stage I, particle initiation; stage II, particle growth; stage III, final monomer consumption, C, and C, are the polymer and monomer concentrations respectively and p is the number of particles.

Fig. 2. Schematic diagram of data acquisition system. The specially designed vapour probe is connected to a gas chromatography sampling system (GC) and TRC is the temperature controller.

agitator. A flow of nitrogen (certified grade, 99.99%) was provided to maintain the inert atmosphere free from oxygen. Efflux gases from the headspace could be vented through a sampling system equipped with an electropneumatic valve, and analysed by gas chromatography every 5 min (Fig. 2). A computer-controlled system was incorporated later for data information retrieval and manipulation, which was capable of immediate action on the monomer feeding pumps, as will be discussed in a later publication (Oliveres et al., in preparation). Polymerizations were carried out at 338 K, and at the agitation rate of 360 rev min-i. After the initiator solution was introduced in the initial batch, microfeed pumps were employed for monomer emulsion addition, when continuous operation was desired. The schematic set-up is shown in Fig. 3. Reagent grade sodium lauryl sulphate (SLS, Doesder) was used as emulsifier and laboratory grade potassium persulphate as initiator in all experiments. Commercial grade AC and ST were used; ST was distilled under reduced pressure (30 mmHg, 54 C) and then stored in a refrigerator until used. Distilled water was used as the dispersion medium, and hydroquinone to quench the poiymerization. The polymer yield was determined gravimetrically by reacting with hydroquinone, and average compositions in copolymer and unreacted monomers were found by gas chromatography. The detailed experimental apparatus, procedures and analysis are described in another publication [ 151.

Fig. 3. Simplified flowsheet of the computercontrolled (mC) laboratory-scale plant, provided with temperature meters TA, TR, TC and TO and flow meters F, FA and FB.

3. THE KINETIC MODEL

The Haskell-Settlage [lo] model for emulsion polymerization was used as starting point, and subsequently improved for better fitting to the experimental data obtained in our microplant and from other reliable literature sources. Besides the usual assumptions made in the Smith-Ewart hypotheses, the liquid phase distribution of acrylonitrile between monomer droplets and the aqueous phase is also taken into account. It is assumed, as a first approximation, that the number of free radicals per particle is one-half (Smith-Ewart case 2). According to the qualitative picture of the three-interval scheme, the model equations have been developed as follows. 3.1. Nucleation of polymer particles Emulsion copolymerization takes place in a heterogeneous system composed originally of water (W), micelle-solubilized monomer and

monomer droplets which constitute the organic phase (0). The free radicals are generated, at the expense of the initiator, at the rate of dR = 2KdI,, exp(-Kdt) dt (1)

(7) The rate of disappearance of each monomer inside the particles follows second-order kinetics as given by the expressions

where K, ana I, are the initiator rate constant and initial concentration respectively. Polymerization in the soap micelles S, is initiated when the radicals generated in the aqueous phase diffuse into them. The nucleation mechanism can be described by second-order kinetics
dN

-~B=WBAN%~)(AR)

+ (KBBWL~)W)

(9)

and the total polymerization rate will be RA = (-rA)(w) RB = (-rB)( VP) (10)

(11)

= KIRS, dt

(2)

where N is the total number of particles, K1 is the collision rate constant. S, is given by S,=S-S,-N (3) where S is the initial number of micelles and S, the number of micelles adsorbed on growing polymer particles, S, is therefore dependent on the effective surface of the polymer S, =AS6 (4) where 6 is the number of micelles required to cover 1 cm* of polymer surface. The total polymer surface area AS can be obtained by considering N spherical particles of density p and overall weight PO: AS = 4.85(POl~)*~N~ (5) The nucleation process stops when S, = 0, so that new particles are not generated, and thus N remains constant. 3.2. Copolymerization A copolymerization reaction as such always takes place inside the initiated particles. Particle growth terminates whenever a second free radical from the aqueous phase enters the polymer particle. Accordingly, one may conclude that in the general case, one-half of the particles are active and the other half inactive at any given time (Smith-Ewart case 2) [ 131: +N=AR+BR (6)

Therefore, the total mass of polymer obtained is PO= j(RA

0

+ RB) dt

(12)

3.3. Volume of polymer particles In order to solve the set of eqns. (8) - (12) it is necessary to determine the volume of polymer particles VP in the reactant phase and the monomer concentrations A&VP and B&VP of acrylonitrile and styrene respectively. The particle volume is made up of polymer, styrene and acrylonitrile volume contributions VP = VAP + VBP + VP0 The styrene contribution lated from the expression
$)=

(13) VBP can be calcu-

Bp
P,+BP

(14)

which relates the adsorbed styrene in the particles BP and the polymerized styrene P,. The non-reacted styrene balance at stages I and II gives BO=BG+BP (15) In the last stage (III), the monomer in the droplets is exhausted, BG = 0, and only the styrene adsorbed remains in the particles until reaction is completed. To calculate the volume of acrylonitrile adsorbed in the particles VAP, we must take into account its water solubility. During the first two stages the AC balance gives A=AO+AW (16)

where AR and BR are the number of free radicals of acrylonitrile and styrene respectively. Furthermore, when a chain-end transfer from A to B occurs, we can write

which, together with the expression for the partition coefficient y for the liquid phase equilibrium between AC in the organic phase and that in the aqueous phase AOIBO = AW/W (17) allows the calculation of acrylonitrile in the organic phase A0 and that in the aqueous phase Aw. It is assumed at this stage that the monomer fractions in the droplets and in the polymer particles are identical; therefore A0 -=BO AP BP (18)

the droplets, then the unconverted styrene will be adsorbed in the polymer particles. Conversely, when acrylonitrile is the first to be depleted, part of it will be adsorbed in the particles and the balance of the same will be dissolved in the aqueous phase. In fact, the following eight possible situations may arise [16]: 1. 2. 3. 4. 5. 6. 7. 8. AC in droplets AC in droplets AC in droplets AC adsorbed in particles AC adsorbed in particles AC adsorbed in particles ST adsorbed in particles ST in droplets ST ST ST ST in droplets adsorbed in particles totally converted in droplets

ST adsorbed in particles ST totally converted AC totally converted AC totally converted

Solution of the last three equations together with eqn. (13) will give the value of VAP and will lead to the calculation of Amp, B mp, the actual concentrations in the polymer particles and the associated rates. 3.4. Relative depletion of the two monomers in the droplets When droplets containing both monomers are still present, the equilibrium postulated between adsorbed styrene BP and polymerized styrene P, implies that BP=

PS

Mws(l - @) However, when one or both monomers are consumed in the droplets, eqn. (17) is no longer valid. In this case the AN present will consist of the monomer fraction with particles and that dissolved in the aqueous phase. Furthermore, from experimental data the following empirical relationship can be written AP = A 1 + 1.05Aw/Ap

(19)

The corresponding decision tree is shown in Fig. 4. When the block diagram shown in this figure is taken into account in the overall mass balance calculation of the monomers during a reaction run, the actual situation is identified and the appropriate volume of polymer particles is thus calculated. The particle volumes as well as the mole fractions of the monomers in them are then used to calculate the actual reaction rates.

4. COMPUTER

PROGRAM

which relates the actual AC adsorbed on the particles with the total acrylonitrile present. At stage III the reaction rate is substantially reduced due to the slow diffusion process of the monomers through the particles, so that eqn. (6) takes the form 0.15N = AR + BR (21) The set of eqns. (8) - (21) can be used to solve the different situations that may arise depending on the specific conditions under which the polymerization reaction takes place. Suppose that styrene is depleted first in

The quantities involved in the course of copolymerization are the degree of polymerization, particle volume, conversion and composition of the polymer, both instantaneous and cumulative, and the composition of the monomer phase due to the relative depletion of the monomers in the droplets. The COPOLY program system was used. The system consists of an assembly of subroutines including integration, differentiation, function generators and an efficient package for solving non-linear equations. Plotting and interactive graphic blocks are included to be used also as part of the control package whose development is under way (Oliveres et al., in preparation). Information handling and transmission between subroutines was achieved through COMMON blocks. A block diagram of the

Fig. 4. Detail of the decision tree in the algorithm that calculates the particle volume. *
RADICALS dRldt=ZXdIO 1 PARTICLES dN/dt=K1 R-S,,, I ADSORBED (-Kd.t) -

,
PARTICLE AREA 4 A=4.fiS(PO/p)23N13 1 MICELLES

VOLUME VAP=Ap.14,a/D,

OF

1~IICILL.~
PARTICLES

Sp=A.6 r

VBP=Bp.14&Db VP=(VAP+VBP+VPO)/lOOO Amp=VAP.DalFlwaVP BmP=VBP.Db/MuaVP

tr PARTITION A =A0+4rJ (STAGES

I ljand

I
I

S,=S-Sp-N

I
REACTION 9 RB=KbaBmPAR+K PO=I(RA*RB)

I
RATE RA=K~aAmpAR+K,bqnpBR.VP bb dt B nip BR.VP

1 (STAGE

1
I I RESlDUAL A=AinitialA B=BinitialMONOMERS I RA.dt I RB.dt I

2,

PARTITION Bp=B Ap= lt1.05

3,

BP=Ps.@/H,,.(l-$1 AO/BO=AP/BP y=(AO/BO)/(Aw/S5.5) t I

AW/AP

Fig. 5. Simplified block diagram of the computer program COPOLY. computer program is shown in Fig. 5, in which the essential parts of the program are included. The program was run in a Digital VAX 11/780 series computer. A typical simu-

lation run takes 4 s and requires a memory size of 22 kbytes. The kinetic parameters and physical constants obtained from experimental data,

6 TABLE 1 Data for the experimental evaluation of the model Distribution coefficient of acrylonitrile: y = 8.3 Reactivity ratios: ra = 0.09, r), = 0.30 Density: D, = 0.81, Db = 0.91 p = 0.494 I0 = 1.11 x 10-j mol (KzOsSz) 1-l K A* = 265.7 x lop3 1 polymer mol- min- at 338 K KBB= 14.8 x 10e3 1 polymer mol- min- at 338 K Kd = 5.14 x 10M3 min- K1 = 1.67 x 1O-23 g 150.

loo_

and used to test the model Table 1.

5. RESULTS AND DISCUSSION

are given in

5q

Comparative results for experimental and computer runs for different initial charges, are represented in Figs. 6 - 8 where the calculated values are expressed as solid lines. In all cases the experimental values agree well with the calculated ones. Table 2 shows the comparison between calculated and experimental limit conversion and the overall stage II reaction rates. The excellent agreement between calculated and experimental values is further emphasized by the fact that the mean error of deviations obtained is less than 8% in the worst case, as

0 50 100 ml 150 Fig. 7. Comparison of experimental conversion with predicted conversion: case b, batch with 46 - 54 mol% ST-AC.

g/

15q

100

100 ml 150 50 Fig. 8. Comparison of experimental conversion with predicted conversion: case c, batch with 28 - 72 mol% ST-AC.
0

50

Fig. 6. Comparison of experimental conversion with predicted conversion: case a, batch with 30 - 70 mol% ST-AC.

LL?-?A
min

shown in the last column of the table. Furthermore, the validity of the model resides also in the fact that both reaction rate and limit conversion, which are the most useful parameters in practice, show a high degree of accuracy that qualify the model as a good predicting tool for operation and control purposes.

TABLE 2 Comparison between calculated and experimental limit conversions

911

Initial (mol% ST)

Experimental $1-1 Zlimit (%) 95 95 87

Calculated rP 2

Mean error deviation* (%I

151

min-' )
70 46 28 2.60 3.93 4.82 2.65 3.85 5.02 92 92 85 +4.1 k6.2 +7.0 lO(

5( sed in %, where yer and yc are the experimental and predicted values respectively of polymer being formed at the N intervals.

One general feature, also found by other researchers [ 121, is that the limit conversion decreases with increasing AC content in the initial charge. It is also well known that AC monomer favours a higher polymerization rate, possibly by a mechanism of autocatalysis. Although the homopropagation constant for AC is about 10 times higher than that for ST, this fact by itself does not sufficiently explain the increased rate observed in high AC-content polymers. The continuous nucleation process introduced in the model as an additional source of catalytic sites tends to confirm the acceleration observed in practice. The results obtained also indicate that it is possible to use the proposed model as a practical tool in describing the overall behaviour of both the limit conversion and the polymerization rate. In particular, it is interesting to note that for high AC-content copolymers, styrene is used up rapidly in the droplets as predicted by the model, which makes the reaction slow down. This gives rise to a limited conversion of the remaining AN that stops the reaction. Figs. 9 - 11 show this effect for the same case studies used previously. The underestimation in the exogenous nucleation rate inherent in previous models led to undercalculation of the limit conversion. Undoubtedly the inferior number of particles that result are compensated by an increase in the effective number that are active at a given time. This would mean that the Smith-Ewart assumption (average number of radicals per particle is 0.5) may be an

f Fig. 9. Variation with time of unreacted monomers (case a).

-r---_

Fig. 10. Variation with time of unreacted monomers (case b).

unreasonable hypothesis, particularly for high AC-content copolymers where there seems to exist a limited number of very active particles most of the time. In this sense the proposed model has used a new approach by using a diverse partition mechanism at stages I and II rather than at stage III, thus obtaining quite satisfactory results in all cases. Our present studies indicate the validity of the model in semicontinuous operation, which constitutes

8
10 V. C. Haskell and P. H. Settlage, Proc. AIChE 63rd Ann. Meet. Chicago, IL, 1970, AIChE, 1970, Vol. 25b. 11 T. Kikuta, S. Omi and H. Kubota, J. Chem. Eng. Jpn., 9 (1) (1976) 64. 12 A. Guyot, J. Guillot, C. Pichot and J. Rios, Am. Chem. Sot. Symp. Ser., 165 (1981) 415. 13 G. Ham, J. Macromol. Sci., Chem., Al 7 (3) (1982) 369. 14 S. Omi, M. Negishi, K. Kushibiki and I. Mamoru, Coll. Int. Copolym. et Copoly. Emuhion, Centre National de la Recherche Scientifique, Lyon, France, 1984, p. 51. 15 M. Alonso, M. Oliveres, L. Puigjaner and F. Recasens, Feasibility of estimating copolymer compositions from on-line headspace analysis in emulsion polymerization, submitted to Ind. Eng. Chem. Fundam., April, 1985. 16 M. Alonso, M. Oliveres, L. Puigjaner and F. Recasens, Further advances in the computer modeling and simulation of the styrene--acrylonitrile emulsion polymerization kinetics, in Proc. 3er Cong. Int. Infor. et Genie Chimique, ESCE Pub. Series, 27 (l), Paris, 1983, p. 40-l. 17 M. Oliveres, M. Alonso, F. Recasens and L. Puigjaner, Simulation and control of copolymerization reactors in semi-continuous operation, (in preparation).

Fig. 11. Variation with time of unreacted monomers (case c).

a valuable control.

tool for process

operation

and

ACKNOWLEDGMENTS

APPENDIX A: NOMENCLATURE

The authors wish to acknowledge financial support from the CAICYT (Grant No. 4010/ 79). The provision of commercial monomers by Arrahona, S.A. is also gratefully acknowledged. The free computer time provided by Instituto de Cibernetica is also appreciated.
REFERENCES 1 W. D. Harkins, J. Chem. Phys., 13 (1945) 381. 2 D. R. Basset and A. E. Hamielec, Emulsion Polymers and Emulsion Polymerization, Am. Chem. Sot. Symp. Ser., American Chemical Society, Washington DC, 1981, Vol. 165. 3 R. M. Fitch and C. H. Tsai, Polymer Colloids, Plenum, New York, 1971. 4 W. V. Smith and F. R. Ewart, J. Chem. Phya, 16 (6) (1948) 592. 5 J. L. Gardon, J. Polym. Sci., A-l, 6 (1968) 623. 6 J. Ugelstad and F. R. Hansen, Rubber Chem. Technol., 49 (1976) 536. 7 K. W. Min and H. W. Ray, J. Macromol. Sci., Rev. Macromol. Chem., Cl 1 (1974) 177. 8 C. Kiparissides, J. F. McGregor and A. E. Hamielec, J. Appl. Polym. Sk., 23 (1979) 401. 9 M. Morbidelli, G. Storti and G. Carra, Dynamics and modeling of emulsion polymerization reactors, in G. F. Frument (Ed.), Proc. Meet. ChemicaE Process Analysis and Design Using Computers, Antwerp, 1982, p. 29.

A A mP

A0 AP AR

AS AW B BG B mP

BO BP

acrylonitrile concentration (kmol ( m3 water)- ) acrylonitrile concentration in unit volume of total polymer particles (kmol (m3 polymer particles)-) acrylonitrile concentration in the organic phase (kmol (m3 water)-) acrylonitrile concentration in polymer particles (kmol (m3 water)-) number of free radicals with acrylonitrile end unit (kmol (m3 polymer particles)-) total surface area in unit volume of polymer particles (m2 (m3 water)-) acrylonitrile concentration in the aqueous phase (kmol (m3 water)-) styrene concentration (kmol (m3 water)) ) styrene concentration in monomer droplets (m3 (m3 water)-) styrene concentration in unit volume of total polymer particles (kmol (m3 polymer particles))) styrene concentration in the organic phase (kmol (m3 water)-) styrene concentration in polymer particles (kmol (m3 water)-)

BR

D* Db IO & K1

KIJ

Mwa

M WS N PO Ps
-rA

-rB

ri

R RA

number of free radicals with styrene end unit (kmol (m3 polymer particles)) acrylonitrile density (kg me3) styrene density (kg me3) initial initiator concentration (kmol (m3 water)- r) initiator rate constant (s-l) collision rate constant between radical and micelle (polymer particles m3 water (kmol radicals number of micelles h)-) monomer disappearance rate constant in the particles (m3 polymer particles (kmol h)-) acrylonitrile molecular weight (kg mol-) styrene molecular weight (kg kmol-) number of polymer particles total polymer weight (kg (m3 water)-) styrene concentration in polymer (kg ( m3 polymer particles)-) rate of disappearance of acrylonitrile (kmol (h m3 polymer particles)-) rate of disappearance of styrene (kmol (h m3 polymer particles)-) reactivity ratios total free radical concentration (kmol ( m3 water)-l) acrylonitrile polymerization rate (kmol (h m3 water )-I)

RB S S, S* VAP VBP VP VP0 W z

styrene polymerization rate (kmol (h m3 water)-) initial number of micelles number of soap micelles left at time t number of adsorbed micelles acrylonitrile volume in particles (m3 ( m3 water)-) styrene volume in particles (m3 (m3 water)-) volume of particles (m3 polymer particles (m3 water)-) volume of polymer particles (m3 (m3 water)-) amount of water (kmol (m3 water)-) total conversion

Greek symbols a partition coefficient for acrylonitrile between aqueous phase and monomer droplets partition coefficient for acrylonitrile between styrene and water number of micelles required to cover 1 cma of polymer surface ratio between adsorbed styrene in the particles and polymerized styrene polymer density (kg mm3)

Y 6 $J
P