ERIULSIOS POLYUERIZ.ITIOS O F S T Y R E S E
1175
E;\IL-ISIOS POLYMERIZA4T1OX OF STYRESE* C R S S T A H;IUSER
Department of Chemical Engineeizng, Jlassachitsetts Znstzticte of l'echnologij, Canibr zdge, Jlassachiisetts 4iYD
r;r>I PI:RRI dIonsanto Chernzcal C'o,irpan?/,Spiingjleld, dlnssach~isetts R e c e z ? e d F e b , iitrril 20. 1948
That emulsion polymerization should be considered a surface phenomenon has already been stated by several colloid chemists. The question, hon-ever, if polymerization is initiated by the use of oil-soluble or n-ater-soluble catalyst$, or starts at the oil-water interface. has never been answered 'satiqfactorily. With this in mind, tn-o series of polystyrene emulsions were prepared, one n-ith a n oil-holuble catalyst and the other with a water-soluble catalyst. The concentration of the emulsifying agent WRS the only variable factor. The data shonthat polymerization ocwirred simultaneously in the oil and miter phases, but not :it the oil-water interface. The value of the oil-phase polymerization was definitely established. 'This is in direct contrast to those theories of emulsion polymerization 11hich a s u m e that only aqueous-phase reactions are involved. Further work showed that a change in the amount of emulsifier and type of catalyst causes the molecular ireight distribution to vary over a ide range. The effect of molecular weight distribution on the physical properties of the polymer \vas recognized qiialitatively, and confirmed by studying the morphology of mised fractions ~iltramicroscopically11 ith incident light. These results are in line with irhat has already been demonstrated Trith other high-polymeric substances. This paper is concerned with the emulsion polymerization of styrene. On the basis of the investigations of Harkins and coworkers ( l ) ,T'inograd and co\~orkers(8), Mark and coworkers (3), and Price and -%dams (4), it has been siiggeqtcd that polymerization was initiated in the aqueous phase. X-ray data supposedly indicated that polymer particles 11-ere expelled from the soap micelles after they had reached a certain size and that they then continued t o grow at the e ~ p e n s eof monomer, diffusing from the monomer droplets n-hich serve as a reservoir to qupply monomer to the aqueous phase to initiate the growth of the small part icleq. ;1qurvey of theliterature revealed that these theories and deductions are based on work carried out with water-soluble catalysts. I t was believed that this type of catalyst obscured the contribution of oil-phase polymerization which was also sccurring. The present ork m s therefore undertaken in an attempt to resolve This paper is based on the AI S thesis of Eli Perry in t h e Department of Chemical 1:nginecr ng of the Ilassachusctts Institute of Technology. June, 1947. It was presented wforc t h e Division of Colloid Chemistry a t t h e 112th AIeeting of t h e ;\merican Chemical qocictv, n h i c h n a s hcltl in Sen- T o r k C i t v , September, 1947
1176
ERSST A . H.‘ITSER . I S D ELI PERRY
this problem. For this purpose t x o series of emulsions were prepared, one with an oil-soluble catalyst (benzoyl peroxide) and the other Tvith a water-soluble catalyst (potassium persulfate). The systems used xere extremely simple and consisted only of styrene monomer, emulsifying agent (sodium stearate), catalyst, and distilled water. The recognized variables, such as catalyst concentration, ratio of oil to water phase, temperature, agitation, and type of atmosphere, were held constant. The only variable factor for each series of polymerizations was the amount of emulsifying agent used. In order to avoid any difficulties which might occur due t o variations in the r a materials, ~ enough of each constituent was obtained so that the same lot could be used throughout the n-hole investigation. To this end the styrene monomer n-as used as received (i.e., with inhibitor) so as to avoid any purification Ivhich might give slightly different grades of monomer for the various runs. Polymerization was conducted under a nitrogen atmosphere in a t hree-necked flask equipped n-ith ground-glass joints and a mercury seal. The procedure conskted in weighing the ivater into the flask and dissolving the emulsifying agent and catalyst (if Ivater-soluble) in the water. The monomer was added (with the catalyst dissolved in it if oil-soluhle) and the dispersion ]vas agitated for 25 min. ,Z sample TI-as withdraivn. and the constituents of the flask were brought to temperature in 13-20 min. The temperature \vas maintained within i0.3OC. until the monomer ivas 93-95 per cent converted to polymer. The emulsion n-as then cooled to 35‘C’. and strained into jars for storage. -4ny lumps ivere air dried and ireighed. The course of the reaction ~vasfollon-edhy pipetting samples out of the dispersion and analyzing them for unreactetl monomer. The analytical procedure used in this n-ork v-as a niodification of the method of I-hrig and Levin (T). -4 rough estimate of the degree of conversion to polymer \vas ohtained as follon-s: 5 ml. of emulsion \vas mixed with 40 ml. of water. Ten milliliters of precipitating reagent (12.5 cc. of concentrated sulfuric acid, 50 g. of sodium chloride, and 1 1. of n-ater) n-as added slowly to the diluted emulsion. If the conversion was less than 40 per cent no precipitate resulted, hiit an oily layer of monomer vas salted out. -kt 30-60 per cent conversion small particles of polymer separated and formed a dispersion resembling shredded cotton rags in a liquid medium. Xt 70-80 per cent conversion the polymer lumped to form a single, solid, sticky ball. ;Ibove 90 per cent conversion a tine, hard precipitate formed, leaving a clear or almost clear supernatant liquid above i t . Sitrogen TI-as bubbled through the dispersion from the time the ivater vas charged to the very end of the run. The follon-ing data n-ere obtained for each polymer: ( 1 ) total time of reaction at the reaction temperature; ( 2 ) the per cent conversion as a function of time; (J) the approximate time of the induction period; ( 4 ) the amount of lumps formed; ( 5 ) the approximate particle size in the final product; (6) the “gross” molecular n-eight of the product polymer; ( 7 ) an approximate molecular weight distribution curve.
EMULSIOS P O L k ~ ~ E R I Z h T I OO SF STYRESE
1177
PARTICLE-SIZE MEASUREMENT
The particle size was estimated ultramicroscopically and by means of light transmission measurements. Keither method allowed an absolute determination of particle size, but a comparison between the various emulsions was all that was wanted. FRACTIOKATIOS PROCEDURE FOR iMOLECVLSR W E I G H T DISTRIBUTIOX
Fifty grams of polymer was purified by dissolution in 1000 cc. of methyl ethyl ketone and reprecipitation with 2500 cc. of methanol. Thirty grams of reprecipitated polymer was dissolved in 1500 cc. of methyl ethyl ketone in a 2000-cc. Erlenmeyer flask. Methanol was added in small increments t o precipitate the different molecular weight fractions of the polystyrene. The recovery of polymers as a result of fractionation was always 97 per cent or better . MOLECULAR W E I G H T D E T E R M I S A T I O S
The molecular weight was determined viscometrically, giving a n-eight average. The specific viscosity ( N a p )was calculated and a plot of Ar8,/C vs. C allowed the intrinsic viscosity, ( N ) , to be found. The molecular weight was calculated from the formula (5):
( N ) = 4.94
x
10-5
_110.78
I t is believed that emulsion polymerization is initiated sjmultaneously in the oil and water phases, and that the use of the oil-soluble catalyst has emphasized the contribution of the oil-phase polymerization to such a degree that it could be easily distinguished from the polymerization which was initiated in the aqueous phase. Most of the polymerization occurs as a zero-order reaction for both types of catalyst. Other investigators have demonstrated that the reaction rate should increase as the solubility of monomer is increased in the aqueous phase. The length of the induction period decreased as the concentration of the emulsifying agent was increased. If one pictures the induction period as lasting until all of the inhibitor reacts with actii-e nuclei, anything that promotes the rate of formation of nuclei (e.g., increased amount of emulsifying agent) shortens the induction period. The extra soap also solubilizes more of the inhibitor t o increase its rate of consumption. The particle size of the polymer decreased with an increased soap concentration, as indicated in figures 1 and 2 (3). The increased concentration of monomer results in more rapid formation of nuclei, each of which may become a new particle. Also, more soap is available for adsorption on the surface of new particles before the emulsifying agent disappears. This increase in the number of particles for a given charge implies that each individual particle is smaller.
1178
EltSST I. FI i U S E R A S D ELI PERRY
The initial sample of dispersion from the flask separated into tlvo layers immediately, rhowing that the emulsion polymer 11as not formed totally by mechanical means. The “two-layer sample” formed an emulsion polymer when allowed to stand unagitated on the 4elf tor 3 weeks at i’oom temperature. This phenomenon i* a consequence ot the picture of polymerization occurring in the micelles and the molecules being expelled :LS particles when they reach a certain size (3). The experimental facts u hich the t Inrury of aqueous-phase polymerization does not explain but which the picture of dual-phase polymerization can explain is
t
0
WATER
P
-
SOLUBLE
CATALYST
I
t
4
6
PERCPNr
8
SOAP
Fic,. 1. 1,H’rct of soap coricentratioii o n particle size (estimated microscopically)
lump formation. Figure 3 s h o m that lump formation decreases with increasing soap concentration. Any satisiactory theory must be able to explain this decrease and also the great differences found when using oil-soluble and Jvatersoluble catalysts. Lump formation mi.; appreciable for resins made with an oil-soluble catalyst, but resins made with a Jvatcr-soluble catalyst gave very little lump formation. In fact, ivitli the latter actual lumps did not form, but instead an “oily” spot was seen on the surface at 73-85 per cent conversion. This “oily” spot disappeared :ti the reaction proceeded and was present in the final polymer as a lump in. t o & in. in diameter for the range of soap concentrations used. For this reason, the size of the ‘,oily” spot was recorded qualitatively instead of weighing the lumps ( c j . figure 3).
EMTLSIOX POLYRfERIZ.-ITIOS
1179
OF S T T R E S E
For an explanation of lump formation one must also take into consideration that polymerization occurs simultaneously in the oil and aqueous phases. The droplets which are “oil-phase initiated” differ in one important respect from particles which are initiated in the aqueous phase in that they must vary in solids content from 100 per cent to 0 per cent monomer during the course of the reaction. This variation in solids content is a consequence of simultaneous polymer 60
WATER 40
- SOLUBLE
CATALYST
OIL- SOLUBLE
CAfALYJT 0
i.
LO -
0,
,000 A. f0)4000
t
4 PLTRCENr
6
8
SOAP
(6) 7 0 o a i . 80
OIL
60
40
-
WATER
- SOLUdLE
CATALYST
- SOLUBLE
CATALYJT
1
I
I
ormation and loss of monomer by diffusion. Therefore, a t some stage in the eaction these latex-like particles must exist in a “sticky” state. If, at this stage, he particle size of the droplets is so large that the adsorbed soap film cannot tabiliee them, the agitation will bring the droplets together and cause them o coalesce. Therefore, only certain particles in the emulsion coalesce and form umps. Since an oil-soluble catalyst accentuates the contribution of oil-phase iolymerization, the amount of lump formation vi11 be greater than when using water-soluble catalyst. With both an oil-soluble and a water-soluble catalyst he over-all time of reaction is controlled by the fastest rate, namely, that in the
1180
ERR'ST A. HAUSER AND ELI PERRY
aqueous phase. This implies a decrease in over-all reaction time when the soap concentration is increased, and less polymerization occurs in the oil phase with the resultant decrease in lump formation. If this picture of lump formation is correct, agitation should be important in regard t o the amount and nature of the lumps formed. The experimental data confirm this deduction.
8 ,
9,
WAMR
- SOLUBLE
CATALYST
( 9 UA L ITAT/ V F )
0
I
0
2
I
6
4 Pk-RCENT
8
SOAP
FIG.3 . Effect of soap concentration on lump formation
As shown in figure 4, the extrapolated value of the reaction rate constant is not zero a t zero soap concentration for an oil-soluble catalyst. For the watersoluble catalyst, the extrapolated intercept was negligible. With a watersoluble catalyst a t zero soap concentration one essentially separates the monomer and catalyst into different phases, so that very little oil-phase or aqueous-phase polymerization occurs. Therefore, the reaction rate constant must be very small. For the oil-soluble catalyst a t zero soap concentration one excludes monomer and catalyst from the aqueous phase, but brings them into contact in the oil phase. The large reaction rate constant is due t o the appreciable oil-phase polymerization, and demonstrates that it makes a significant contribution to the whole reaction.
EML-LSIOS POLYJZERIZATIOS O F STPRESE;
1181
More low-molecular-weight material was produced in the polymer as the soap concentration was decreased. This is seen in figures 5 and 6. The data of these figures are plotted in such a way that a curve tending away from the abscissa represents a preponderance of high-molecular-weight material, a straight line represents uniform molecular weight distribution, and a curve tending toward the abscissa represents a preponderance of low-molecular-n-eight material. Thc
-
WATER S O L UBL€ CA TAL Y J T
100 100
0 d O
0
6
4 PCACEN'T
8
SOAP
FIG.4 . Effect of soap concentration on t h e reaction ratc constant
trend toward increased low-molecular-weight material with a decreased soap concentration is unmistakable for both the oil-soluble and water-soluble catalysts. It must be emphasized that the molecular n-eights n-ere estimated viscometrically by finding the intrinsic viscosity, (AT). The recorded molecular weights therefore represent a measure of the difference of degree of polymerization rather than absolute values. -1 satisfactory picture of the variation of the molecular weight distribution curves with the soap concentration is presented by considering dual-phase polymerization. Less soap results in a longer reaction time, allows oil-phase polymerization t o occur, and reduces the Concentration of monomer and catalyst
1182
ERNST
.\.
CO
HhUSElZ A K D E L I PERRY
40 PERCENT
60 OF
80
L
/oo
POLYMER
FIG.5 . Effect of soap concentration on the molecular m-eight distribution in the polymer (water-soluble catalyst i .
I I -
01 0
0
t
20
60
40
PrRcrhrr
OF
I
I
80
/eo
POLYMZR
FIG.6. Effect of soap concentration on t h e molecular weight distril7ution in the polymer (oil-soluble catalyst).
(if oil-soluble) in the aqueous phase (so that less aqueous-phase polymerization occurs). Therefore, less soap implies more low-molecular-weight material and
E J K L S I O S POLPRZERIZ.ITION O F STYRESE
1183
less high-molecular-weight material in the polymer (cf. figures 5 and 6). -11though the shape of the curve changes for the polymers made ivith an oil-soluble catalyst as the soap concentration decreases from 7 to 1 per cent (figure B), the trend is not as apparent as in figure 5 for two reasons: ( a ) With an oil-soluble catalyst, the amount of oil-phase polymerization is large for any soap concentration, so that one is trying to detect a small addition to a large quantity. ( h ) With increased oil-phase polymerization (i.e., decreased soap concentration), the amount of lump formation increases greatly (figure 3 ) . Since lump for-
01
0
-
I
i?
PERCENT
01 0
L
f
6
8
6
8
SOAP
I
B
4 PERCENT
aOAP
FIG 7. Ilflect of soap concentration on t h e molecular weight of t h e unfractionated polymer Photomicrographs of polystyrene fractions
mation is due to oil-phase initiation, the more oil-phase polymerization which occurs, the more of that type of polymer is removed as lumps, which do not remain in the product. A study of figure 5 indicates that the mechanism of aqueous-phase polymerization may change as the soap concentration is varied. This idea is suggested by the difference in the molecular weight fractions of the polymers, and variation in its gross molecular weight (cf. figure 7). I n any case, the present report does not pretend to deal with the mechanism of polymerization, but only its loci. The mechanism is still unknown and requires much more research.
1181
ERSST -1. HAUSER .1SD ELI PERRI-
One must also consider the evidence against polymerization taking place at the oil-water interface. ( I ) Csing a water-soluble catalyst, the reaction rate constant was negligible a t zero soap concentration. As explained above, the catalyst and monomer are in different phases, so that if interface polymerization n.ere important the reaction rate constant would be significant under these conditions. ( 2 ) With a water-soluble catalyst lump formation nab small, and with an oil-soluble catalyst lump formation was large. If interface polymerization w r e important one n-ould expect the same reaction mechanism and, hence, approximately the same amount of lump formation iyith both types of catalyst. ( 3 ) Polymer particles of colloidal size were formed without agif ation. Interface polymerization does not provide any mechanism for the formation of small particles. It was therefore concluded that the experimental facts could be explained only by the theory of dual-phase polymerization. The molecular n-eight disiributiori in the polymers is of special interest. Qualitative tests on films made from mixtures of fractions of different molecular weight confirmed the idea that the toughness and elasticity were markedly influenced by the molecular m i g h t distribution. Thus, films containing 50-75 per cent of low-molecular-weight fractions a-ere much tougher than films of pure high or pure lon- molecular weight material. Much work remains to be done, but once the optimum molecular weight distributions for the various properties are known, it should be possible t o produce any molecular weight distribution desired in any molecular weight range by controlling such variables 2.s type and amount of catalyst, per cent emulsifier, temperature of the reaction, and concentration of the monomer (6). The variation in the molecular weight distribiition obtained in the present work indicates a general method of attack. Some ultramicroscopic observations n-ere made on the fractions of polystyrene, using the technique described by Kauser and le Beau (2). Films of the polymer were deposited on fine-mesh wire screen and viered under a Leitz Cltropak under controlled conditions of time and temperature. Photomicrographs published previously (2) shon-ed glob formation for many substances, due probably t o syneresis (flon- Jyhereby the low-molecular-weight material was squeezed out from between the high-molecular-weight chains). Therefore, the formation of globs depends in part on the molecular weight distribution of the pdymer. The freedom of movement of the chains and how that movement is affected by side groups are other facts t o be considered. Various fractions of high-molecular-weight and low-molecular-weight polystyrene were mixed in the proportions of 75 per cent low molecular weight and 25 per cent high molecular weight. The individual unmixed fractions rarely showed glob formation a t either room or elevated temperatures. Occasionally. globs were observed but they xere attributed to the fact that the fractions were not sharp. Often, fragments or particles were observed which were similar t o globs in appearance but which had sharp corners instead of smooth rounded surfaces. The mixed fractions did not form globs a t all times, especially a t room
