literature Cited
Eastman Kodak Co., Japanese Patent 409,197 (July 31, 1963). Hanna, R. J., Ind. Eng. Chern., 49, 208 (1937). Ikeda, K., Sekine, Y., Ind. Eng. Chem., Prod. Res. Develop., 12, 212 (1973a). Ikeda, K., Sekine, Y., Ind. Enq. Chem., Process DES.Develop., . , in press (1973b). Kolensnikov, G. S., Said, E., Khassen, A , Smirnova, 0. V., T’gsokomol. Soedin., Ser. A , 7, 129 (1965). Korshak, 1‘.V., Frunze, T. M., Kozlov, L. Y.,B i t / / . Acatl. Sci. USSR. Diu. Cheni. Sei.. 1969 il962a). Korshak; I r ,V., Frunze, T . AI., I?ozlov,’L. V., ibid., 2128 (1962b). lIatsugaiie, Tahara, S., KatB, S., “Polycarbonate Resin,’‘ p 71, Nikkan Kogyo Shimbun-Sha, Tokyo, 1069. Morgan, P. W., “Condensation Polymer: By Interfacial and Solution Method,” p 342, Wiley, Kew York, X. Y., 196Sa. Morgan, P. W.,“Condensation Polymer: By Interfacial and Solution Method,” p 472, Wiley, S e w York, K.Y., 196.ih.
Morgan, P. W., “Condensation Polymer: By Interfacial and Solution Method,” p 497, Wiley, New York, N. Y., 196th Noguchi, -$., Yuki Gosei Kagaku Kyoh-az Shi, 21, 928 (1963). Noguchi, A., Koseki, K., Tanimoto, K., Honna, K., Shimada, Y., Sodeyama, H., ibid., 21, 694 (1963a). Noguchi, 9., Koseki, K., Tanimoto, K., Yasuda, A., Fukamachi, K., Watanabe, S., Honna, K., Shimada, Y ., Igawa, T., ibid., 21, 621 (1963b). Sekine, Y., Ikeda, K., Sawai, K.,Yoshizaki, H., Kogyo Kagaku Zasshi, 74, 2,550 (1971). Sekine, Y., Ikeda, K., Taketani, H., zbid., 73, 429 (1970). Smirnova, 0 . Y.,Losev, I. P., Khorvat, E., T’ysokomol. Soedin., Ser. A , 6 , 594 (1964). Tomikawa, AI., Fujimoto, S., Kobunshi Kagaku, 25, 625 (1968). RW~:ITI:Dfor review Sovember 1, 1972 ACCEPTEDJune 4, 1973
Molecular Weight Distributions of Polymers in the Synthesis of Bisphenol A-Tetrachlorobisphenol A Polycarbonate Copolymers by the Method of Successive Additions of Monomers Koji lkeda and Yoshiro Sekine” Department of Clzenzistry, School of Science and Engineering, T a s e d a Cniversity, Sishi-Ohkubo, Shinjuku-ku, Tokyo, J a p a n
Using the precipitation turbidimetric titration of Morey-Tamblyn, the relationship between polymerization conditions and molecular weight distributions in the synthesis of polycarbonate copolymers was studied, and f(M), the reaction process was analyzed. It was found that k, calculated from the equation y = k log C where y i s the volume ratio of the precipitant at the point of beginning of precipitation, C the concentration, and f(M) a function of the molecular weight (M), varied depending on M. The molecular weight distribution curve obtained b y this method was in good agreement with that obtained b y the fractional precipitation method and also that of equivalent mixtures. In the earlier stage of polymerization, the molecular weight distribution accorded with Poisson’s distribution, and under proper reaction conditions (pseudohomogeneous state) the distribution could b e represented finally with the distribution function of Schulz-Zimm ( h = 1 ). O n the contrary, under improper conditions, the distribution curve was a polypeaked one, which suggested the overlapping of Poisson’s distribution and Schulz-Zimm’s distributions ( h = 1 , 2), and the final polymers were less uniform in their compositions and were inferior to those copolymers of the simple distribution in their thermal stabilities and also in their modulus of elasticity.
+
In the polycondensation step of the bispheiiol -1 @PA)tetrachlorobisphenol -1 (TCBP.1) polycarbonate copolymer syntheses using the method of successive addit’ions of nionomers, it was recognized t’hat the composition and the properties of tlie copolymer were seriously affected by improper reaction coiiditioiis (an excess of tlie emulsifier, hydrolysis caused by a too high p H value, and a n excess of the catalyst). Therefore, t o syiit’hesizeco~iolymersof high quality, tlie reaction state iii the polycondensation step must be studied more in detail. For this liurpose, it ma)- be quite iiecessary to determine tlie polyineriza tioii conditions and the time del)eiidence of the molecular weight distributions i l l the course of the polycondensation reaction, to analyze the reaction process. and to elucidate the relationship betneeii the molecular 2 12 Ind.
Eng. Chem. Prod. Res. Develop., Vel. 12,
No. 3, 1973
weight distributions and the compositions and propert’ies of ‘the copolymer. Xolecular weight distribut’ions of polycarbonates have been studied hitherto mainly by determining the distributions of the final products of polymerization (Schnell, 1956; Tomikawa, 1963). Therefore, studies on the time dependence of molecular weight distributions in the course of this interfacial copolycoiidensatioil reaction may serye not only to s h o the ~ changes of molecular xeight distrilsutions and t o analyze the reaction process in the synthesis of polycarboiiates by the phosgene process but also to be helpful to work with general interfacial iiolycoiideiisatiolls. The authors deterniined the molecular weight distributions of the copolymers, with the samples for which the time deliendence of X iii the course of the copolymer synthesis had
been measured in the preceding report (Ikeda and Sekine, 19i3), and analyzed the reaction processes. X s i t was necessary for t h e determination of the time dependence of molecular weight distributions to measure with a relatively small amount of the sample xithin a short time, the turbidimetric titration process of ;\Iorey-Tamblyn, which has been known as very useful for the measurement of molecular weight distributions of BPA polycarbonate resins (Kroser, et al., 1960), was chosen as the most suitable process for this purpose. - i t first, the usefulness of this process in its application to chlorinated polycarbonate resins was tested, aiid then the relationship between the polymerization conditions and the time dependence of molecular u eight distributions was determined to elucidate the change of molecular weight distributions in the period from the earlier to the last step of polymerization. From these experiments, it TT as elucidated that copolymers of simple molecular weight distributions with uniform compositions and of evcellent thermal stabilities and other properties could be obtained only under proper conditions.
Figure- 1. Relationship between chlorine content (per cent) and of fractionated sample
y a t the point of the beginning of precipitation and the concentration C
Experimental Section
Samples. Copolymers synthesized b y the interfacial polycondensation process, described in the preceding report (Ikeda and Sekine, 19i3), ivere used. The calibration curve was drawn and examined v i t h the copolymer of = 42,860 and C1 content = 20.23%. Polymers were dissolved in methylene chloride, washed with n-ater in a separating funnel, dried completely, aiid then used for the experiments. The sample used to obtain the calibration curve \vas prepared by a successive fractionation method in the usual wal-! using methylene chloride as a solvent and n-heptane as a precipitant. Methylene chloride was washed with water, dried over calcium chloride, and distilled; n-heptane was distilled before use. Calculation of Molecular Weight. T h e int'riiisic viscosity [ q ] was obtained from the viscosity at 2 5 O , measured with Ubbelohde's viscometer. for the methylene chloride solutions of fractionated samples, and the osmotic pressure was measured with a Shibata osmometer for polymers. From these data molecular TTeights were calculated v i t h the equation [ q ] = 1.585 X 10-4Ar0.75 (25"). Precipitation Turbidimetric Titration and Method of Calculation of Molecular Weight Distribution A Kotaki T y p e ST-3 measuring apparatus for molecular weight, distribution was used. T h e inteimity of scattered light, was measured in t'he direction of a n intensive region, namely, 135". Through Filter-NV-II2, the light Ivith 365-mp rravelength was used. -1s a solvent-precipitant system, considering refractive indices, solubilities, miscibilities, and specific gravities, methylene chloride-cyclohexane was adopted. Cyclohexane was dist,illed before its use. Every possible care was taken to prevent a n invasion of any impurity in the course of preparation and dissolution, and all samples were filtered with a glass filter KO. 4. hdding the precipitant, as described in t'he literature (Higashide, 1961), a t a rate of 1 m1L30 sec, solutions were agitated constantly for 1 min, and 30 sec after the addition of the precipitant and stopping of the agitation, the indication of the meter was read off; these operations were then repeated. The calibration curve was obtained by the procedure described in the literature (Iwamoto, 1968). The samples, after being measured for their molecular weights, were dissolved to prepare some solutions, the initial concentrations Co of which ranged from 0.001 to 0.02 $100 ml; turbidimetric titrations of every solution were carried out. From the ernpirical relation between the volume ratio of the precipitant
Au
y =
k log
c + f(M)
(1)
(n-here k is a eotist'ant, f(M) is a function of niolecular weight, and C = Co(l - y ) ) , a group of parallel straight lines may be obtained for y - log C, correspondiiig to each value of molecular m-eight; then the values f(M) can be obtained from the section log C = 0 and the values of k from the inclination. Consequently, the calibration line, which shows the relationship between and f ( J f ) , can be obtained. I n t,his procedure, as i t was quite difficult to take exactly the point a t which the precipitation just began, considering that the turbidity curve rose almost linearly, the precipitation point was estimated approximately by the extrapolation as described in the literature (Higashide, 1961). As for samples for molecular weight distribution measurements, the titration was carried out with 0.006 g 100 ml solutions, the turbidity (T,) was read off from the deflection of the meter, regulating the highest turbidity T,,, (the highest turbidity corrected for the dilution effect of the precipitant addition) as 100, the turbidity curve (relation between T aiid y ) was plotted, T 1va.s determined graphically a t the volume ratio Ay = 0.005, and TI', was determined from the equation of LIorey-Tamblyn
From the equatioii f(X)
= y -
k log [1r,cO(l - r)l
13)
f ( N ) was also determined and \\as recalculated to 31 with the molecular calibration line; from the values of V and TT', weight distribution curves could be obtained. Results and Discussion
Preparation of t h e Calibration Curve for Molecular Weight Distribution Measurement. The relationship bet\T een chlorine content and the average molecular w i g h t Ji of the fractionated sample, used to prepare the calibration curve, is shown in Figure 1. I t waq found t h a t t h e fractionated samples were of almost uniform compositions in the range of 37 used for this experiment. namely, 15.000100,000. The volume ratios y a t the precipitation points of Ind. Eng. Chem. Prod. Res. Develop., Vol. 12, No. 3, 1973
21 3
-
-
Table II. l i m e Dependence of M , M,, M,, -
- -
M,IM,, and 6 under Various Reaction Conditions
K/MW
Time, min
M
M W
7 7 TB-TBb TB-TB
120 120 120 120
42,860 42 860 44.540 44.540
41,120 49,000 47,100 44,100
53,800 62,700 52 , 400 48 , 500
1 1 1 1
340 280 088 100
0 0 0 0
581 521 295 315
Fractional precipitn Turbidimetry Fractional precipitn Turbidimetry
2
0 40 80 120
17,440 25,460 24,550 27 730
18, 800 27,720 28,250 30 , 530
23.120 35,100 37,600 38,800
1 1 1 1
230 268 330 272
0 0 0 0
475 516 572 520
*Iddn curve I1 (emulsifier 0 2%)
1
0 40 120
15,190 31,710 71,880
15,460 26,730 68,890
19,900 35,600 95 800
1 290 1 341 1 391
0 537 0 583 0 623
-1ddri curve I Effect of amt of emulsifier (0.06%)
9
0 40 80
3 , 840 3 , 900 4 , 650
4,180 6,000 7,130
4,620 6,860 8,650
1 103 1 143 1 212
0 353 0 431 0 557
Effect of amt of emulsifier (0%)
6
0 40 120
20,730 29,980 45,350
27,080 26,280 40 , 000
32,200 33,600 52,000
1 190 1 280 1 30
0 434 0 522 0 545
Effect of amt of emulsifier (0.1%)
7
0 40 80 120
10,440 39,820 47,630 42,860
9,940 42,110 38 630 49.000
11,740 46,700 62.700 53.800
1 1 1 1
180 207 280 340
0 0 0 0
394 433 528 580
Effect of pH (12 0-12 4)
0 40 80 120
8 270 19,230 17,130 18,240
11,460 29,800 16.180 17 240
15,000 40 , 100 21,760 23,100
1 1 1 1
362 342 341 340
0 0 0 0
602 583 583 582
Effect of gH (13.0-13.4)
0 30 60 90 120
6 280 13,430 11,800 24 410 30 , 370
6.660 16,530 I7 , 230 23,180 31 , 860
7,300 20 , 600 23,000 30,330 42,200
1 1 1 1 1
096 246 292 320 323
0 0 0 0 0
307 494 540 566 567
Effect of methanol addn (0 5 nil)
30 90 120
15.520 16.990 17,410
16.880 16.050 20,810
21,100 20,400 24 950
1 248 1 215 1 20
0 490 0 447 0 446
Effect of methanol addn ( 1 . 5 ml)
0 80 160 240
3 630 3 , 970 4,400 3,240
3,170 4,230 6,540 800
0 40 80 150 180
6,480 9,090 10.500 21 , 820 43,510
4,990 6,920 11,110 17.540 43,800
No.
8
15
17
20 21 23
~
~
~
~
~
~
4
~
6
Remarksa
Effect of large amt of catalyst (3 55%)
6,020 8,520 14 430 24,100 56,800 ~
1 206
1 1 1 1
232 30 373 298
0 0 0 0 0
443 480 546 609 545
Variation of monomer ratio TCBPA’UPX (75 ’25 molar ratio) (addn curve 111)
Variation of morioiner ratio TCBPX/BPA 1 192 0 436 0 11,900 10.810 9.980 (25175 molar ratio) (addn curve IV) 0 631 40 1 401 25,880 21,440 18.200 0 508 80 1 260 18,730 17,020 14.890 1 243 0 452 120 23.280 18,720 20,660 a Additional experimental conditions: for addition curve I, emulsifier 0 . 1 5 (based on monomer), pH 10.0-10.4, catalyst 2.3% (based on monomer). * See Sekini, et al. (1971). 25
close to that of the radical polymerization (Toda, 1968), and. especially a t the early period of polymerization, owing to high concentrations of monomers and oligomers, having functional groups, and also oning to the low viscosity, effects of termiiiation reactions may be small and the reaction may proceed rapidly a t random. [See the discussion of time dependence of per cent C1 in chloroformate groups and of in the preceding report (Ikeda and Sekine, 1973).] Consequently. the molecular weight distribution of the product mag be in accord with Poisson’s distribution curve, and sharppeaked curves can be observed. To prove this, using Poisson’s distribution functloii 216 Ind. Eng. Chem. Prod. Res. Develop., Vol. 12, No. 3, 1973
(eq 4) and some assumptions, the molecular weight distribution curve was calculated. TT’,
= [v, (V
+ 1)]Pe-”~~-l:’P!
(4)
IT, is the weight percentage distribution, P is the degree of liolymerization, and v is equal t,o the number-average degree of polymerization P,. The weight-average degree of polymerization P , is giveii by P , = 1 V . Then, taking the molecular weight of the monomer of copolymer (B/T = 1/’1mole ratio) to be 323 and obtaining the weight-average molecular weight W, in the cases of v = 10 and 20, the molecular weight dis-
+
F r a c t i o n a l precipitation
N X 10
Figure 5. Comparison of molecular weight distribution curves obtained b y the fractional precipitation method and by turbidimetry (AT = 42,860)
."If
2
x 10-6
Figure 7. Amount of emulsifier vs. time dependence of molecular weight distribution curves (no. 9, 0%): -0-1 0 min; I 4 0 min; -X-l 80 min
65
50I
P-
1
0
-1
these distribution curves were 41,120 by fractional precipitation and 49,000 by turbidimetry, and thus pretty good agreement was observed. In Figure 6, are shown the turbidity curve of the mixture consisting of the same amounts of fractionated copolymers = 26,400 and J 7 = 81,600 and the molecular weight of distribution curve calculated from the turbidity curve. This curve clearly represented the feature of the mixed state: peaks of this curve almost agreed in their position with those of each component of the mixture. The turbidity curve was smooth and shoulderless and served as one piece of evidence for believing that the polymers used were true copolymers. Molecular Weight Distributions i n t h e Interfacial Copolycondensation Reaction. From t h e result of t h e examination of the calibration line, it was ascertained t h a t the molecular weight distribution of the polymer to be studied could be approximately represented by t h e molecular weight distribution curve, which was obtained b y using this calibration line. Therefore, with turbidimetric titrations of the samples, the time dependence of whose -lf, (viscosity-average molecular weight) had been measured under each polymerization condition described in the preceding report (Ikeda and Sekine, 1973), molecular weight distribution curves were obtained, and the xeight-average molecular weights, 3TV(calcd), calculated from these molecular weight curves were compared with Li7v.As shown in Table 11, both values were in fair agreement, and the effects of changes of the polymerization conditions could be elucidated clearly. On the basis of this fact, under both proper and improper reaction conditions, the reaction processes in each early, middle, and last stage of polymerization were analyzed from their molecular weight distributions, and the effects of compositions and properties of samples were also studied. Early Stage of Polymerization. A t the first period when phosgenization was completed, sharp-peaked molecular weight distribution curves were obtained in all cases, as shown in Figures 7-13. This may suggest t h a t t h e reaction proceeded with a mechanism different from t h a t of t h e ordinary homogeneous polycondensation. It is thought that the reaction rate of the interfacial polycondensation may be
Lv
Figure 6. Turbidimetric titration curve ( A ) and molecular weight distribution curve (0) of mixtures of equal amounts of fractions i v = 26,400 and 81,640
somewhat higher than those obtained by turbidimetry ; however, the former is somewhat smaller than the latter a t ,V > 60,000. The most important origin of this deviation may be attributed to the fact that the values of k applied were determined from the calibration line, using the values of average molecular weight LV of the sample before fractionation. Such differences may be considered as a n inevitable consequence of the following fundamental characters of both procedures. Namely, in the fractional precipitation, owing to the dropwise addition of the precipitant to relatively concentrated solutions and also owing to the tail effect, precipitates of low molecular neights may be contaminated with those of relatively high molecular weights, and accordingly, the average molecular weight may tend to larger values. On the contrary, in the turbidimetric process, when the precipitation of low molecular weight polymers occurs, precipitates of polymers of high molecular weights may already exist in the solution and the newly precipitated polymers may aggregate with the latter; consequently, the results may be less accurate especially in the low molecular weight region. The weight-average molecular weights ,V,(calcd) calculated from
Ind. Eng. Chem. Prod. Res. Develop., Vol. 12, No. 3, 1973
2 15
Table II. l i m e Dependence of -
-
- -
- -
M I M,, M,, M,fM,, and 6 under Various Reaction Conditions -
No.
Time, min
M
M,!
M,
7 7 TB-TBb TB-TB
120 120 120 120
42,860 42,860 44.540 44,540
41,120 49,000 47,100 44,100
53,800 62,700 52 ! 400 48 , 500
1 1 1 1
340 280 088 100
0 0 0 0
581 521 295 315
Fractional precipitn Turbidimetry Fractional precipitn Turbidimetry
2
0 40 80 120
17,440 25 , 460 24! 550 27! 730
18, 800 27 , 720 28 , 250 30,530
23.120 35 , 100 37 , 600 38,800
1 1 1 1
230 268 330 272
0 0 0 0
475 516 572 520
-1ddn curve I1 (emulsifier 0 2%)
1
0 40 120
15,190 31 !710 71,880
15,460 26,730 68,890
19,900 35,600 95 800
1 290 1 341 1 391
0 537 0 583 0 623
-1ddn curve I Effect of amt of emulsifier (0.06%)
9
0 40 80
3 840 3,900 4,650
4.180 6 !000 7,130
4,620 6,860 8,650
1 103 1 143 1 212
0 353 0 431 0 557
Effect of amt of emulsifier (0%)
6
0 40 120
20! 730 29 ! 980 45,350
27,080 26,280 40 , 000
32,200 33,600 52,000
1 190 1 280 1 30
0 434 0 522 0 545
Effect of amt of emulsifier (0.1%)
7
0 40 80 120
10 , 440 39 , 820 47,630 42 860
9 , 940 42,110 38,630 49.000
11 ,740 46,700 62 700 53.800
1 1 1 1
180 207 280 340
0 0 0 0
394 433 528 580
Effect of p H (12.0-12.4)
0 40 80 120
8,270 19,230 17,130 18.240
11,460 29,800 16.180 17,240
15,000 40 , 100 21 760 23 100
1 1 1 1
362 342 341 340
0 0 0 0
602 583 583 582
Effect of gH (13,0-13,4)
15
0 30 60 90 120
6,280 13,430 11,800 24 , 410 30,370
6,660 16,530 17 , 230 23,180 31 ,860
7.300 20 , 600 23,000 30,330 42,200
1 1 1 1 1
096 246 292 320 323
0 0 0 0 0
307 494 540 566 567
Effect of rnethanol addn (0 5 nil)
17
30 90 120
15,520 16,990 17,410
16,880 16,050 20,810
21 100 20,400 24,950
1 248 1 215 1 20
0 490 0 447 0 446
Effect of methanol addn (1 5 ml)
20
0 80 160 240
3,630 3,970 4 ! 400 3 , 240
3,170 4,230 6,540 800
0 40 80 150 180
6,480 9,090 10.500 21 , 820 43,510
4,990 6.920 11,110 17.540 43,800
8
21 23
~
~
~
~
~
~
~
GSIM,
6
Remarksa
Effect of large amt of catalyst (3 55%)
6.020 8.520 14 430 24,100 56,800 ~
1 1 1 1 1
206 232 30 373 298
0 0 0 0 0
443 480 546 609 545
Variation of monomer ratio TCBPA,’13PAi (75 ’25 molar ratio) (addn curve 111)
Variation of monomer ratio TCBPA/BPA 1 192 0 436 0 10,810 9,980 11,900 (25175 molar ratio) (addn curve IV) 0 631 1 401 40 21,440 18,200 25,880 1 260 0 508 80 18 730 17,020 14,890 1 243 0 452 120 23.280 18,720 20,660 a Additional experimental conditions: for addition curve I, emulsifier 0 . 1 5 (based on monomer), pH 10.0-10.4, catalyst 2.3% (based on monomer). See Sekini, et ai. (1971). 25
~
close to that of the radical polymerization (Todn, 1968)) and, especially at, t,he early period of polymerization, owing to high concentrations of monomers and oligomers, having functional groups, and also oning to the low viscosity, effects of termiiiation reactions may be small and the reaction may proceed rapidly a t random. [See the discussion of time dependence of per cent C1 in chloroformate groups and of X in the preceding report (Ikeda and Sekine, 1973).] Consequently, the molecular weight dist,ribution of the product mag be in accord with Poisson’s dist,ribution curve, and sharp-peaked curves can be observed. To prove this, usiig Poisson’s distribut’ionfunction 2 16 Ind.
Eng. Chem. Prod. Res. Develop., Vol. 12, No. 3, 1973
(eq 4) and some assumptions, the molecular weight distribution curve was calculated.
K,
= [v, ( Y
+ ~)]P~-”V~-‘~’P!
(4)
TV, is the weight percentage distribution, P is the degree of i,olynierizat,ion, and v is equal to the number-average degree of polymerization P,. The weight-arerage degree of polymerization P , is given by P, = 1 Y . Then, taking the molecular weight of the moiiomer of copolymer (B/T = 111 mole ratio) to be 323 and obtaining the weight-average molecular weight Xwin the c%sesof v = 10 and 20, the molecular weight dis-
+
1 Iilr
80,
loo
141
I1 pH 12.0-12.4(No.7)
I pH
13.0-13
4tNo.8)
-2 6 50 c
%
c
'0
2
4
6 8 ,cI x 10-4
10
0 '
12
Figure 8. Amount of emulsifier vs. time dependence of molecular weight distribution curves. (I) NO. 1 , 0.06%, addition curve I: -A-, 0 min; --0--, 40 min; -X-, 120 40 min; -A-, min. (11) No. 6, 0.1 %: -0-, 0 min; -O-, 120 min
2
4
6 .I1 x 10-4
Figure 10. Effects of pH: -0-, 80 min; -X-, 120 min
8
12
10
0 min; -A-,
40 rnin;
-n-,
I
.Mx1@-:
Figure 9. Time dependence of molecular weight distribution curve for addition curve II (no. 2): -O-, 0 min; ---A---, 40 min; -0-, 80 min; X ---, 120 min
---
tribution curve was calculated. The result calculated from Poisson's distributions in the case without emulsifiers is shown in Figure 14. Assuming that the polymerization pro3550 Poisson's distribution and ceeded up to 50% with then 50% with 8,6780 Poisson's distribution, the calculated curve shown in Figure 14 was obtained. Perhaps, this curve corresponds to no. 9 in Figure 7 . Actually, it was ascertained to be similar to the experimental results. However, these Poisson distributions may espress only the distributions of unstabilized polymers a t the intermediate stage, and the same molecular weight distributions cannot always be obtained a t the more advanced stage of polymerization. As the molecular weight distribution curve may shift to higher molecular weights inthe course of the reactiontime'effects of termination reactions may become conspicuous, owing to the concentration decrease of functional groups and also to the viscosity increase of the polymer produced; so the reaction rate may become
ti
Jln10
'
Figure 1 1 . Effect of methanol addition. (I) NO. 15, 0.5 ml: -0-, 0 min; -A-, 30 min; -D-, 60 min; ---O---, 90 min; -X-, 1 2 0 min. (11) No. 17, 1.5 ml: -A-, 30 min; ---O---, 90 min; -X-, 1 2 0 min
lower than that of the early stage and, a t the subsequent middle to last stage, may approach that of the ordinary polycondensation reaction. Consequently, the molecular weight distribution may be espressed with the distribution function of Schulz-Zimm
g(M)
+
=
i/r(h
+ 1)yh+1M' exp(-yyM)
(5)
where r ( h 1) represents a r function, of which the parameters are y and h, and there is a relation LVw= (h l ) / y .
+
Ind. Eng. Chem. Prod. Res. Develop., Vol. 1 2 , No. 3 , 1973
2 17
400,
Figure 12. Time dependence of molecular weight distribution in the case of using large amounts of catalyst
M x 10-4 Figure 14. Molecular weight distribution curves calculated from Poisson's distributions: ---O---, 100% ItT, 3550 Poisson's distribution; ---A---, 1 00% 6780 Poisson's 50% 3550 Poisson's distribudistribution; -O-A--, tion 50% &fw 6780 Poisson's distribution
+
Y x 10-'
Figure 13. Effect of compounding the ratio of monomers. (I) -0-1 0 min; -A-l 40 min; - - - 0 - - - ,80 min; -U-l 150 min; -0-1 180 min. (11) ---0---,0 min; ---A- I 4 0 min; -o-~ 80 min; -X-l 120 min
Middle Stage of Polymerization. At the middle stage of the polymerization, i t was observed t h a t , while these reaction systems under agitation were in a viscous, emulsified state, the systems left standing were a p t t o be separated in all cases. The separation may be caused by copolymers consisting of strong polar monomers. However, while the systems were considered to be in well-emulsified, nearly homogeneous states (perhaps, pseudohomogeneous states; see Japanese Patent from PPG (1966)) under proper conditions, the systems under improper conditions were in badly emulsified states (perhaps, heterogeneous states) and were a p t to promote the disproportion of the methylene chloride phase with the witer phase in their microstates; therefore, some differences in concentrations and in reaction rates resulted. The effects of these differences on molecular weight distribution curves were subsequently observed. 2 18 Ind.
Eng. Chem. Prod. Res. Develop., Vol. 1 2, No. 3, 1973
a,
n,
(1) In Case of the Reaction System Being in the Pseudohomogeneous State. As shown in Figure 8-1, under proper conditions, no. 1, i t was observed t h a t the sharp distribution a t the early stage of polymerization became broader and broader and shifted to higher molecular weight re,'oions and could be characterized wit,li overlapped distribution curves of Poisson's distribution a t the early stage of polymerization and that of Schulz-Zimm ( h = 1) a t the middle stage. The complex curve is shown in Figure 15. This curve was obt'ained by assuming that a t the early stage, a 2 6 7 , ii, 16,470 ( Y = 50) Poisson distribution and a 76% AVw 16,470 SchulzZimm dist'ribution were applied and, a t the nest stage, a 7 5 7 , xw32,600 ( v = 100) Schulz-Zimm distribution vias overlapped wit,h a 267, complex distribut,ion obtained a t the former stage. It is clear that this result agreesrell, in the mode of changes, with the experimental result shown in Figure 8-1. (2) I n Case of the Reaction System Being in the Heterogeneous State. Under improper condit'ions, such as a n excess of emulsifier (see Figures 8 and 9 ) ) too high p H (see Figure l o ) , a n addition of methanol (see Figure 11))and a n excess of catalyst (see Figures 12 and 13-11), molecular weight distributions were polypeaked and accorded with molecular weight distributions obtained a t heterogeneous polymerizations. I t is well known that the molecular weight increases a t a n interfacial polycondensation are greatly affected by the ratio of number of terminal chloroformate groups to that of terminal hydrosvl groups a t the interface of methylene chloride and water (Noguchi, e t al., 1963). Therefore, conditions which may disturb the balance of this ratio may strongly affect particularly the molecular weight distributions. I n case a n escess of the emulsifier hPPE (trade name Triton X-
5
I
l i
.zIxlo-~
I\. :: 10
Figure 1 5. Complex molecular weight distribution curves of Poisson’s distribution -and Schulz-Zimm’s distribution ( h = 1): ---0---, 100~oM, 16,470 Poisson’s distribution; ---X---, 100~o$fW 16,470 Schulz-Zimm’s distribution; _ . _ . _ _ . _ . - 100% 32,600 Schulz-Zimm’s distribution; , 55% iVn,16,470 Poisson’s 75% iqW 16,470 Schulz-Zimm’s distribution (1); -D-, 25% distribution (1 ) 75% *Vn.32,600 Schulz-Zimm’s distribution; -X-, 1 00% ilTW75,800Schulz-Zimm’s distribution
x --
ivW
+
+
100) is used, where HLB (hydrophile-lyophile balance) of this emulsifier is 13.5, it caii be expected t,hat, a t the interface, lyophilic and hydrophilic groups are relatively well balanced, and the existence of a large amount of hydroxyl groups directed toivard the xvater side may inhibit oligomers in contacting the chloroformate aiid sodium phenoxide groups. Consequently, several phenomena, such as temporary cessation of growth etc., caii be observed. 111case that the heterogeneity of the reaction system, caused by the emulsifier, is observed, with a high pH value or a n addition of methanol, the polymerization may be stopped by the decomposition of terminal chloroformate groups, arid the amount of oligomer of both terminal groups HOROH may increase and a part of it may react with the oligomer CICOR’OH
I’
0
Supposing that the polymerization proceeds with this mechanism, it may be expected t’hat the molecular weight distribution of the product may be the relatively sharp distribution of Schulz-Zimm (h = 2). Distributions of Schulz-Zimm may vary with trro parameters, namely, h, which may correlate wit’h the degree of freedom of the functional groups, aiid y, which may correlate with the rate and the velocity of the reaction, to complicate the mode of overlapping of distributions. On the contrary, sharp distributions which were observed under conditions of pH 12.0-12.4 (see Figure 10-11)>a small amount of metlianol being added (see Figure 11-1) aiid a n excess of the catalyst being applied (see Figure 12), can be considered to be the distribution close to that of Poisson, owing to the fact that a random polymerization may rapidly proceed similarly to that in the early stage of polymerization (see Figures 7-13). Thus, in case a small amount’of methanol is added, the regulation of molecular weights i3 made difficult locally in the course of reactions. I n case of relatively high pH values as 12.0-12.4,
Figure
1 6. Complex
molecular weight distribution curves
of Poisson’s distribution and Schulz-Zimm’s distributions (h = 1, h = 2):---o---,100% 16,470 Poisson’s distribution; ---A-, l_OO% aw 32,600 Poisson’s distribution;
--- X ---,100% Mw 1 6,470
distribution; distribution; 16,470 and 10% 32,600 Poisson’s 3Oy0 iVw16,870 Schulz-Zimm’s distribution 50% 32,600 Schulz-Zimm’s distribution
---a---, 1 00%
--, 10%
distributions (h = 1 ) ( h = 2)
+
Lqw 32,600
lvw +
Schulz-Zimm’s Schulz-Zimm’s
nW
ivw
the BPA component could be activated. The sharp distribution observed in the presence of large amounts of the catalyst caii be explained with the local progress of the molecular growth. That is, while the molecular growth a t the chloroformate group of BP-i may be inhibited owing to the hydrolysis of that group caused by the excess of the catalyst, the molecular groxth a t the chloroformate group of TCBPA may progress, because the intermediate product of the chloroformate group of TCBP