Polymeric Plasticizers Preparation and Characterization of a Series of Terminated Polyes t e rs JOSEPH E. KOROLY AND ELLINGTON &I. BEAVERS Rohm 6% Haas Co., Philadelphia, Pa.
P
RIOR to 1942,the primary plasticizers of commercial importance for polyvinyl chloride and chloride-acetate copolymers were simple esters of monohydric alcohols andpolybasic acids, such as dibutyl phthalate and dibutyl sebacate. For some purposes, plasticizers of this kind (now generally described as “monomeric”) have the good feature of plasticizing efficiency, particularly desirable a t low temperatures, but also exhibit a “fugitive” character that is undesirable. The monomeric plasticizers are relatively small polar molecules, mobile and migratory, generally classified as good solvents and soluble in turn in many oils, greases, lacquers, etc., with which the plasticized film may come in contact. Investigators in the Rohm & Haas Laboratories discovered in 1942 that linear polymeric esters, derived from properly chosen dihydric alcohols and dibasic acids, are compatible with and excellent plasticizers for polyvinyl chloride and polyvinyl chlorideacetate (8). Because their average molecular weights are higher than those of the monomeric plasticizers, the polymeric esters were found to have greater permanence as plasticizers, with very low volatility, low extractability by solvents, and low tendency to migrate into other organic materials. Because of these characteristics, polymeric plasticizers have found wide commercial acceptance for applications in which durability and stability are primary considerations. Polymeric plasticizers pose some challenging and interesting theoretical questions. Present-day concepts of the mechanism of plasticization can hardly account for the good efficiency of polymeric plasticizers (6). It is the purpose of this and later communications to present the results of studies carried out in these laboratories to improve appreciation of the basic chemical and physical factors which influence the now well-known performance of polymeric plasticizers. The preparation of a family of polymeric esters is described here. Most of the linear polyesters described in the literature (1, 7, 9) (exclusive of patents) have had either hydroxyl or carboxyl groups, or both, at the chain ends. The concentration of terminal functional groups in the products has in these cases been inversely proportional to the number-average molecular weight. “Terminated” polyesters, as defined here, are those having the general schematic structure ( 5 ) :
L
- G-(-A
- G-)n-L
where A represents a dibasic arid, L a monobasic acid, G a dihydric alcohol, and n the number of repeating units in the number-average molecule, each “unit” being the esterification product of one molecule of glycol and one molecule of dibasic acid, In this study, n will have various values from 0 t o 10. The monobasic and dibasic acids chosen will be essentially nondistillable under the conditions employed in the polyesterification. The terminated polyesters were selected for this study for several reasons: When the reactants are of good purity, and if no important side reactions can occur, the polyesterifications proceed to terminal viscosities and molecular weights which are highly reproduciblei.e., the reaction is by nature self-limiting and requires no critical control after the proportions of reactants have been specified.
Within the range selected, the viscosities of the products are capable of precise measurement. Also, average molecular weights are of magnitudes that allow the application of more than one direct method of estimation, The products are essentially neutral. Theoretically, a terminated polyester contains no free hydroxyl or carboxyl groups (typical alcohols and acids are known from earlier experience to show poor compatibility with polyvinyl chloride). I n practice, terminated polyesters of this type retain only a low population of free functional groups. In general, the chemical composition and average polymer size of these products (none of which actually represents any commercial product) can be closely reproduced and physical properties of the products assessed with reasonable precision. EXPERIMENTAL WORK
Adipic, azelaic, and succinic acids were used as supplied coinmercially in excellent quality. A technical grade of sebacic acid was distilled in vacuo to yield a center cut having melting point of 133-134” C. and neutral equivalent of 101 (theory 101). Thiodipropionic acid was prepared by the method of Keyssner (16). yMethyl-yacetopimelic acid was prepared by cyanoethylation of methyl ethyl ketone (6). Commercial lauric acid was redistilled and a center cut of high purity used. The commercial grades of ethylene glycol and 1,2-propylene glycol were used as supplied. Commercial trimethylene glycol was thrice distilled without entirely eliminating a slightly J ellow color and an odor of stale hay. 2,2-Dimethyl-l,3-propanediolwas prepared by the procedure of Whitmore et al. (20). %But>l-2-ethyI-lJ3-propanediol was made by a procedure similar to that described by Tribit (18). The two substituted propanediols were prepared by C. 13. McKeever, Rohm & Haas Laboratories. The intermediates were charged to the reactor in amounts calculated t o yield a product having a value of n specified by the relationship:
where [A] and [L]are moles of dibasic acid and lauric acid, respectively. The amount of glycol required in each case was t h e calculated functional equivalence plus a 25 mole % excess. Zinc chloride (0.1% on the weight of carboxylic acids) was t h r esterification catalyst in all cases. The process was essentially that which has been described (2, 4, 17). The reactor was a three-necked flask fitted with agitator, thermometer, carbon dioxide inlet, and an upright, steam-cooled, three-bulb hllihn condenser packed with borosilicate glass helices. In the first stage of reaction. the batch temperature was allowed t o rise to 200’ C. while water in the vapors distilled, and vaporized glycol was returned by reflux. The pressure was then lowered over a period of 4 hours to 100 mm., and the batch temperature of 200” held until the acid number of the product was 10 or less. (Acid number is expressed as milligrams of potassium hydroxide per gram of sample; determined by dissolving 5 grams of sample in 75 ml. of benzene and 25 ml. of ethyl alcohol and titrating with 0.1 N alcoholic potassium hydroxide to a phenolphthalein end point.) At this point, the batch was cooled to 175” C., the condenser replaced with a downward take-off, and the system evacuated to a pressure of 1 to 2 mm. The pot temperature was then raised to 220“ C. and these conditions were maintained until the melt viscosity of the product, determined a t 25” C,, reached a constant value. Suitable precautions were taken at all times to avoid exposing the product t o air until the reaction was complete and the product cooled. Abnormal behavior was apparent during the preparation of two Some cleavage of thiodipropionic acid, or of
of these polymers. 1060
May 1953
INDUSTRIAL AND ENGINEERING CHEMISTRY
its ester, waa indicated by the formation of a milky precipitate of zinc sulfide. Further evidence of partial decomposition in this case was the fact t h a t 323'% of the theoretical amount of distillate was obtained, which was shown t o be neither water nor glycol. (While this figure seems large, the theoretical ratio of weights of distillate and product is small, and this represents a relatively small amount of decomposition of the acid.) The keto group in y-methyl-r-acetopimelic acid also seems to have led t o complications, as is evidenced by the abnormally high acid number of this polyester and by other properties discussed later.
g
less convenient, indication. ) Samples removed periodically during the second stage were examined for melt viscosity a t 25" i 0.2" C. [The relationship between melt viscosity and molecular weight (10) of these polymers will be discussed in another report.]
;
grooo 3 d Y m
4
DISCUSSION
Z
The purpose of the first reaction stage is direct esterification with the formation of glycol half-esters and hydroxyl-terminated low polymers:
p
P
0
The rate and degree of completion of this stage are conveniently followed by periodic analyses of the reaction mixture for the concentration of unesterified carboxyl groups. These values are expressed in Figure 1 as the acid number, In the second reaction stage, the polymer is formed by transesterification, For example, the following reactions occur:
u
HO-R-OOC-R'-COO-R-OH
+ HO-R-OOCR" + HO-R-OH HO-R4OC+R'+ HO-R-OH t
HO-R-OOC-R'-COO-R-OH
~HO-R-OOC-R'-COO-R-OOCK."
2 HO-R+--OOC-R'-COO-R-OH COO-R -O0C-R'-C00-R-OH
30-0 34-1 34-3 34-5 34-10 36-1 36-3 36-5 36-10 310-1 310-3 310-5 310-10 35-3 38-3 39-3 3K-3
None Succinic Succinic Succinic Succinic Adipic Adipic Adjpjc Adi~i~
210-3 T10-3 410-3 54-3
Sebacic ' Sebacic Sebacic Glutaric Thiodipropionic Azelaic y Methyl yacetopimelic Sebacic Sebacic Sebacic Succinic
56-3
Adipic
610-3
Sebacic
910-3
Sebacic
-
-.
Glycol 1.2-Prouvlene 1'2-Propjilene 1'2-~ropylene 1 '2-Propylene 1:a-Propylene 1 2-Propylene 1'2-Propylene 1'2-~ropylene 1:2-Prowlene 1 2-Propylene 1:2-~ropylene 1.2-ProDvlene
n 0
1 3 5 10 1 3 5 10
1
3 5 10 3 3 3
1 2-Propylene Ethylene Trimethylene 1,%Butylene 2,2 Dimethyl 1,3propanediol 2,2 Dimethyl 13propanediol Dimethyl 1,3/2,2 propanediol 2 Ethyl 2 butyl propanediol
-
-
-
-
- -
-
-
Acid Number 0.7 1.0 1.6 1.3 1.0 0.6 0.6 1.7 1.1 0.9 1.4 1.7 3.6 1.7 1.1 1.7
SEb
... 163
116 103 92 148 117 98 100 172
iii 135 ... ...
...
... ... ...
12.9 1.7
3
1.0
...
3
1.7
3
1.0
3
1.4
... ... ...
1.0
. I .
a aimule indication of the comDosition of the nolvmer ---?hi-fi&t- two digits represent the number of Earbon atoms iG-&e glyo'ol and dibasic acid, respectively. The digit following the dash re resents the value of n. Thus, the product coded "30-0" is 1,2-propyKne dilaurate. In a few cases where a number would not be adequately dietinotive a letter is used in the code number. b kaponification equivalent. a T h e m d e number is
10
5
4
0
8 19 16 REACTION TIME, HOURS
24
PO
Figure 1. Direct Esterification 0
3 3 3 3
1.1
6
0
TABLE I. LAURATE-TERMINATED POLYESTERS Dibasic Acid
E t
U
t
I n this stage, the stoichiometric excess of glycol is removed. The ultimate result to be expected is an equilibrium distribution (11) of polymer molecules of different lengths, terminated by the monofunctional reactant, R"C0OH. The possibility of ring formation is discussed below. The most convenient indication of rate in the second stage is the change in melt viscosity of the reaction mixture. (Change in hydroxyl content would probably be a more exact, though much
Codea
100
Y
+ nHOOC-R'-COOH + 2R"COOHe R"CO0-R-OH + . .+. etc. + 1.25(n + 1)HZH t
1.25 (n +!I) HO-R-OH
1061
Melt visoosity Acidnumber
Figure 1 illustrates a typical condensation.
The polyester
in this example is a laurate-terminated 1,8propylene adipate in whichn = 3:
r
1
CiiHzaCOO-CH~-CHOOC(CH~)~C00---CHzCHOOCC~~H~~
1
I
CHI
I
]a
CHs
Chemical composition of all the produrts prepared for this investigation is shown in Table I. Except in the case of polymer 3K-3, the products have very low concentrations of free carboxyl groups. The values in Table I for acid number of the products represent almost entirely the contribution of the zinc soaps formed from the transesterification catalyst (these soaps titrate as acid also, under the conditions of the detkrmination). An effort was made to determine the concentration of free hydroxyl groups in the products by the method of Verley and Bolsing (19) and by various modifications of this procedure, b u t highly reproducible results could not be obtained. Values for the hydroxyl number of any particular sample were found t o vary from -10 to +20, which eliminates confidence in small differences, but nevertheless indicates t h a t the concentration of hydroxyl groups is of a low order of magnitude. (As'a point of reference, it might be noted that the hydroxyl number of propylene glycol is 1476.) In principle, the method is carried out b y reacting a weighed sample of the polymer with a known amount of acetic anhydride in the presence of pyridine. T h e excess anhydride is then hydrolyzed with water, and the amount of unreacted acid determined by titration. The amount of esterified acetic acid is calculated by difference. The result is expressed as the hydroxyl number, in the same units as the acid number. The certainty that the melt viscosity of the products (with the exception of 3K-3) had reached a constant value for several
1062
INDUSTRIAL AND ENGINEERING CHEMISTRY
hours before polyesterification conditions were discontinued is a further assurance that the concentration of reactive groups had been reduced to a low magnitude. It is significant that the same polyesterification conditions will produce poly( l,2-propylene sebacate) of 20,000 to 25,000 molecular weight in the same reaction time, which represents about 99% completion of reaction.
Bo00 -
I
I
I
I
I
I
i
I
l
l
/
I
J
i
-
Vol. 45, No. 5
TABLE 11. NUMBER-AVERAGE MOLECULAR WEIGHTS Code 30-0 34-1 34-3 34-5 34-10 36-1 36-3 36-5 36-10 310-1 310-3 310-5 310-10 35-3 35-3 39-3 3K-3 210-3 T10-3 410-3 54-3 56-3 510-3 910-3
Theoretical Mol. W t .
Mol,
wt,
by
Sap. Eq.
440 598 914 1230 2020 626 998 1370 2300 682 1166 1650 2860 956 1130
608 928 1236 2024 592 936 1176(?) 2200 686
600 849 i1156 1668 620 It 1008 r 1150 1939 It
963 1037 971 1057 746 1266 901 973 1053 1111
1346 1136 902 1371 1367
1100
1278 1502
10 22
1 1 6 0 . k 17
10
1930 f 90 660 i 7 , .
......
2485 i- 113
888 r 8
984
1208 1110 1166 1222 1026
430 i 2 624 i 11 805 + 5 1165 1674 f 40 616 j= 5
18
1150 r 38 1523 =t17
1584 2970
1110
Ebulliometric bxol. wt. I n acetone In benzene
= 7
=
5
& 3 =t 5
+
7
=!= 20
*i 33 2
f 13 f 2
This is illustrated for the series 3 4 1 to 34-10 in Figure 2. Ring formation (14, 15) is a possible explanation suggested by T. G. Fox, of these laboratories, Saponification equivalent, essentially an end-group analysis, would not distinguish between A and B, for example:
500
‘ 0
r
1
C,,H,,CO~CH,cHoO(CH,),Coo--CHz-CHOOCC1€~~
I
2
4
6 8 THEOETICAL x
Figure 2.
10
12
/
I
1,
14
I
CII,
Ring Formation
Line represents theoretical values.
0 Values of molecular weight derived from saponification 0
equivalent Ebulliometrlc values (acetone)
The saponification equivalent of each polymer is a characteristic from which number-average molecular weight of the polymer can be calculated. The significance of the method depends upon the fact that the saponification equivalent of terminal ester groups (L-GL) is substantially different from that of the recurring diester units ( -GA-). Thus, it can be shoim that 12
=
WLGL- 2(SE) 2 ( S E ) - W7GA
where WLGL= molecular weight of the diester derivable from the glycol and monobasic acid, WGA = .sum of the atomic weights in the recurring unit derived from the glycol and dibasic acid, SE = saponification equivalent of the polymer, and n has the meaning previously described. Number-average molecular weights determined in this manner are presented in Table 11. Molecular weights were also determined ebulliometrically (by D. R. Conlon, H. W. Douglass, and H. F. Mason, of these laboratories) by a modified Menzies-Wright method ( I d , IS). These data are shown in Table I1 with a comparison of results produced by benzene and acetone as the solvents. The molecular weights by saponification equivalent are less influenced by analytical errors in the case of the succinates than in the case of the sebacates (the greater the difference between the saponification equivalents of the recurring unit and the terminal group, the more dependable is the derived value for molecular weight of the polymer). In general, the molecular weights by saponification equivalent agree fairly well with the theoretical molecular weights, and the disparities appear t o be randomly above and below the theoretical values. An unexpected observation was the regular downward deviation of ebulliometric molecular 17-eights from the theoretical.
B On the other hand, a colligative property of solutions of A or B, such as boiling point elevation, would distinguish between the molecule A and the^ combination of molecules, B, having lower average molecular weight than A. The lower molecular weights by the ebulliometric method are consistent with some ring formation, therefore. It is also intuitively reasonable t h a t the amount of ring formation should be greater in the polymers of high molecular weight (lower concentrations of monofunctional reactant). Jacobson and Stockmayer ( 1 4 , 1 5 )have shown t h a t the proportion of rings does not increase appreciably with molecular weight, in polyesters which are not “terminated.” I n the present case, however, the termination reaction can be considered to compete with ring closure, provided that rings do not arise by intramolecular ester exchange-( within the molecule ,4,for example. to yield the two molecules, B). If intramolecular ester exchange were as likely as the possible condensation reactions, then the terminated character of these polyesters would have little or no influence upon their ring content. On the other hand, if the Condensation reactions were more likely, the higher concentrations of “terminator” a t lower molecular weights should depress ring formation. Conversely, as the theoretical molecular weight is increased by lowering the concentration of terminator,
May 1953
INDUSTRIAL AND ENGINEERING CHEMISTRY
1063
(3) Ibid., 2,499,983 (March 7, 1950). (4) Biggs, B. S., Erickson, R. H., and Fuller, C. S., IND. ENG. CHEM.,39,1090 (1947). * ( 5 ) Boyer, R. F., T a p p i , 34, 357 (1951). (6) Bruson, H. A., U. S. Patents 2,386,736 (Oct. 9, 1945)’ 2,342,606 (Feb. 22, 1944). (7) Carothers, W. H., Chem. Reus., 8,353 (1931). (8) Conyne, R. F., and Felske, F. G., unpublished results: Brit. Patents 586,826 (April 1 , 1947), 588,574 (May 28, 1947). (9) Flory, P. J., J . Am. Chem. SOC.,61, 3334 (1939); Chem. Reus., 39,137 (1946). (10) Flory, P. J., J. Am. Chem. SOC.,62, 1057 (1940) (11) Ibid., 64, 2205 (1942). (12) Hanson, W. E., and Bowman, J. R., IND.ENQ.CHEM.,ANAL. ED., 11, 440 (1939). (13) Hurwits, M. D., Conlon, D. R., Mason, H. F., Meunier, V C., and Auten, R. W., “Approximate Molecular Weights of Alco-
the normal amount of ring formation for conventional polyesters should be approached. The divergence illustrated in Figure 2 increases not only in absolute magnitude with molecular weight, but also percentagewise; this suggests that the ring content actually increases with molecular weight in the case of these terminated polyesters. Polymers 35-3 and 3K-3 (derived from P,P’-thiodipropionic acid and ?-methyl-r-acetopimelic acid, respectively) are poorly accredited members of the series because they have undergone other reactions as well as the intended polyesterification. Precipitation of zinc sulfide in the first case is evidence of “betacleavage.” The melt viscosity of polymer 3K-3 did not reach the normal constant value during the polyesterification, but rather continued to rise throughout a long reaction period. This might indicate t h a t the keto group in methylacetopimelic acid activated a n intermolecular condensation, perhaps resembling the Dieckmann reaction. ACKNOWLEDGMENT
The authors wish to express their appreciation t o Louis P. Hammett, who consulted with them throughout the course of the study of which this is a part. LITERATURE CITED (1) Baker, W. O., Fuller, C. S.,and Ileiss, J. H., J . Am. Chem. SOC., 63,2142 (1941). ( 2 ) Reavers, E. M., U. S. Patent 2,445,553 (July 20, 1948).
hol-Modified Urea-Formaldehyde and Melamine-Formaldehyde Resins,” presented before Division of Paint, Varnish, and Plastics Chemistry, Symposium on Urea, Melamine, and Related Resins, 119th Meeting, AM. CHEM.SOC., Boston 1951. (14) Jacobson, H., Beckmann, C. O., and Stockmayer, W. H., J . Chem. Phys., 18, 1607 (1950). (15) Jacobson, H., and Stookmayer, W.H., Ibid., 18,1600 (1950). (16) Keyssner, E., Ger. Patent 669,961 (Jan. 7 , 1939). (17) Rothrock, D. A,, Jr., and Conyne, R. F., U. S. Patent 2,437,046 (March 2, 1948); Brit. Patent 588,834 (June 4, 1947). (18) Tribit, S., U. S. Patent 2,413,803 (Jan. 7, 1947). (19) Verley, A., and Bolsing, F., Be?., 34, 3354 (1901). (20) Whitmore, F. C., Popkin, A. H.. Bernstein, H. I., and Wilkins, J. P . , J . Am. Chem. SOC., 63,124 (1941). RECEIVED for review October 7, 1952. ACCEPTED January 17, 1953.
Phenol-, Urea-, and Melamine= Formaldehyde Plastics J
RECENT DEVELOPMENTS P. 0. POWERS Pennsylvania Industrial Chemical Corp., Clairton, Pa.
T
HESE plastic materials are now well established items and the past 5 years, which are taken as including recent developments, have seen little that was revolutionary. Many workers have indicated there were no new developments. However, we find an expanding volume of products, new applications, and, what is of particular interest, a renewed activity in the chemistry of these plastics. Phenolic resins were important industrial materials long before the principles of polymer chemistry were formulated. Perhaps for t h a t reason, the chemistry of the reactions of phenol and formaldehyde has nevkr been completely clarified. Excellent work now going on in this field has added greatly t o our knowledge and promises t o make our understanding more nearly complete. PHENOL AND FORMALDEHYDE
Phenol is an important ,organic chemical and the plastic industry is the largest consumer. Recent surveys (8) have indicated t h a t the present capacity in the United States of 344,000,000 pounds annually is not sufficient] and further capacity is planned or building t o bring the total to 624,000,000 pounds. Much of the new production will employ a novel method, involving the peroxidation of cumene. Because neither sulfuric acid nor chlorine, materials which have recently been in short supply, is required, the process is particularly attractive. Other phenols can be made b y this method and i t is contemplated t h a t p-cresol
will be produced by the peroxidation of cymene, obtainable b y the dehydrogenation of dipentene or other terpenes. Cumene is oxidized with air t o the hydroperoxide (7), which on decomposition in the presence of acids and a catalyst yield8 phenol and acetone (Figure 1).
H HsCCCH3
Benzene
Propylene
(CHa)&OOH
(R)
(CH3) Cumene (Cymene) OH
f)
v
+$H3COCH3
(CHI) Phenol (p-Cresol ) Figure 1
&
Cumene Hydroperoxide
Acetone