E. C. STEINER, R . K.PELLETIER, A N D R.0. TRUCKS
46i8 [CONTRIBUTION F R O M
THE
VOl. 86
EDGARc. BRITTOXRESEARCH LABORATORY, T H E D O I V CHEMICAL CO., MIDLAND, MICHIGAN]
A Study of the Polymerization of Propylene Oxide Catalyzed by Anhydrous Potassium Hydroxide BY E. C . STEINER. K. R. PELLETIER, A N D R. 0. TRUCKS RECEIVED F E B R U A R24, Y 1964 T h e polymerization of propylene oxide catalyzed by solid anhydrous KOH is shown to have essentially the same mechanism a s the well-known horriogeneous base-catalyzed polymerizations. The reaction is not surfacecatalyzed and the role of solid KOH is to convert hydroxylic groups almost quantitatively to alkoxide groups. Unsaturated end groups are shown to conie from rearrangement of propylene oxide to allyl alcohol, which initiates new polymer chains. Allyl ethers thus formed are converted t o &-propenyl ethers a t rates dependent upon their molecular weights. A kinetic analysis shows t h a t the proposed reaction sequence accounts for all the characteristics of the polymerization including reaction rates, induction period, average molecular weight, and molecular weight distribution, and the arnount and type of unsaturation in the polymer.
Alkylene oxides may be polymerized in a number of ways, the type of product obtained being dependent on the catalyst system used. Most commercial polymerizations are carried out by treating alkylene oxides with hydroxylic initiators (e.g., propylene glycol) in the presence of small amounts of KOH a t temperatures of 100° or higher. The products under these conditions are mostly telomers of the initiator and alkylene oxide and the average molecular weight is approximately determined by the ratio of alkylene oxide to initiator. When propylene oxide (PO) is polymerized under these conditions the product is always contaminated with a small amount of unsaturated, monohydroxy polymer. The unsaturation has been shown to be allyl and propenyl ether end groups.’ This impurity is detrimental to the use of polyglycols for the preparation of polyurethans. Xnhydrous KOH, in the absence of hydroxylic initiators, also causes propylene oxide to polymerize, but the reaction proceeds readily a t room temperature and the product consists largely of unsaturated, monohydroxy polyethers whose average molecular, weight is about 5000 regardless of the monomer-to-catalyst ratio.2 Several other features of the polymerization in the presence of anhydrous KOH are different from the commercial process: (1) the polymerization rate is much higher; ( 2 ) the average molecular weight is relatively constant throughout the polymerization ; (3) NaOH and LiOH are ineffective catalysts while KOH, RbOH, and CsOH are e f f e ~ t i v e . ~The marked changes in the characteristics of the polymerization caused by elimination of the hydroxylic initiator led St. Pierre and Price to suggest the possibility of a difference in reaction mechanism between the two polymerization systems. This paper deals with a study of the mechanism of the polymerization catalyzed by anhydrous KOH and will show that no basic change in mechanism is necessary to account for the change in characteristics of the two polymerization systems. The generally accepted mechanism of polyrnerization in the commercial process is4 ROH
+ HO-
fast
RO-
4- H,O
0
/ \ R O - f CH2-CHCHa _____
-+
PO
R O C H p C H ( C H I ) O - -+ polymer
(1) G. J . Dege, K. L. Harris, a n d H. S. hfacKenzie, J A m . Chem S O L, 81, 3374 il96Yj. ( 2 ) I, E. S t . Pierre a n d C. C. Price, ibid , 78, 2-132 (1966) (:O W . H. S n y d e r , J r . , Ph.11 Thesis, University of Pennsylvania, 1961. ( 4 ) G. Gee, W. C. E. Hiaginson, K . J T a y l o r , a n d M \V Trenholme, J Chem Soc.. 4298 (1961)
0
HO-
/ \
PO
+ CHZ-CHCH3
-+- H O C H P C H ( C H P ) O --+
+ RO-
ROH
fast
RO-
polymer
+ ROH
where R = any of the possible organic moieties including the initiator. The details of the mechanism are determined by the relative acidities of the various hydroxylic materials and the rate of nucleophilic attack of the different anionic species on propylene oxide. The source of unsaturation has been uncertain, but Price and S n ~ d e r ~ , ~ and Simons and V e r b a d suggested t h a t it is produced by the reaction 0
RO-
/ \
+ H-CHZCH-CHz
ROH
+ CH,=CHCH,O-+
PO
CH*=CHCHd OCsHs),,O-
They showed that allyl ethers will rearrange to propenyl ethers under basic conditions and thereby accounted for both types of unsaturation. Gee’ had suggested earlier that isopropenyl ethers could be formed by an alternate rearrangement of propylene oxide 0
RO-
0-
/ \
+ CH2-CCHs 1
-+
ROH
+ CH2=CCH3
PO --f
H CH2=C--(
I
OC3Hs)J-
CH3
This reaction is not likely since there is no evidence for the presence of isopropenyl ethers in polyglycols. However, the third analogous reaction has not been ruled out as a mode of formation of propenyl ethers
’
0
/ \
RO-
+ H-CH-CHCH;
---+
ROH
+ CH3CHzCHO-
PO
+
CHaCH=CH( OC3H6),0-
The mechanism of the polymerization catalyzed by anhydrous KOH is not as well understood. St. Pierre and Price2 initially proposed that the reaction was catalyzed by the surface of the KOH and involved a chain-type reaction which yielded polymer of nearly constant molecular weight. I t was suggested that the chain termination was an elimination reaction which amounted to a chain-transfer reaction. (51 C. C Price, lecture delivered a t S o r t h w e s t e r n University, Dec. 8, 1959 I6J I ) M. Simons and J. J. I’erbanc, J . P o l y m e r Sci , 44, 803 (1!>60) ( 7 ) G. Gee, Chem I n d . fI.ondonj, 678 (19.59).
Nov. 5, 1964
CATALYZED POLYMERIZATION O F P R O P Y L E N E
0
OH
/ \
KOH(s)
+ CHsCH-CH?
OXIDE
4679
I
I
+CHzCHCHzOK 0
OH
1
CHICHCH~OK
+
A
CHsCH-CH2
KOH(s) ___f
OH
I
t-’
CHaCHCHz( OC3He)nOK -0
0
z
W V
a W
a.
T I M E ,HOURS
Fig. 2.-The polymerization of propylene oxide by anhydrous K O H : 0 , propylene oxide remaining, percentage of starting 0, mmoles unsaturated polyether chains formed material; from 200 g. of propylene oxide; X , mmoles of dihydroxy polyether chains formed from 200 g. of propylene oxide; --, calculated curves.
is almost completely reacted. The total amounts of hydroxyl groups and of double bonds follow similar sigmoidal curves with the amount of hydroxyl always being greater than the amount of double bond. T h e average molecular weight of polymeric product, as determined by end group analysis, is high even a t low conversion, and changes by a factor of only about two throughout the polymerization (see Fig. 3 ) . This is in essential agreement with the results based on viscosity measurements reported by St. Pierre and Price.2
0 0
TIME,HOURS.
Fig. 1.- Effect of catalyst concentrations on r a t e of polymerization of propylene oxide by anhydrous K O H ; mole ratio of K O H to propylene oxide: 0, 0.022; 0 , 0.044; 8 , 0.087.
a fourfold change in amount of catalyst was made. Even a t a catalyst-to-propylene oxide mole ratio of one there was no significant change in reaction rate nor in product characteristics. These data rule out the possibility of any significant amount of surface catalysis. The kinetic course of the polymerization is shown in Fig. 2. In these reactions, the mole ratio of KOH to propylene oxide was 0.044, which was enough to eliminate the complication reported b y Snyder3 t h a t polymerization stops prematurely if insufficient catalyst is used. It was found t h a t there is an induction period during which the amount of organic hydroxylic material increases to a significant level. The polymerization then accelerates markedly until the monomer
I 20
40 60 T I M E , HOURS.
80
IO0
Fig. 3.-Molecular weight buildup during anhydrous K O H catalyzed polymerization of propylene oxide.
T h e polymer obtained when reaction is complete is a viscous oil with an average molecular weight of 50006000. T h e molecular weight distribution is extremely broad as may be seen in Fig. 4. The lowest average molecular weight fraction obtained was about 1600 g./mole, whereas the highest was about 31,000 g./mole. This distribution is much broader than the “Flory” distribution (also included in Fig. 4) which is approximated in commercial polyglycols.* The Flory distributionQ assumes t h a t all of the initiator is present a t the beginning of the polymerization, that there is no chain termination reaction, and that the propagation rate constant is independent of molecular weight. (8) ( a ) L. C. Case, J . P h y s . Chem., 62, 895 (1858); ( h ) I< J Morris and H . E. Persinger, Paper No. 6 7 , D i v . of Polymer Chem., 135th National Meeting of t h e American Chemical Society, Boston, Ivlass., April, 1959. (9) P. J. Flory, J. A m . Chem. SOC., 62, 1661 (1940).
E. C. STEINER, R. R. PELLETIER, A N D R . 0. TRUCKS
4680 IO0
0
a
0
0
do'"
8
0
10,000
5000
15,000
20,000
M O L E C U L A R WE I GHT.
Fig. 4.- Cumulative molecular weight distribution of two polypropylene oxides prepared with anhydrous KOH catalysis. Solid line represents a "Flory" distribution for polymer of t h e same average molecular weight.
In order to explain the available data, the following reaction sequence, a modification of that suggested by Snyder3and Price, is proposed fast
KOH(so1id)
KOH(so1ution)
0 KOH(so1ution)
(1)
OH
/ \
+ CHsCH-CHz --+C H I CI H C H z O K kd
(2)
The rearrangement of substituted epoxides to m , p unsaturated alcohols under basic conditions is a wellknown reaction. '" However, no definitive evidence has been reported for such a rearrangetnent under these reaction conditions. T h a t the reaction does indeed go has now been shown in two ways. Attempts were made to copolymerize propylene oxide and trans2,3-epoxybutane in the presence of anhydrous KOH. In these experiments varying mixtures of the two epoxides were allowed to react a t 30° until no further change occurred and then the volatile and polymeric components were analyzed separately. The volatiles contained, along with unreacted butylene oxide, small amounts of 1-buten-2-01, the product of a rearrangement corresponding to reaction 3 . The alcohol was 0
KO-
/ \
+ H-CHzCH-CH-CHa
ROK
+ CHaCH-CHz
isolated by gas-liquid chromatography and identified by comparing its infrared spectrum with that of an authentic sample. Essentially no propylene oxide remained a t the end of the polymerizations. Added butylene oxide, also, has a marked effect on the properties of the polymeric product (Table I ) , whereas inert
+ CHFCHCHZOK
(3)
EFFECT OF
+ CH3CH-CH2
k.
+ROCHzCHOK I
14)
R O K f HzO
(5)
+ KOH(so1id) e K O H . H 2 0 ( s o l i d )
(6)
Yield based on PO, 70
Molecular weight
Unsaturation, mrnole,'g.
0 10 20 30
97.2 97.5 95 8 100.5 99 5
5650 3910
0.141 ,218 ,291 ,393 ,616
50
fast
k,, --f
CHaCH=CH(OC3H6),0K
(7)
where R = any of the possible organic moieties. d kinetic model derived from this sequence has been tested and found to fit the data in Fig. 2 . Before discussing the kinetic analysis, however, the qualitative reasons for choosing this model should be mentioned. The reaction is not surface catalyzed; therefore there is no apparent reason why the well-known mechanism of commercial polymerizations should not apply. Hence, the homogeneous reactions 2 and 4 are included. If these two reactions apply, then the over-all rate of polymerization will depend on KOH concentration. For the rate to be independent of amount of solid KOH present, the KOH concentration must also be independent of it. Therefore, it is assumed that the dissolution process 1 is fast compared to the polymerization reactions and t h a t KOH concentration is practically constant. Reactions 5 and 6 have been shown to be relatively fast and to go almost quantitatively to the right. Thus low molecular weight polypropylene glycols in tetrahydrofuran solution equilibrate with excess anhydrous KOH within 30 min. to give about %5% yield of the corresponding alkoxides. NaOH under the same conditions yields only %yoalkoxide. It is felt t h a t the efficient dehydrating ability of KOH causes reaction 5 to shift to the right.
PROPYLEXE OXIDE( P O )
Mole 70 BO in PO
fast
ROH f KOH(so1ution)
CHn=CHCHz(OC3HBjnOK
t r a n s - ~ , 3 - E P O X u B U T A S E ( B O j O N POLYMERIZATION OF
/ \
CH3
H20
0-
+ CHz=CHLCH3
TABLE I
ki
+ROH 0
ROK
+ KOH
0
/ \
Vol. 86
2830 2220 1535
solvents have only a small effect. These results are readily understandable on the basis of the proposed reaction scheme. Butylene oxide would not be expected to enter into the propagation reaction since Gee and his co-workers have shown that secondary alkoxides do not add a t a significant rate to substituted oxirane ring carbon atoms. However, it can rearrange to the unsaturated alkoxide and initiate new chains, and since its concentration remains high, new chains will continue to form even toward the end of the polymerization. Furthermore, its presence will slow the propagation reaction 4 by dilution of the propylene oxide. The net result is that the higher concentrations of butylene oxide lead to lower molecular weight products with higher double bond concentrations. 0
RO-
/ \
+ HCHzCH-CH?
+
1-+
[
.:O, RO~~~H~..CHz-CH-CH2 ROH
Br
rio-
+ H C H ~I C H C H+ ~
+ CH~=CHCHIO-
1-+
R O . . . H . . . C H ? LBr- ~ H C H ~-+ ROH CH2==CHCH, Br-
+
(IO) R L Letsing-er, J. G . T j a y n h a m , and l
