I -
W. B. SATKOWSKI and C. G. HSU Inorganic Chemicals Division, Research Department, Monsanto Chemical Co., Everett, Mass.
Polyoxyethylation of Alcohol A mechanism is postulated for the polyoxyethylation of alcohols which is of current industrial application in nonionic surfactants
POLYOXYETHYLAiION is now very widely used in the synthetic detergent industry. This reaction has been extended in various directions through the development of nonionic detergents. The polyoxyethylation of alcohol was first reported by I. G. Farbenindustrie (7), who condensed methanol, ethanol, propanol, and 2-butanol with ethylene oxide in the presence of an acidic or basic catalyst. Later Schoeller and Wittwer (79) also successfully condensed a long chain alcohol such as octadecyl alcohol with ethylene oxide. Since then, many patents have been issued, but no systematic study of this reaction is reported in the literature. Therefore, it seemed desirable to investigate this reaction in detail in order to obtain a reasonable mechanism.
Experimental
Starting Materials. Oxo alcoholsiso-octyl alcohol, decyl alcohol, and tridecyl alcohol-were obtained from Enjay Co. (New York) and pure normal alcohols-1-pentanol, 1-decanol, and 1octadecanol-were purchased from Eastman Kodak Co. (Rochester, N. Y.), Ethylene oxide (Union Carbide Corp., New York), and reagent grade of potassium hydroxide, sodium hydroxide, potassium carbonate, sodium carbonate, and metallic sodium are all commercial products. Sodium methoxide and sodium ethoxide were prepared from metallic sodium and the corresponding absolute alcohol according to the methods of Allen ( 2 ) and Adams (7) and their coworkers. Polyoxyethylation Procedure. Apparatus similar to that described by Miller, Bann, and Thrower (74) was used. In order to minimize the varia-
tions due to techniques the same equipment and procedure were used in all experiments. The general procedure for polyoxyethylation was illustrated in a typical example as follows : Tridecyl alcohol (200 grams, 1 mole) and potassium hydroxide (2 grams, 0.036 mole) were placed in a three-necked, 2000-ml., 35/25 ball-jointed flask equipped with a gas tight stirrer, a Kjeldah1 trap, a gas inlet tube, and a thermometer. A mercury manometer was connected with the Kjeldahl trap. The whole assembly was mounted on a triplebeam balance. The system was closed, purged with dry nitrogen, vented through the Kjeldahl trap, and heated with a heating mantle. When the temperature reached 100" to 110' C. it was purged with dry nitrogen again 8 to 10 times with the stirrer running. The heating was continued. When a reaction temperature of 135" to 140°C. had been reached the stirrer was stopped, the system was vented and purged three times with ethylene oxide. Then the stirrer was started again and the weight of the whole assembly was taken. The ethylene oxide flow was adjusted to keep the pressure at 30 =t1 cm. of mercury on the system or as otherwise indicated. During the experiment the temperature of the reaction was maintained within a 5' C. range by heating with a mantle or cooling with a water bath as necessary. When the calculated amount of ethylene oxide was added the ethylene oxide flow was stopped and the reaction was allowed to
was cooled with a water bath. The system was purged with nitrogen three times when the temperature reached 100" C. At 65' C. the flask was dismounted and ready for neutralization and drying. The weight of ethylene oxide introduced was checked by weighing the flask and its contents before and after the reaction.
Discussion of Results
Several alcohols were used in this investigation. Tridecyl alcohol, obtained from the oxo process, was studied extensively. I t does not react with ethylene oxide alone at 195' to 200' C. under a pressure of 9 to 40 cm. of mercury. In the presence of potassium hydroxide, sodium hydroxide, sodium methoxide, or sodium ethoxide, tridecyl alcohol condenses smoothly with ethylene oxide. The rate of the addition of ethylene oxide was found to be practically the same at 195' to 200' C. for all the catalysts mentioned (Figure 1). However, a t 135' to 140' C., there is a difference in the rate of addition. Sodium hydroxide was less effective than the other catalysts (Figure 2). Potassium hydroxide shows a slightly delayed action at the beginning of the reaction. These phenomena indicated that the first step of the polyoxyethylation of alcohol is probably similar to that in the polyoxyethylation of phenols postulated by Boyd and Marle (4)-namely, the reaction between the ethylene oxide and the alkoxide ion, RO-. This postulation was also accepted by Schechter and Wynstra ( 7 8 ) in their study of glycidyl ether reactions. RO-
+
CHz-CH?
+
\/
following equations: ROH + R,OM ROM + R,OH ROM e ROM+ (2) R ' = H, CHs, or CzHs; M = Na or K ~
+
Reactions 1 and 2 are reversible (73). Thus, Reaction 1 can be shifted to its VOL. 49, NO. 1 1
NOVEMBER 1957
1875
350T
3507
I
--
300--
dr
0
250-n 4
w
n
2 200--
MOLE OF C13-ALC. USED = I MOLE OF CATALYST USED '0.036
0 W
z W
J
MOLE OF C13-ALC. U S E D * I MOLE OF CATALYST USED 50.036
150--
I-
W
m A e
Na NaOCH3 NaOC2H3 X KOH NaOH
LL 0
$ loo-P
s 0
2 3 T I M E (hr.)
I
5
4
Figure 1. Effect of catalyst on the rate of polyoxyethylation at 195" to 200" C.
4
0 0
I
2 3 TIME (hr.)
5
Figure 2. Effect of catalyst on the rate of polyoxyethylation at 135" to 140" C.
will then be expected to be lower in the case of sodium hydroxide than in sodium methoxide or sodium ethoxide, as shown in Figures 2, 3, and 4. At the high reaction temperature this difference between the various catalysts becomes less important and they will be expekted to act similarly. I n comparing potassium hydroxide and sodium hydroxide
right by a complete removal of R'OH. Since methanol and ethanol have a lower boiling point than water, they can be more easily removed a t a moderate temperature. Also, because of the greater basicity of CH30- and C2H60in comparison with HO-, Reaction 1 will go faster and to further completion. Therefore, the concentration of ROM
4
according to Richards and Rowe (77), potassium hydroxide has a higher basicity than sodium hydroxide. Hence, a t a lower temperature the ionization of ROM will be expected to go toward further completion for the potassium salt than for the sodium salt. Also, the reversible reaction between potassium hydroxide and tridecyl alcohol explains
350
30C
-
d
0
n
250 n U
w
n
x0 20c w
z
w
J
150
MOLE OF C13-ALC. USED = I MOLE OF KOH USED = 0.036
I-
Y LL
0
5
100
1.
105~-110~C 105°-1100C
2.
135O -140% 135°-140%
3. 165°-1700C
OLE OF C 1 3 -ALC. USED = I MOLE OF CATALYST USED e 0.036 H3 AT 195*-2OO0C
4. 195°-2000C 195' - 2 0 0 %
H j AT 135°-140'C
50
H3
C
I
I
2 3 T I M E (hr.)
4
5
(Figure 3. Effect of temperature on rate of polyoxy.ethylation with KOH as catalyst
1 876
INDUSTRIAL AND ENGINEERING CHEMlSTRY
0
I
2 3 T I M E (hr.)
AT
4
105*-110~ C
5
Figure 4. Effect of temperature on rate of polyoxyethylation with N a O H and NaOCH3 us catalyst
POLYOXYETHYLATION OF ALCOHOL
OF C13-ALC. USED OF KOH USED:
MOLE MOLE
- 0.018 - 0.036 3 - 0.072 I
2
0.143
4-
.,.
0
I
4
2 3 T I M E (hr.)
5
Figure 5. Effect of KOH concentration on rate of polyoxyethylation at 135' to 140' C. the delay in the reaction a t the beginning of the polyoxyethylation. As the tridecoxide ions are slowly condensed with ethylene oxide the equilibrium of this reversible reaction maintains the constant concentration of tridecoxide ions while the concentration of the oxyethylated tridecoxide ions is increasing. Because of this increase in the total concentration of reactive ions-namely, tridecoxide ion and oxyethylated tridecoxide ion-the rate of condensation becomes accelerated. Similarly, this phenomenon was also observed with sodium hydroxide. Potassium carbonate has been applied successfully as a catalyst for the polyoxyethylation of acids ( 3 ) . However, it was found to be unsatisfactory in catalyzing the polyoxyethylation of tri-
decyl alcohol. Karabinos, Bartels, and Kapella (9) also found this same result in the polyoxyethylation of other alcohols. This is to be expected because the reaction between alcohol and potassium carbonate or sodium carbonate will not readily form the desired alkali alkoxide as the alkoxide ion is more basic than the carbonate ion. T o further substantiate the necessity of the tridecoxide ion in this polyoxyethylation reaction metallic sodium was tried instead of sodium hydroxide. Ethylene oxide was introduced into the alcohol solution after all the metallic sodium was reacted and the hydrogen had been vented out of the system. The rate of the reaction was the same as with the others. The only products formed from the reaction of metallic sodium and tridecyl alcohol are sodium tridecoxide and hydrogen which was liberated from the system before polyoxyethylation. Therefore, the tridecoxide ion is the important reacting species in the polyoxyethylation according to the following equations: I
.
+
2C18Hz70H 2Na+ 2C13Hz70Na C13H270Na
e
C13H270-
+ HZT
+ Naf
(3)
(4)
As the temperature of the reaction was an important factor in the rate of the addition of ethylene oxide, several temperatures were examined with different catalysts (Figures 3 and 4). T h e rate of the addition of ethylene oxide increases with the reaction temperature. However, their relationship is not linear. The rate of the addition of ethylene oxide a t the same temperature increment is greater a t higher temperatures than a t lower temperatures. The disappearance of the initial delayed reaction when potassium' hydroxide and so-
dium hydroxide were used at high temperature can be explained by the findings of Engel ( 5 ) , Lescoeur (72), Williams and Bost (27)-namely, that the reaction between alcohol and potassium or sodium hydroxide is completed more rapidly a t high temperatures. The concentration of catalyst is also important (Figure 5). The rate of the reaction increases with the catalyst concentration. However, as the concentration of potassium hydroxide was progressively doubled from 1.8 mole yoto 3.6, 7.2, and 14.3 mole %, the rate of the reaction was more sensitive a t low concentrations than a t high concentrations. This indicates that the ethylene oxide molecule reacts with alkoxide ion rather than the nonionized metallic alkoxide, because the degree of ionization of metallic alkoxide is higher a t lower concentration compared with that a t higher concentration. According to the mass action law the pressure of ethylene oxide will probably have some effect on the rate of the reaction because the pressure of ethylene oxide is directly proportional to the concentration of ethylene oxide present in the system. However, in this low pressure study the effect of pressure was found not to be great. Figure 6 shows that difference is very small in the reaction rate a t 135' to 140' C. between 7.5 and 15 cm. of mercury. The speed of the reaction definitely increases as the pressure is raised to 30 and 45 cm. of mercury. However, in higher pressure studies the rate of the addition of ethylene oxide was found to increase considerably with the increasing of ethylene oxide pressure (7 7). The initial reaction species is believed to be alkoxide ion. Therefore, the structure of the alkoxide ion will have some effect on the rate of polyoxyethylation. Hence, a difference in speed be3 5 0 ~
300
-
300.-
s-
D
0
g 250-
0
g 250-
P
a
0
0 W
W
g 200--
2 200-
0
2 w
W W
1
c
150.-
$
-
0 LL
+
100-
'"-v _0.
c
MOLE O F C13-ALC. USED = I MOLE OF KOH USED 0,036 p MANOMETER READING (H$)
I
I = 45
=
2
3
I-
E
%
50t/
l5cm Z5cm
a
0 L
MOLE OF "OXO"-ALC. USED = I MOLE OF KOH USE0 0.036
cm
2 = 30cm 3 4
-
4
5
T I M E (hr.)
Figure 6. Effect of ethylene oxide pressure on rate of polyoxyethylation at 135' to 140' C.
0
3
150-
5
100-
50 --
5 - 00 '
I TIME (hr.)
T I M E (hr.1
Figure 7. Effect of oxo-alcohols on rate of polyoxyethylation at 135' to 140' C.
Fiaure 8. Effect of normal alcohols on- rate of polyoxyethylation at 105' to 1 10' c. VOL. 49, NO. 11
NOVEMBER 1957
1877
tween two alcohols would be expected. From the results shown in Figures 7 and 8 a generalization about the effect of structure of alcohol can be drawnthat is, in a homologous series the rate of the polyoxyethylation of an alcohol decreases as the carbon chain Iength in the alcohol increases.
Based on the data obtained in this investigation the authors postulated the following mechanism for the polyoxyethylation of alcohol using potassium hydroxide as the illustrating catalyst. I t consists of a series of consecutive S.y2 type reactions (8, 75):
+ KOH + ROK + HzO k kl
-1
k2
ROK k
RO-
RO-
+K+
-2
+ CHz-CH2
k0-E
+
\0/
R ( OCHzCH2)O-
+
k’-”
R(OCHzCH2)OROH R(OCHzCH2)OH R(OCH2CHz)O-
+ RO-
+ CHgCHe + ki-E
\/
0 R(OCH2CH2)20-
2
+
R(OCH~GHZ)ZO- ROH R(OCHzCHz)20H
+
+ ROkz
-1
R(0CHzCHz)zOR(OCHzCHz)OH+ R(OCHzCH2)zOH R(0CHzCHz)OR(OCH2CHz)zO-
+
+ CHzCH?+ k2-E
\0/
R( OCH2CH2)aO
+
acidic than the corresponding monoalkyl glycol ether (6). Alcohol unlike phenol will not give a quantitative yield of monoalkyl glycol ether during the oxyethylation (14, 20). After one unit ka-z R(OCH2CHz)aO- R(0CHzCHz)zOH --+ of oxyethylene joined to the alcohol molecule or alkoxide ion the close enR(OCHzCH2)30H R( 0CHzCHz)zO- vironment around the reacting center (--OH or -Ow) becomes very similar. ka-~ The reactivity of these molecules or ions R(OCH2CHz)zO- CH2CH2 --+ should then be the same. Experimentally it was found to be true. Figure 9 0 shows that the ethylene oxide consumpR( OCH2CHg)40 tion (grams per minute) became constant after 8 or 9 moles of ethylene oxide per mole of alcohol were introduced. The low consumption at the beginning 0 of the reaction is due to the existence of R(OCHzCHz),Othe less active unreacted alcohol.
ka-o
R(0CHzCHz)qOROH --+R(OCHzCHz)3OH
+ RO-
ka-1
+
+
+
Proposed Mechanism
ROH
+
R(OCHzCH2)aO- R(OCHzCHZ)OH+ R(OCHzCH2)aOH R( 0CHzCHz)O-
+
\/
Hence the polyoxyethylation of alcohol is a complicated reaction. The product obtained from this reaction is a mixture of monoalkyl poly(ethy1ene glycol) ethers containing various numbers of oxyethylene units. Karabinos and Quinn (70) reported that the amount of the various ethers closely followed the Poisson distribution law. So all the reactions listed above are in competition with each other. The extent of these reactions will be determined by the law of probability and dependent on the ease of the reaction and also the concentrations of different molecules and ions present. The over-all rate of the reaction will therefore be governed not only by the speed of the anion attacking ethylene oxide molecule ( k , - E) but also by the ease of the anion to combine with the acidic hydrogen in the neutral molecule to form a new anion ( k % - 0 and K, Although no kinetics data are available for this reaction it is believed that like in the polyoxyethylation of phenol (76) most of the rates are equal under a given condition and can be grouped as follows: %?),
k1-o = k2-o kz-1 = k3-2 k1-E = k2-E k0-E f ~ L - E
,........ . , k,,-o = k3--1 = . . . . = k,-,, = ka..~ = . . . . . . = kn-E
However, the alcohol is usually less
a
0
W
i
5
0
d
i
6 4 4 k
+
Q b Ib
1’1
112
I5 Ib
I5
16
MOLES OF ETHYLENE OXIDE ADDED
Figure
1 878
The authors are indebted to F. T. Lense, W. B. Bennet, and R. D. Swisher for their friendly criticism and suggestions. literature Cited (1) Adams, R., Kamm, R. M., “Organic Synthesis,” 1, p. 250, Wiley, New York, 1941. ( 2 ) Allen, C. F. H., Abell, R. D., Normington, J. B., Ibid., p. 205. ( 3 ) Ballun, A. T., others, J . Am. Oil Chemists’ Soc. 31, 20 (1954). ( 4 ) Boyd, D. R., Mark, E. R., J . Chem. Soc. 105, 2117 (1914). ( 5 ) Engel, hi., Comfit. rend. 103, 156 (1886). ( 6 ) Hine, J., Hine, M., J . Am. Chem. SOG. 74, 5266 (1952). ( 7 ) I. G. Farbenindustrie, Brit. Patent 271,169 (Feb. 22: 1926). ( 8 ) Ingold, C. K., “Structure and
Mechanism in Organic Chemistry,” p. 341, Cornel1 Univ. Press, Ithaca, N. Y . , 1953. ( 9 ) Karabinos, J. V., Bartels, G. E., Kapella: G. E., J . Am. Oil Chemists’
SOC.31. 419 11954). (10) Karabinos, 3. V., Quinn, E. J., Ibid., 33, 223 ( 1 956). (1 1) Koegle, J. S., private communication
frcm this laboratory.
(12) Leecoeur, H., Comfit. rend. 121, 692 (1895). (13) hladsen. E. H., J . Chem. Soc. 103, 966 11913). (14) -idllrr. ’S.~ A,.~Bann. ~ B..~ Thrower. ~ R . D.,-Ibid.,’l95, 3623 ( I 950). (15) Natta, G., Mantica, E., J . A m . Chem. SOC.74, 3152 (1952). 116) Patot, F., Cremer, E. Bobleter. O., Monatsh. Chem. 83, 322 (2952). 117’1 Richards, T. W., Rowe, A. IV.. J . Am. Chem. Soc. 44. 684 11922). Schechter. L., Wynstra, J., IND.ESG. CHEM.48, 86 (1956). Schoeller, C., Wittwer, M. (to I. G. Farbenindustrie), U. S. Patent 1,970,578 (Aug. 12, 1934). Weibull, B., h-ycander, B., Acta. Chem. Scand. 8, 847 (1954). Williams. D.. Bost. R . Mi.. J . Chem. Phys. 4; 251 (1936). \ - . I
w
w
Acknowledgment
9. Rate of ethylene oxide consumption at 135” to 140” C.
INDUSTRIAL A N D ENGINEERING CHEMISTRY
1;
RECEIVED for review September 15, 1956 ACCEPTEDApril 24, 1957 Division of Industrial and Engineering Chemistry, 130th Meeting? ACS, Atlantic City, N. J., September 1956.
