Oddon, A., Wylde, J., Bull. SOC.Chim. France 1967, 1603. Reppe, W., Nicolai, F., Ger. Patent 631,016 (June 13, 1936). Schoenberg, A., in Houben, J., Weyl, T., “Methoden der Organischen Chemie,” Band IX, pp. 149 ff, Georg Thieme Verlag, Stuttgart, 1955. Schulz, H. S., Freyermuth, H . B., Buc, S. R., J . Org. Chem. 28, 1140 (1963); and references cited. Snyder, H., Stewart, J., Ziegler, J., J . A m . Chem. SOC. 69, 2672 (1947).
Staudinger, H., Siegwart, J., Helu. Chim. Acta 3, 833 (1920). Swern, D., Billen, G. N., Scanlan, J. T., J . A m . Chem. SOC.68, 1504 (1946). Umbach, W., Mehren, R., Stein, W., Fette-SeifenAnstrichmittel 71, (KO. 3), 199 (1969). Van Tamelen, E. E., J . A m . Chem. SOC.73, 3444 (1951). RECEIVED for review February 26, 1969 ACCEPTED July 19, 1969
AROMATIC ACID ANHYDRIDES BY A DIRECT OXIDATION PROCESS M I C H A E L
A .
T O B I A S
A N D
W A R R E N
W .
K A E D I N G
Research and Development Laboratories, Mobil Chemical Co., Edison, N . J . 0881 7 A novel process for the preparation of aromatic anhydrides involves direct oxidation of the corresponding methyl-substituted aromatic hydrocarbon with oxygen in the liquid phase at atmospheric pressure. Aromatic anhydrides have been produced from the xylenes and mesitylene in 20% yields. Coproducts are the corresponding monobasic acids and substituted benzyl esters.
ANHYDRIDES of carboxylic
acids possess physical and chemical characteristics which make then more attractive industrial intermediates than their parent carboxylic acids. They are usually lower melting and are more soluble in common solvents than their corresponding free acids, and when used for the production of polyesters they react with alcohols with a significant enhancement in rate and produce only half as much water as the carboxylic acids from which they are derived. Acetic anhydride is the only compound of this class which is prepared on an industrial scale by a liquid phase process. Ketene, prepared from acetic acid or acetone, reacts with acetic acid to form the anhydride, or an alternate method is used which involves the autoxidation of acetaldehyde. A method has now been developed for the preparation of aromatic acid anhydrides by the direct reaction between certain methyl-substituted benzenes and oxygen. The corresponding monoaldehyde, benzylic ester, and monocarboxylic acid are the major coproducts produced during the oxidation process. Since the monoaldehyde and benzylic ester can be easily autoxidized to the monocarboxylic acid, production of aromatic anhydrides by this method can be visualized as an industrial process integrated with the production of the corresponding free acid. Experimental
General Procedure. The oxidation procedure and method of product isolation were identical for each reaction. The hydrocarbon and catalyst were charged into a tubular glass reactor and brought quickly to reflux. Oxygen was introduced through the bottom of the reaction vessel, and any water formed was removed by a Dean-Stark trap. At the end of the desired oxidation period, the 420
I & E C PRODUCT RESEARCH A N D DEVELOPMENT
reaction mixture was drained from the reactor and allowed to cool overnight. Crystalline acids were removed, washed with pentane, and dried t o constant weight. Any dissolved acid remaining in the hydrocarbon oxidate was removed with aqueous sodium bicarbonate. The acids were determined by gas chromatography of their methyl esters. The remaining neutral hydrocarbon was washed with dilute hydrochloric acid and dried. Unoxidized hydrocarbons, as well as any aldehyde produced, were isolated by distillation a t reduced pressure. The high boiling residue was again washed with aqueous bicarbonate, and an aliquot treated with a reagent prepared from boron trifluoride and methanol. Anhydride and benzylic ester present in this residue were converted to products which were easily analyzed by gas chromatography.
.
0 0
II I1
ArCOCAr
+
I/
C H S O H 2 2ArCOCH3
0
II
o
ArCHzOCAr + C H 3 0 H
BF3
0
/I
ArCOCH3 t
ArCHzOH + ArCH20CH3 All reaction products were identified by infrared and nuclear magnetic resonance spectroscopy. EXAMPLE1. A tubular, glass reaction vessel, 4 x 100 cm., was charged with 500 grams of p-xylene and 100 p.p.m. of cobalt 2-ethylhexanoate. The solution was brought t o reflux and oxygen was introduced through the bottom of the reactor a t a rate of 500 cc. per minute. Vapors which distilled were condensed, the water was separated and removed, and the p-xylene was returned
to the reactor. Vigorous reflux was maintained a t atmospheric pressure and 138" to 150°C. for 4 hours. ,Upon cooling overnight, 44.6 grams of p-toluic acid crystallized from the reactioin mixture. Additional p-toluic acid (10.5 grams) was removed by washing with a saturated aqueous solution of sodium bicarbonate. The neutral p-xylene solution was, washed with dilute hydrochloric acid, and the unreacted p-xylene removed by distillation. Further distillation yielded 13.9 grams of p-tolualdehyde. Upon skanding, 9.7 grams of crude p-toluic anhydride crystallized from the neutral high boiling residue. The crude anhydride was shown to contain 2.9 grams of p-methylbenzyl p-toluate. The remaining mother liquor was shown to contain an additional 12.3 grams of p-toluic anhydride and 9.6 grams of p-methylbenzyl p-toluate. EXAMPLE2. An identical oxidation of neat p-xylene was continued for 15 hours. Each hour, after the second hour of reaction, 100-gram aliquots were removed from the reactor and replaced with 100 grams of fresh p-xylene. The material removed from the reactor, as well as the material remaining in the reaction after 15 hours, was processed as described in Example 1. A total of 511.6 grams of p-toluic acid, 41.6 grams of p-tolualdehyde, 86.6 grams of p-methylbenzyl p-toluate, and 135.0 grams of p-toluic anhydride was; produced. EXAMPLE3. The glass reactor was charged with 540 grams of o-xylene and 120 p.p.m. of cobalt 2-ethylhexanoate. Oxygen was allowed to bubble through the hydrocarbon a t a rate of 500 ml. per minute for 7 hours a t 144" to 150". Vapors which distilled were condensed, the water was separated and removed, and the o-xylene was returned to the reaction vessel. Of the o-xylene charged, 350.5 grains were recovered. The reaction products were separated into neutral and acidic fractions, and the acidic fraction (114.3 grams) was shown to be o-toluic acid. The neutral material was composed of 10.2 grams of o-tolualdehyde, 7.0 grams of phthalide, 36.0 grams of o-methylbenzyl o-toluate, and 45.0 grams of o-toluic anhydride. Results
When pure oxygen was bubbled through refluxing p-xylene containing trace amounts of soluble cobalt salts, and any water formed in the oxidation was removed, an appreciable concentration of p-toluic anhydride was
discovered among the reaction products. Other major products were p-tolualdehyde, p-toluic acid, and p-methylbenzyl p-toluate. Small amounts of terephthalic acid, p-carboxybenzaldehyde, and terephthaldehyde were also isolable. The preparation of aromatic anhydride by this method was readily extended to the other isomeric xylenes and to mesitylene. Table I summarizes the results and experimental conditions for typical reactions. The observed difference in the relative rates of oxidation of these hydrocarbons is in general agreement with that found in other studies in our laboratories. There is a direct correlation between these rates and the amount of free monoaldehyde and monoacid isolated as one proceeds from o-xylene to mesitylene. The slight reduction in the amount of monoacid isolated in going from p-xylene to mesitylene may be due, in part, to oxidative decarboxylation (Starnes, 1966) taking place a t the higher temperature of the mesitylene reaction (Equation 1). The isolation of 2,3',4,5',6-pentamethylbiphenyl (11) from this reaction is good evidence for this supposition (Equation 2). coon
j &
,
t
c0+3
-
-
OPCO.
+
C f 2
+
Hf
+B
(2)
mesitylene
The production of anhydride and benzylic ester is essentially constant throughout the series. Having demonstrated the generality of this technique for oxidizing the xylenes and mesitylene directly to their anhydrides, two approaches were used in an effort to increase the ratio of anhydride to monocarboxylic acid produced in these oxidations. First, assuming p-tolualdehyde to be an essential intermediate in the production of p-toluic anhydride, p-xylene was oxidized in the presence of a large initial excess of its monoaldehyde. Table I1 compares a result obtained from an oxidation of neat p-xylene with that from an oxidation of p-xylene plus p-tolualdehyde. I n the latter case, the production of p-methylbenzyl p-toluate (50%) and of p-toluic anhydride (18%) increased. However, most of the initially added p-tolualdehyde had been oxidized to p-toluic acid and not incorporated into either bimolecular product.
Table 1. Oxidation of Xylenes and Mesitylene Products, G.
React ion Reacted
Relatiue Rateh
Monoaldehyde
Monoacid
7
49.2
1.12
28
130
26
27
2,3',4,5',6pentamethyl biphenyl (5.5)
138--513
7
43.8
1.0
23
140
34
51
Terephthalic acid (8.5)
140-580
7
38.5
0.88
15
126
31
47
Isophthalic acid (6.4)
144-513
7
35.0
0.80
10
114
36
45
Phthalide (7.0)
Temp, Substratea
A
&
Time, hr.
160-713
c.
C/c-
Benzylic ester Anhydride
Others (g)
Reactor charged with 500 grams of hydrocarbon. Based on hydrocarbon consumption after 7 hours
VOL. 8 NO. 4 DECEMBER 1969
421
~~~
Table 11. Effect of Added p-Tolualdehyde on a p-Xylene Oxidation
~
500.0 415.3 84.7 0.0 13.9
500.0 416.6 83.4 100.0 59.6 54.3 (0.452 mole) 110.6 (0.812 mole) 22.3 (0.177 mole) 18.8 (0.157 mole)
...
p-Toluic acid isolated p-Toluic anhydride isolated p-Methylbenzyl p-toluate
55.1 (0.405 mole) 19.1 (0.150 mole) 12.5 (0.104 mole)
@ @ CH3
CH,
+
1/20i-
o\
cri3
+
h20
(3)
cw3
Ill
422
Total p-xylene charged Total p-xylene recovered Total p-xylene reacted
1800.5 1139.1 661.4 (6.24 moles)
4290.3 3664.2 626.1 (5.91 moles)
p-Tolualdehyde isolated
41.6 (0.35 mole) 511.6 (3.76 moles) 86.6 (0.36 mole) 135.0 (0.53 mole) 0.264 1.56
122.0 (1.02 moles) 309.7 (2.28 moles) 123.2 (0.51 mole) 163.3 (0.64 mole) 0.525 1.32
138 -50 94.4
13-40 95.2
Anhydrideiacid ratio Anhydride! ester ratio
H3
@ @
Reaction B
p-Toluic anhydride isolated
The identification of significant concentrations of anhydride in our hydrocarbon oxidation system prompted us to try to gain some insight into the mechanism of the reaction. Three approaches were examined. The possibility of direct oxidation of p-methylbenzyl p-toluate to p-toluic anhydride was investigated initially (Equation 3). The presence of I11 in the system might be rationalized as a consequence of a reaction between the corresponding alcohol and acid, possible oxidation intermediates, since the reactor was designed to remove water rapidly, conditions which should favor ester formation. /
2OOg. 0.75 hr.
Reuction A
p-Methylbenzyl p-toluate
Discussion
O *c
100g. hr.
Acidsb isolated
A second approach toward increasing the ratio of anhydride to acid isolated involved periodic removal of a portion of the oxidation mixture from the reaction vessel, followed by its replacement with fresh hydrocarbon. Since the formation of water as an oxidation product is a yieldlimiting factor for the production of anhydride in this system, any suitable technique to remove anhydride continuously, as well as water, from the reaction zone would decrease the degree of anhydride hydrolysis t o free acid. Table I11 compares the oxidation products produced in two different oxidations of p-xylene, carried on for 15 hours. In the first experiment (Reaction A ) , 100 grams of material were removed from the reaction vessel every hour, while in a second experiment (Reaction B), 200 grams of material were replaced with p-xylene every 45 minutes. In the latter case, fresh hydrocarbon was added as soon as p-toluic acid began to precipitate from the oxidizing mixture. In Reaction B, the anhydride-acid ratio increased 100% over that obtained in Reaction A. Concurrent with this fact are the substantial increases in the quantity of isolated p-tolualdehyde and p-tolyl p-toluate and the significant decrease in the amount of p-toluic acid produced. The isolation of larger quantities of aldehyde and ester in the second case results from their more rapid removal from the reaction vessel before further oxidation t o p-toluic acid could occur. The increase in the anhydride-acid ratio is a result of this fact, as well a5 the decreased opportunity for anhydride to undergo hydrolysis to 2 moles of p-toluic acid.
ch;o\CH~
~~~~
W i t h d r a d Hate"
Oxidation of Oxidation of p-Xylene plus A'eut p-Xylerle, G. p-TolualdehSde, G. p-Xylene charged p-Xylene recovered p-Xylene reacted p-Tolualdehyde charged p-Tolualdehyde isolated p-Tolualdehyde reacted
~
Table Ill. Effect of Withdrawal Rate on a p-Xylene Oxidation
l & E C PRODUCT RESEARCH A N D DEVELOPMENT
Reaction temperature, C. Material balance, ( i
'Reactor initially charged zcith 550 grams of p-xylene. p-toluic acid.
*>
95'/;
However, all attempts to oxidize benzyl benzoate (a readily available model compound), under similar conditions of reactions, failed to give benzoic anhydride. Oxidations of the neat ester, as well as oxidations in a series of hydrocarbon solvents, yielded benzoic acid. The reported cobalt-catalyzed oxidation of p-methylbenzyl p-toluate (Katzschmann, 1966) to mono-p-tolyl terephthalate, terephthalic acid, and p-toluic acid (Equation 4) is in agreement with our results.
Further attempts to elucidate the mechanism of anhydride formation were directed a t the possible simple dehydration of p-toluic acid. Experiments along these lines showed that p-toluic anhydride was not formed from p-toluic acid when the latter was placed in refluxing tertbutylbenzene in the presence of cobalt salts and excess oxygen, or when it was heated in refluxing p-xylene in the presence of cobalt salts and nitrogen. The reaction between perbenzoic acid and p-tolualdehyde was also briefly examined. The uncatalyzed interaction of these compounds, in refluxing benzene or p-xylene, afforded no detectable 4-methylbenzoic anhydride. However, when this reaction was carried out in the presence of trace quantities of cobalt(I1) 2-ethylhexanoate, characteristic infrared anhydride absorption peaks were observed in the product mixture (Equation 5).
(5)
This last observation, coupled with the findings of Brihta and Vrbaski (Brihta and Vrbaski, 1956; Vrbaski and Brihta, 1952, 1954) concerning the autoxidation of aliphatic aldehydes to aliphatic anhydrides, supports the credibility of a mechanism for aromatic anhydride formation
involving the reaction of an aromatic peracid with an aromatic aldehyde. Thus, the kinetic data of Vrbaski and Brihta led them t o formulate the following reactions for the formation of aliphatic anhydride (Equation 8) and showed that this reaction is strongly catalyzed by cobalt salts and proceeds very slowly in their absence.
aliphatic and aromatic primary hydroperoxides with their respective aldehydes.
0
I/
I1
+
RC. 0
-+RCOO*
0
II
RCOO.
II + RCH
0
/I
RCOOH
-+
RCOOH
+ RCH
I/
+ RC.
(7)
+ H20
(8)
0 0
I/ II
/I
+
-+
I
CH3C-O-O-CHCH3 0 0
RCOCR
CH3COCCH3 + HzO
(6)
0
0
0
II
OH
II
I1 II
II
0 2
0
II
CH3COOH + CH3CH
0
0
0
A pair of reactions similar to that shown in ,Equation 8 can, of course, be postulated to account for the formation of p-toluic anhydride and p-methylbenzyl p-toluate during the cobalt-catalyzed autoxidation of p-xylene.
l i
+
(11)
An alternate mechanism has been proposed to account for the formation of p-toluic anhydride (Katzschmann, 1966) which involves a combination of aroyl peroxy and benzoyl radicals to give the benzoyl peroxide (VII), which subsequently decomposes to give anhydride.
CH3
CH3
CH3
CH3
VI1
However, it seems that the oxygen concentration in our system should be sufficiently high to oxidize most benzoyl radicals rapidly to benzoyl peroxy radicals and therefore leave too low a concentration to account for the substantial amount of anhydride found in our reaction mixture. Conclusions
More specifically, the formation of p-toluic anhydride or p-methylbenzyl p-toluate can be envisoned as proceeding through hydroperoxides IV or V, by way of a mechanism involving considerable polar (path A) or radical (path B) character.
The cobalt-catalyzed oxidation of neat alkyl aromatic hydrocarbons, in a system which allows for rapid and continuous removal of water, affords up t o a 20% yield of the corresponding aromatic anhydride. The use of a cyclic process, in which products are removed from the reaction zone soon after their formation, affords a means of raising the ratio of anhydride to monocarboxylic acid produced. I t is suggested that the aromatic anhydrides are produced from the reaction of an aromatic peracid with an aromatic aldehyde. literature Cited
IV,
v,
x x
:0 : H2
It
Y CH,@-:-O-:@-CH3
+
i
+
O
+
HE
CH3&g-O-!-@CH3
+
H20
Although no known experimental evidence exists to demonstrate the intermediacy of IV in the formation of p-toluic anhydride, the hydroxyperoxide (VI) has been shown to play a significant role in the reaction between peracetic acid and acetaldehyde on the way to forming acetic anhydride (Bawn and Williamson, 1951; Kagan and Lobarsk, 1935; Phillips et al., 1957; Starcher et al., 1961). I n addition, the inteymediacy of hydroxyperoxides analogous to V has been demonstrated during the formation of aliphatic (Durham et al., 1958; Wurster et al., 1958) and aromatic esters (Farrissey, 1961) in the reaction of
Bawn, C. E. H., Williamson, J. B., Trans. Faraday SOC. 47, 721 (1951). Brihta, I., Vrbaski, T., Croat. Chem. Acta 28, l a (1956); C A 50, 16318 (1956). Durham, L. J., Wurster, C. F., Mosher, H. S., J . A m . Chem. SOC.80, 332 (1958). Farrissey, W. J., J . A m . Chem. SOC.84, 1002 (1961). Kagan, M. J., Lobarsk, G. D., J . Phys. Chem. 39, 837, 847 (1935). Katzschmann, E., Chem. 1%. Tech. 38, No. 1, 1 (1966). Phillips, B., Frostick, F. C., Jr., Starcher, P. S., J . A m . Chem. SOC.79, 5982 (1957). Starcher, P. S., Phillips, B. Frostick, F. C., Jr., J . Org. Chem. 26, 3568 (1961). Starnes, W. H., Jr., J . Org. Chem. 31, 1436 (1966). Vrbaski, T., Brihta, I., Arhil; Kem. 24, 111 (1952); CA 49, 163 (1955). Vrbaski, T., Brihta, I., Arhil; Kem. 26, 267 (1954); CA 49, 15411 (1955). Wurster, C . F., Durham, L. J., Mosher, H. S., J . A m . Chem. SOC.80, 327 (1958). RECEIVED for review May 8, 1969 ACCEPTED October 1, 1969 Division of Petroleum Chemistry, 156th Meeting, ACS, Atlantic City, N. J., September 8 to 13, 1968. VOL. 8 NO. 4 DECEMBER 1969
423