pressure of water when plotted in like manner. Through the temperature range 20 to 60°, a straight line with log p and l / T as variables reproduces the experimental pressure values for either variety of water with error riot in excess of 0.5%. In Table I1 are shown values for relative pressures in systems A and B, in computing which we employed the ordinary water pressures fouiid in “International Critical Tables,”22the deuterium water pressures of Miles and Menzies,lG the pressures for system A given by Collins and M e n ~ i e s ‘and ~ the various pressures for system B as reported in Table I. ‘r.4EI.E
11
RELATIVE PRESSURES IN PERCENTAGE FOR SYSTEMS A AND R COMPUTED FROM DATAOF VARIOUS OBSERVERS I’, “ C
20 25 30 35 39
40
A B B C and M. M.and M.S and R . 30.5 32.1 29.2 32.9 35.0 32.1 35.5 3i.7 35 6 38.3 40.6 40.6 43.0
43.6
50
43.2 47.6
50.3
42 ’i 50.5
60
54.6
57 7
59.3
B
B
Bell
P.and 5
31.6 33.4 39.9
32.3 33.4
41.7
These values are graphed in the manner de(22) “I. C. T.,” Vol. 111, 1926,p. 212.
[CONTRIBUTION FROM
THE
scribed in Fig. 1, where we have added also our lines for the relative pressures in systems C and D. If the modest extrapolation, which we have shown in Fig. 1, from the values of Partington and Stratton be allowable by reason of their utilization of four significant figures in expressing pressure, we can state that the relative pressure lines for system B as derived from the experimental data of all the o t l m clhsrrvers intersect the relative pressure line for system A as graphed ill every case, contrary to our finding in this regard. It will be seen that, although it is true that the relative pressure lines of the other workers pass close to the same point at 2 5 O , this is by no means the case a t other temperatures.
Summarv Dissociation pressures in the systems consisting of deuterium water vapor and cupric sulfate penta- and trideuterate and also strontium chloride hexa- and dideuterate are reported in the temperature range 2.3 to SOo, and compared with analogous hydrate pressures. The related heat values are deduced and compared with those of others, and attention is called to certain discrepancies of our values from those obtained in three investigations by other workers. PRINCETON,
RECEIVEDNOVEMBER 15, 1937
N. J.
GEORGE HERBERT JONES LABORATORIES OF THE UNIVERSITY OF CHICAGO]
The Amine Catalysis of the Dealdolization of Diacetone Alcohol BY F. H. WESTFIEIMER AND HERZL COHEN
In their study of the dealdolization of diacetone alcohol catalyzed by amines, John Miller and Kilpatrick’ observed that, at constant buffer ratio, the rate of the reaction increased with increasFrom this, they condensation is an example using these words in the .2 Previously, however, Frenchahad found that, in phenol-phenolate ion by the hydroxyl ion buffers, the rate is eontro concentration alone. T esent work was undertaken to clear up this discrepancy. se, the rate of the deddolization
chloride, by dimethylamine and dimethylammonium chloride, by trimethylamine and trimethylammonium chloride, and by triethylamine and triethylammonium chloride, all at 18’ and a t a constant ionic strength. The work of John Miller and Kilpatrick was verified and amplified with the first two amines. Molecular trimethylamine and molecular triethylamine, however, were found to be completely without effect on the rate of the reaction. The experimental results, together with the conclusions to be derived therefrom are presented in this paper. Experimental Apparatus and Edetkod.-The rate of the reaction is easily and accurately measured dilato~netrically.~The instrument employed w a s similar to that of Br6nsted and (4)
Koelichen, 2.phyn’k. Ckm.,m,129 (1800).
Jan., 1938
h I N E CATALYSIS OF THE bEALDOtIZATION OF
Guggenheim,6 using mercury t o seal the stopcocks. Since the half time of the reaction is great, the dilatometer was allowed t o remain in the thermostat for an hour or more before readings were begun; thus a chamber for adjustment of temperature was not employed. The thermostat was maintained at 18.05 * O.O0lo. The velocity constants were calculated by the method of Guggenheim,6 using decadic rather than natural logarithms. Materials.-Eastman diacetone alcohol, freshly distilled for each run, boiled a t 65-67” a t 20 mm. Eastman methylamine hydrochloride was recrystallized from nbutyl alcohol.’ Kahlbaum trimethylamine hydrochloride, when treated with 10 mole per cent. of sodium nitrite and hydrochloric acid, yielded no nitrosoamine on steam distillation. Commercial Solvents Company aqueous dimethylamine was converted into the hydrochloride. This salt was completely soluble in dry chloroform,8 from which solvent it was crystallized by addition of acetone. In each case, the hydrochloride was treated with an excess of alkali, and the amine distilled into carbon dioxide-free water. Eastman triethylamine was fractionated with a n eight-bulb column, and the fraction boiling at 88.7 to 89.0’ a t 755 mm. chosen. Mallinckrodt Analytical Reagent sodium chloride was used to maintain a constant ionic strength of 0.15.
Results
DIACETONE ALCOHOL
91
TABLE I A TYPICAL VELOCITY DETERMINATION Run 9 0.05 m Methylamine .0167 m Methylammonium chloride .I333 m Sodium chloride .080 m Diacetone alcohol r = 7 hours 1,
min.
v
k = 80.4 10-6 rnin.-’ log (w - w‘) c’
log ( 8 - - 0 ’ )
15 30 45 65 85 135 155 175 195 226 245 265 285 315
81.52 79.66 77.89 75.60 73.40 68.32 66.42 64.64 62.88 60.38 58.80 57.25
46.08 45.26 44.48 43.42 42.42 40.05 39.25 38.42 37.62 36.50 35.80 35.05 55.74 34.39 53.58 33.42
1.5495 1.5366 1.5239 1.5076 1.4911 1.4482 1.4341 1.4186 1.4024 1.3780 1.3617 1.3464 1.3294 1.3045
Dev.
graph
0 83.46 46.94 1.5625 1.5608
+0.0013 1.5489 .OW6 .OOOO 1.5366 .0002 1.5241 .0006 1.5082 .0010 1.4921 .0038 1.4520 1.4355 ,0014 1.4190 ,0004 1.4028 .0004 1.3790 .0010 1.3627 - .0010 1.3464 .WOO 1.3301 .OW7 1.3056 - .0011 Average O.OOO9 or 0.2%
+
-
It had been shown previously that the reaction in the presence of alkali hydroxides is pseudounimolecular. The variations in k caused by a twen- librium. From a portion which had been neutraty-fold change in concentration of diacetone al- lized, a 95% yield of pure acetone p-nitrophenylcohol are only a few per cent., and would not in- hydrazone was obtained. validate this statement.q That this is also the With these fundamentals established, the rates case in the presence of amines is shown not only of the reaction in the presence of the amine buffers by the accuracy with which the data fit a first can be considered. The rates were measured at order equation but also by the fact that variation three or more buffer concentrations, and, except of the concentration of diacetone alcohol from for triethylamine, a t two different buffer ratios, 0.5 to 2%, keeping other factors constant, caused keeping the ionic strength constant. The slopes no appreciable change in the rate constant. A typi- of the straight lines obtained by plotting the rate cal velocity run is recorded in Table I. While the constant against amine concentration are the speaverage deviation was always of the order of a few cific rate constants for the molecular amines, and tenths of a per cent., successive runs differed as were found independent of the concentrations of much as 4%. Some wide deviations found in our amine, substituted ammonium ion and hydroxyl early work were eliminated by scrupulous cleaning ion. The intercept of these lines on the rate conof the apparatus. stant axis represents the part of the rate constant In the presence of the alkali hydroxides, acetone due to hydroxyl ion catalysis alone. The data is the only product of the reaction. It would are set forth graphically in Fig. 1and analytically certainly be anticipated that the same would ob- in Table 11. tain in the presence of amines. As a final preBy demonstrating that the slope of the line in caution, the following experiment was performed: which ammonia concentration is plotted against a reaction mixture containing 0.15 molar xnethyl- rate constant, is independent of buffer ratio, Milamine, 0.15 molar methylammonium chloride and ler and Kilpatrick proved that the catalysis was 5% diacetone alcohol was allowed to come to equi- due to molecular ammonia. The rate constant ( 5 ) Bronsted and Guggenheirn, THISJOURNAL, 49, 2554 (1927). fitted the equation kobsd. = koH-(OH-) kB(B). (6) Guggenheirn, Phil. Mag., [7] 1, 538 (1926). The results reported (Table 11) show that, as they (7) “Organic Syntheses,” Coll. Vol. I, John Wiley and Sons, Inc., New York, N. Y., 1932, p. 340. assumed, the same is true for methyl- and di(8) Knudesen, Ber., 42, 3994 (1909) methylamine. The values of k g are independent (9) Sturtevant, THISJ O U R N A59, L , 1528 (1937).
+
With trimethylamine, however, an entirely different situation obtains. Here the rate depends on the buffer ratio, but is independent of the buffer concentration. The unvarying rate is eloquent proof that the aldol condensation is not an example of general base catalysis. The rate depends on, and only on, the hydroxide ion concentration. ti. 11is the average rate constant for those buffers in which the ratio of concentration of amirie to concentration of substituted ammonium chloride is ten. When the buffer ratio is seven, the average constant is 5.80. The hydroxide ion concentrations, then, stood in the ratio of 1 to 0.70, while the rate constant stood in the ratio of 1 to 0.72, a satisfactory agreement. Had there been a specific rate constant, due to molecular trimethylamine, as large as that due to ammonia (ail even weaker base) the rate in the more concen(1 1 0.2 0.3 hfolarity of base. trated buffers mould have been almost double that Fig. 1. H I the more dilute. fhe rate coliqtants with triethylamine actually of the concentration of the amine, of the subqtistic )wa slight decrease with increasing buffer cotituted ammonium salt and of hydroxide ions. witration. While the deviation is hardly greater TABLE I1 than the experimental error, it may be real, for it AMIHZCATALYSIS OF THE DEALDOLIZATION OF DIACETONI~is quite possible that the primary and secondary ALCOROL salt effects of the triethyIammonium chloride and Concn. of SDerific ._ .. rate constant Concn. subst'ttuted Concn. of sodium chloride are not identical. k X 10" k g X lo' of ammonium gr NaCl m h - 1 mol.-' m h - 1 chloride amme Further, the absolute magnitude of the rates is Methylamine that which would be expected if the catalysis were Ix i 42.2 O.Od0 0.120 0,030 due vnly to hydroxide ions. Dividing the ob86.0 .ox5 .065 .om served rate by the catalytic constant for hy126.0 .I00 ,050 ,100 175.0 . O OH droxide ions (0.107), the hydroxide ion concen. I 5n ,150 . 133 80.4 .0167 ,050 tration of the solution is obtained. From this 117 133.0 . loo .0333 and the buffer ratio, the dissociation constant 188.0 100 . 1.50 ,0600 of the amine can be estimated. Dimethylamine The most accurate determination of the disso8.8 12.4 0.100 0.050 0.050 ciation constant of the methyl amines was made a t 18.2 ,050 .100 .IO0 2. 5 ' by I-Iarned and Owen.19 Further, Harned and 000 21.0 ,150 ,150 Robinson'l have determined the activity coeB35.7 ,117 b333 ,100 ,100 39.1 .I50 .o500 cients of these amities in the presence of sodium 49.2 ,0833 .iX7 .250 chloride, again a t 25'. Assuming that the differTrimethylamine ence, in both the primary and secondary salt ef5.71 0.00 0.0143 0.136 0,100 fects, between sodium chloride and a tertiary am5.88 .0286 .121 ,200 monium chloride is negligible, and that the activ5.80 .lo7 ,0429 .300 ity coefficients at 25' do not differ appreciably .130 8.08 .0100 .IO0 from those a t 1 8 O , 3.2 x 10-5 is calculated as the ,130 8.12 ,200 ,0200 true- value for the dissociation constant of tri8 13 120 3 00 .0.300 methylamine a t ISo. XTarned and Owen's valuc 7'1 iet h ykdili itic at 35' is 5.45 X lo+. Considering the assump0.00 0 . IUO 17.1 1). 050 0 100 tions involved and the difference in temperature, .050 16.5 .loo I200 _r
.uoo 15.8 .300 .1M) Some of thesP values art=averagt>s.
i
, 5079 (1930). 10) Hariled and Ouen 1 AI? ~ O ~ I R Y A L68, I I f i l r l l c I d : l c ~ Hot,lll,irrl i i l l , f 60. : I i 7 C m a )
Jan., 1938
AMINECATALYSIS OF THE DEALDOLIZATION OF DIACETONE ALCOHOL
93
the agreement is satisfactory, and the conclusion is strengthened that, in trimethylamine-trimethylammonium chloride b d e r s , the rate is determined by the OH- concentration alone. A similar calculation of the dissociation constant of triethylamine involves the additional SLSsumption that the activity coefficients for triethylamine do not differ appreciably from those of trimethylamine. The value obtained from our data is 3.2 X Previous investigators’* have found 6.4 X lo4and 5.65 X for the dissociation constant of triethylamine; their values are not, however, corrected by the use of activity coefficients. Likewise, following John Miller and Kilpatrick we can determine the dissociation constants of methyl- and dimethylamines from the intercepts on the rate constant axis of the plots of rate against buffer concentration. The value thus obtained and for dimethylfor methylamine is 3.2 X The corresponding constants amine 3.4 X as given by Harned and Owen are 4.38 X and 5.26 X Again considering the difference in temperature, the assumptions already cited, and the errors of extrapolation, the agreement is as good as could be expected.
action is not general base catalyzed, we are justified in concluding that the time consuming step is not the removal of the proton, but that the alcoholate ion exists in equilibrium with aldol. The slow reaction is the unimolecular decomposition of the alcoholate ion. Neglecting the reverse reaction, these steps are outlined below
(12) Bredig, 2. $&ysik. Chcm., lS, 191 (1894); Hall and Conant, TEISJOURNAL, 4*, 3047 (1927). (13) Huckel, “Theoretische Grusdlagen der organischen Chemic,” 2nd ed., Vol. I, Akademische Verlagsgesellschaft. Leipzig, Germany,
(14) Usherwood, J . Chem. SOC.,118, 1717 (1923). (15) Knoevenagel, Ber., 81, 2696 (1889). (16) Knoevenagel, Ann., Sal, 26 (1894); Worrall, THIS JOURNAL, 66, 1556 (1934) (17) Cerf, Compt. rend., 196, I084 (1932).
H I 0
+
Fast
~ - C H I C O C H ~ OHCH
0-
+
C H y - C H ~ C O C H ~ H20 CHa 0-
CH3\(!!-CH&OCH~
-+Slow CHaCOCHa f
(CH~COCHS) -
CH 3/ (CHzCOCHJ-
Fast + HzO + CHaCOCHs
OH-
Since the concentration of the ionized aldol would be low, the rate of the reaction should vary with the hydroxyl ion concentration, as observed. That the reaction in no way involves the enolization of the aldol has long been known, since isobutraldol easily reverts to isobutraldehyde.I4 There remains, however, the question of why Discussion some amines catalyze the reaction. From the The results of this investigation confirm those data presented here, a possible hypothesis is that of Miller and Rilpatrick insofar as they show that a necessary (but probably not a sufficient) condithe rate of the dealdolization of diacetone alcohol, tion for catalysis by molecular amines is the prescatalyzed by methyl- and dimethylamine, is the ence of at least one hydrogen atom attached to sum of an hydroxyl ion catalysis and a catalysis the amine nitrogen. Since the rate varies linearly due to molecular amine. Molecular trimethyl- with the amine concentration, it is apparent that and triethylamines, however, do not affect the an intermediate in the reaction must be a cornrate, nor, according to French, does phenolate ion. pound formed between the aldol and the amine, We are then forced to the conclusion that this is and according to this theory, involving the amine not a case of general base catalysis. It is illumi- hydrogen. A substituted ketone ammonia (hynating to discuss the mechanism of the aldol con- drated ketimine) would satisfy the kinetic redensation in the light of these results. quirements. The accepted theory of the aldol condensatio~i~~ The idea of a ketimine as intermediate in conwould demand that the dealdolization of diace- densations similar to the aldol is not new.15 tone alcohol take place in the following steps: Qualitative differences in behavior between prifirst, a proton would be lost from the alcohol mary and tertiary amines have been noted begroup of the aldol. Next, this ion would decom- fore.’e Compounds have even been isolated pose to acetone and the enolate ion of acetone. which contained the aldehyde, amine, and, for exThe latter would then acquire a proton from the ample, nitroparaffin.” But it is difficult to estabsolvent. Since, as has been shown above, the re- lish, without kinetic evidence, that these com-
l W 4 , p 188
94
Vol. (io
FREDC . WRBERAND FRANK J. SOWA
pounds are intermediates, and not the products of Summary a side reaction, in equilibrium with the starting or 1. The rate of the dealdolization of diacetone final products, but not directly concerned in the alcohol has been measured in buffered solutions conversion of the ones to the others. containing methyl-, dimethyl-, trimethyl- and The data here presented indicate the existence triethylamine, at several buffer concentrations of the ketone amine intermediate, but do not set- and buffer ratios. tle the question of its structure, or why it should 2. While molecular methyl- and dimethyldecompose rapidly with respect to diacetone alco- amine catalyze the reaction, molecular trimethylhol. In fact, a meclianism which would depend amine and molecular triethylamine are without upon the concentrations of the substituted am- effect. monium ion and hydroxide ion would be indis3. It is concluded therefore that the aldol continguishable kinetically from one depending on densation is not an example of general base cathe concentration of the amine. talysis. The authors wish to thank Professor M. S. 4. The mechanism of the reaction, in the presKharasch for his helpful suggestions during the ence and the absence of amines, is discussed. course of this work. CHICAGO, ILL. RECEIVED AUGUST26, 1937
[CONTRIBUTION FROM THE
DEPARTMENT O F CHEMISTKY, UNIVERSITY
O F NOTRE DAME]
The Cleavage of Diphenyl Ethers by sodium in Liquid Ammonia. 111. 4,4’-Disubstituted Diphenyl Ethers‘ BY FREDC. WEBERAND FRANK J. SOWA It has been shown2that a solution of sodium in liquid ammonia reacts vigorously and quantitatively with substituted diphenyl ethers. The products formed are substituted benzenes and phenols. The substituted phenols .alone were determined experimentally and it was observed that the relative proportions of the two possible cleavage products formed in each reaction vary with the nature of the su tuents in the phenyl nucleus. It was pointed out that the electron is probably the effective cleavage reagent, since sodium amide in liquid ammonia does not cleave diphenyl ether, thus eliminating the possibility of the sodium ion being the effective reagent. The purpose of this investigation was to determine the effect of various substituents on the manner in which 4,$’-disubstituted diphenyl ethers are cleaved by sodium in liquid ammonia, the method having esperisl merit because one can make a comparison of the effect of two dissimilar groups within the same moleede. Experimental A standard methad of cleavage of the diphenyl ethers, outlined in the first article of this series,2 (1) Original manuscript received July 6, 1937. For previous article see Kranzfelder, Verbanc and Sawa. THISJ O U R N A L , 69, 1488 (1937). ( 2 ) S u t o r e t t o and Son&, r b d , 69, 607 (1!237).
TABLE I PHYSICAL PROPERTIES OF THE DIPHENYLETHERS PREPARED
Compds. K--C~H~-O--~HI--R’ No. R R
B. p. at indic%tFd press.
C.
Mm. M. p.. “ C
1 p-Methyl
p’-Methoxy“ 193-198 22 49-50 2 p-Methyl P’-Nitro 223-227 25 65 3 p-Methyl p’-Amino 195-199 16 119 218-222 16 106-7.5 4 p-Methoxy p‘-Nitro 5 p-Methoxy p’-Amino 191-192 4 754.5 6 P-Methoxy p‘-Hydroxy“ 189-193 16 64-5 7 p-Methyl p’-t-ButyP 174-178 4 b ‘ Compounds previously unreported in the literature. * 1226D1.5488, Sp. gr.26 0.9898. Compounds nos. (2) and (4)were prepared according to the method of 3rewster and Groening, “Organic Syntheccs,” John Wiley and Sons, Inc., New York, Vol. XIV, p. 6 7 ; nos. (3) and (5) by reduction of the corresponding nitro compounds employing the method of Suter, THIS JOURNAL, 51,2538 11929); nos. (l), (6) and (7)were prepared using the method described by Sartoretto and Sowa.Z All of the diphenyl ethers described in this paper were purified by distilla tion under reduced pressure
was used throughout this work. The cleavage products were extracted with ether or benzene and then fractionated using an efficient Wiclrner column. The cleavages of all of the diphenyl ethers studied, with the exception of the 4-methoxy-4‘hydroxydiphenyl ether, require two atom equiva-
