COMMUNICATIONS TO THE EDITOR
Oct. 20, 1964 KIK2KaP=0
+
CI -
H+
//
RIK2R&-OH
KIR~RIPOHC~ I1
K~R~RsP(OH)Z I11
occur a t this stage by internal motion, i.e., flipping of the groups' positions in either a trigonal-bipyramidal structure or similar motion in a square pyramid. Such motion has been observed in other pentacovalent phosphorous compounds and there seems little reason for not expectingit here.j I t is possible, of course, t h a t R 1 K Z R a PC12 is formed in a subsequent step and such a reaction would most certainly lead to racemization. Oxygen exchange can occur via the dihydroxy compound (III).' Similar exchanges have been recently discovered in reactions of' alkoxyphosphonium salts.' It is interesting to note t h a t triphenylphosphine oxide does not exchange oxygen on boiling with waterEor in concentrated sulfuric acid.g
3387
of 180-labeledethanol. In both cases the appropriate phosphine oxide was formed in high yield (86-90~o). The triphenylphosphine oxide contained 78yo of the lKOexpected if complete equivalence of all the ethoxy groups had been obtained in the system. Fifty-four per cent of the theoretical amount of lEO was found in the phenyldipropylphosphine oxide. In a complementary experiment tributylphosphine was allowed to react with diethyl peroxide in 1-propanol. Besides tributylphosphine oxide there was formed ethanol, ethyl propyl ether, and dipropyl ether.5 T h e results of these experiments are explainable in terms of the formation of alkoxyphosphonium alkoxides R3P
R'-OH + (CH3CH20)2-R3h-OC,Hs
+ (R'O- + C2H50-)
which undergo varying amounts of exchange in competition with the irreversible decomposition to phosphine oxides, ethers, alcohols, and possibly olefins. Whether exchange proceeds through a pentacovalent intermediate, R3POROR', cannot be deduced from this evidence. I t is important to note that the amount of ( 5 ) E. L. Muetterties, W. XIahler, and R . Schmutzler, I n o r g . Chem., 2 , exchange is a function of relatively minor structural 613 (1563). changes6 (6) In all of these reactions t h e results can in general be explained b y a pentacovalent intermediate or a transition s t a t e with five groups coordinated The reaction of tributylphosphine with diethyl pert o phosphorus. oxide in the absence of solvent was followed by con(7) D . B. Denney, H . M. Relies, a n d A. K. Tsolis, J . A m . Chem. S o c . , 8 6 , tinuous investigation in the n.m.r. It was easily seen 4487 (1561). (8) M. Halmann a n d S . Pinchas, J . Chem. Soc., 3264 (1968). that the initial products of the reaction were tributyl( 5 ) S. Oae, T. Kitao, and Y . Kitaoka, Chem. Ind. (London), 291 (1961). phosphine oxide, ethanol, and presumably e t h ~ l e n e . ~ SCHOOL OF CHEMISTRY DOSALD B. DESNEY After about 40 min. (ca. 1 half-life) diethyl ether was RUTGERS, THESTATEUSIVERSITY ALEXANDROSK. TSOLIS being formed to the practical exclusion of ethanol. The SEW BRLXSWICK,SEX JERSEY final products, identified by n.m.r. and g.l.p.c.,were triDEPARTMEST OF CHEMISTRY KURTMISLOW PRISCETON UNIVERSITY butylphosphine oxide, ethanol, and diethyl ether. The PRINCETON, SEW JERSEY relative yields of the latter two components were 69 and RECEIVED AUGUST24, 1964 31%, respectively. The change in rate of product formation with time of reaction is particularly unusual and indicates a general change in reaction course. This is Formation and Reactions of most probably due to the formation of ethanol. Two Alkoxyphosphonium Alkoxides' explanations can be offered for its effect. In the absence of alcohol the reaction may yield the pentacovalent Sir : compound I1 in the initial step.x This could decompose Alkoxyphosphonium salts (I) are formed as inter+
R3P-OR' I
X-
mediates in a wide variety of reactions in organophosphorus chemistry.2 Little is known about their stability'and properties other than that they, in general, decompose rapidly b y typical displacement and/or eliminative p r o c e ~ s e s . ~Because of their importance to our understanding of mechanisms in organophosphorus chemistry a study of the formation and properties of these materials has been initiated. The results of these experiments illustrate t h a t simple decomposition of these intermediates is not the only reaction which takes place in these systems. Triphenylphosphine and phenyldipropylphosphine were allowed to react with diethyl peroxide4in an excess (1) Research supported by t h e Yational Science Foundation under N S F GP-202. (2) J. I. G. Cadogan, Quart. Rev. (London), 16,208 ( 1 5 6 2 ) . ( 3 ) D . B. Denney and R . R . DiLeone, J . A m . Chem. Soc., 84, 4737 (1962), have reported on t h e preparation of
( C s H a & P - 0 ~ c1which is stable toward decomposition because of t h e bridgehead nature of t h e oxygen-bearing carbon. (4) L . H o m e r and W Jurgeleit, A n n . , 691, 138 (1955).
(C4H9)3P
+ (C2HiO)z + ( C ~ H P ) ~ P ( O C ~ H S ) ~ I11
+ CHa=CH2
directly to phosphine oxide, ethanol, and ethylene or ionize to tributylethoxyphosphonium ethoxide. Subsequent decomposition would then have to yield the products but as alcohol was formed an increase in S N ~ product, diethyl ether, would be required. On the other hand, as alcohol is formed displacement by the phosphine may yield ions 111 directly. T h e function of the alcohol would be to hydrogen bond with the forming ethoxide ion. Indeed, i t was noted t h a t the reaction in 1-propanol was a t least as fast as the neat reaction despite the large change in concentration. ( 5 ) Small amounts of diethyl ether might have been formed and escaped detection. (6) T h e extensive exchange observed in these systems supports t h e mechanism suggested by C. B. Parisek, C . S . VanderWerf, and W E . M c E w e n , J . A m . Chem. Soc., 82, 5.503 (1960). foi- t h e conversion of optically active methylethylphenylhenzylphosphonium butoxide t o highly racemized methylethylphenylphosphine oxide. See also M .Grayson and P. T. Keough, ibid., 82, 3919 (1960), in this regard. ( 7 ) K O evidence for t h e formation of t r i b u t y l d i e t h o x y p h o s p h o r u s was obtained by this technique. (8) D. E . Denney, W . F. Goodyear, and B. Goldstein, J . A m . ('hum .Sot., 88, 1726 (1561), D . B. Denney and H Relles ibid , 86, 3897 (196-1)
COMMUNICATIONS TO T H E EDITOR
4488
Although it is not possible a t this time to provide a complete mechanistic picture for the reactions described, it is clear that they are subject in a major way to minor changes in structure and reaction conditions. These results show t h a t attempts to generalize from isolated experiments such as a stereochemical investigation will in most cases be meaningless, for the stereochemistry of the reaction may change as the reaction progresses. For example, the reaction of diethyl peroxide with optically active phosphine may yield early in the reaction entirely different optical results from those near the end.g (9) K . F. Hudson a n d >I GI-een, . Angpw. C h r m . I r r l e v n . E d . E n g l . , 2 , 11 (1963j, have indicated some vther p!.oblems which a r e encountered in stereochemical studies with compounds oi this t y p e (10) National Science Cooperative G r a d u a t e Fellow, 11362-1963, Public H e a l t h Service Fellow. 1963-1964.
SCHOOL OF CHEMISTRY RETGERS, THESTATE USIVERSITY SEW BRUXSLVICK, SELV JERSEY
DONALD B . DENNEY HOWARD M. R E L L E S ] ~
.ILEXASDROS I(.TSOLIS
RECEIVEDAUGCST24, 1961
Proton Transfers in Dipolar Aprotic Solvents.
I. Transfers to Fluorenyl and Trityl Anions in Dimethyl Sulfoxide Solution Sir : Recently, both Cram' and Schriesheim2 have presented evidence t h a t proton transfers from t-butyl alcohol and other hydroxy compounds to strong hydrocarbon bases in dimethyl sulfoxide (DMSO) solution are diffusion controlled. In this preliminary paper, we report results of competitive proton abstractions from several types of acids by the triphenylmethyl (trityl) and fluorenyl anions in DMSO solution which indicate that diffusion control may be a general characteristic of proton-transfer reactions in such solvents, indeTendent of the nature of the acid and base involved. U'e suggest a hypothesis which is consistent with the present data and has broad implications as to the general nature of proton-transfer reactions. Fluorenyl anion in DSISO was produced by adding 1 mmole of fluorene to 2 mmoles of potassium t-butoxide in 20 ml. of dry (
