Mechanism of Alcohol to Ester Rearrangement in Phosphorus Compounds1 A. F. JANZEN AND T. G. SMYRL Deparltnenl of Cllemislry, Unicersily of Mat~ilobo,Witltlipeg, Manitoba
Can. J. Chem. Downloaded from www.nrcresearchpress.com by 148.251.232.83 on 05/03/18 For personal use only.
Received October 4, 1971 The kinetics of the base-catalyzed rearrangement of dialkyl 2-hydroxy-1,1,1,3,3,3-hexafluoroisopropylphosphonate, (RO),P(0)C(OH)(CF3), to 1,1,1,3,3,3-hexafluoroisopropyldialkylphosphate, (RO),P(O)OCH(CF,),, where R = methyl, ethyl, n-butyl, have been studied. Energies of activation and entropies of activation have been obtained and the base strength of the catalyst has been varied. A mechanism is proposed to account for the results. Les cinttiques des transpositions baso-catalysees des dialkyl-2 hydroxy-1,1,1,3,3,3-hexaBuoroisopropy1phosphonates (RO),P(0)C(OH)(CF3), en 1,1,1,3,3,3-hexafluoroisopropyldialkyl phosphates, (RO),P(O)OCH(CF,), avec R = methyle, ethyle, n-butyle, ont ete etudiees. Les energies d'activation et les entropies d'activation ont ete obtenues et la force basique du catalyseur a ete modifiee. Un mecanisme est propose pour rendre compte des resultats. Canadian Journal of Chemistry, 50, 1205 (1972)
Introduction Rearrangements involving isomeric alcohols and alkoxides are potentially useful in synthetic chemistry. Depending on the element M, a straightforward procedure may result in the preparation of one, but not the other, isomer.
"Alcohol"
"Alkoxide" (ester, ether)
Considering only one example, namely the addition of metalloid-hydrogen bonds across carbonyl bonds, it is well known that B-H ( l ) , Si-H (2), Ge-H, and Sn-H (3) bonds react with ketones to give alkoxides, whereas C-H, N-H, P-H, and As-H (4) bonds react with ketones to give alcohols. Significant work has been carried out in the area of "alcohol-alkoxide" rearrangements. The Wittig rearrangement (M = C ) involves the conversion of ether to alcohol in the presence of a strong base such as alkyllithium ( 5 ) . The Brook rearrangement (M = Si) involves the conversion of alcohol to alkoxide in the presence of diethylamine (6, 7 ) whereas the reverse rearrangement, i.e. alkoxide to alcohol (M = Si) in the presence of t-butyllithium, has recently been reported by West and co-workers (8). Peddle and Ward have reported the alcohol to alkoxide rearrangement for M = Ge (9). 'Presented at the 54th Canadian Chemical Conference and Exhibition. Halifax, Nova Scotia, May 3 1 -June 2, 1971.
A number of workers have reported basecatalyzed "alcohol-alkoxide" rearrangements involving phosphorus compounds (10) and a kinetic study, by the refractometric method, of phosphonate to phosphate rearrangement has been reported by Pudovik and co-workers (1 1). In 1970 we reported the reaction of dialkyl phosphonates with hexafluoroacetone and found that in each case isomeric products were formed, the relative yield depending on the nature of
"Alcohol"
"Ester"
R (12). We further observed that rearrangement of alcohol to ester did not take place at room temperature in the absence of basic catalyst, i.e. under preparative conditions, and hence a rearrangement was not invoked to explain the relative yield of products. To investigate this reaction further and determine under what conditions rearrangement does occur and what conditions give maximum yield of either alcohol or ester, a kinetic and mechanistic study was undertaken of the basecatalyzed rearrangement, in which R, base, and temperature were systematically varied. 0 OH
0
(RO)~P-C(CF,)~
+( R O ) ~ P - O - C ( C F ~ ) ~
11
I
"Alcohol"
base
11
H
"Ester"
1
C A N A D I A N J O U R N A L OF CHEMISTRY. VOL. 50, 1972
TABLE 1. Rates of rearrangement of (CH,O),P(O)C(OH)(CF,), by pyridine in dichloromethane at 32.6 'C
Can. J. Chem. Downloaded from www.nrcresearchpress.com by 148.251.232.83 on 05/03/18 For personal use only.
Initial alcohol concentration (MI
2.'
r P,
R = methyl
Initial base concentration (MI
B =pyrid~ne
k; (s-I)
R = ethyl
2.5,
l
600 FIG. 1. Plot of log of peak area us. time for rearrange(0.99 M ) with pyridine ment of (CH,O),P(O)C(OH)(CF,), (2.57 M) in dichloromethane at various temperatures.
Kinetic Results The chemical shifts of trifluoromethyl groups of alcohol and ester differ sufficiently (12) that 19F n.m.r. techniques were suitable for measuring concentrations. The base catalyst pyridine was added to the alcohol in dichloromethane as solvent and the disappearance of the alcohol doublet was followed as a function oftime. Pseudo first order rate constants k; were obtained by plotting the logarithm of the peak area against time and the best value of the slope was calculated by the method of least squares. Dividing through by the base catalyst concentration, k;/(base) = k,, gave a second order rate constant k, which is independent of initial concentration of alcohol and base, as shown in Table 1. Alkyl substituents were varied and with the aid of a variable temperature probe on the n.m.r. instrument pseudo first order rate constants were determined at several temperatures for R = methyl (Fig. I), R = ethyl (Fig. 2), and R = n-butyl (Fig. 3). The second order rate constant was obtained as described above and values of k, are given in Table 2.
kz (1 mol-' S - ' )
i
l
B = pyridine
I
I
I
,
I800
l
i
4800
l
6000
TIME, s Frc. 2. Plot of log of peak area us. time for rearrangement of (CzHsO),P(0)C(OH)(CF3)z (1.22 M)with pyridine (2.57 M ) in dichloromethane at various temperatures.
It may be seen from Table 2 that rates of rearrangement decrease in the order R = methyl > ethyl > n-butyl, the second order (45.6"), rate constants being 8.03 x 1.33 x (44.6"), and 1.27 x (51.4") 1 mol- ' s- ' , respectively. Plotting log k, us. 1/T gave linear plots from which the energy of activation was obtained. The entropy of activation and pre-exponential
1
600
1
1
1800
1
1
4800
1
6000
TIME. s FIG.3. Plot of log of peak area us. time for the rearrangement of (tr-C,H,),P(O)C(OH)(CF,), (1.18 M ) with pyridine (2.57 M ) in dichloromethane at various temperatures.
1207
J A N Z E N A N D S M Y R L : ALCOHC)L-ESTER R E A R R A N G E M E N T
TABLE^.
TABLE 2. Rates of rearrangement of (RO),P(O)C(OH)(CF,), by pyridine in dichloromethane
Can. J. Chem. Downloaded from www.nrcresearchpress.com by 148.251.232.83 on 05/03/18 For personal use only.
R
T ("C)
k, (1 mol-' s - ' )
Effect of basecatalyst (1.38 M) on rearrangement of (CH,O),P(O)C(OH)(CF,), ( I . 10 M ) in dichloromethane at 33 "C
Base
PK,*
Dimethylsulfoxide Aniline N-Methylaniline N,N-Dimethylaniline Pyridine Diethylamine Piperidine
13 9.37 9.15 8.85 8.75 3.51 2.88
k, ( I m o l - ' s - ' ) -7 x 1.11 1.47 x 1.02 x 3.00 x
t
lo-4 lo-, lo-4 lo-,
t
*Value for dimethylsulfoxide taken from Anal. Chem.; all others from "Handbook of chemistry and physics" (see ref. 17). tToo fast for measurement by n.m.r. techniques.
TABLE 3. Energy of activation (E,), entropy of activation (AS+),and logarithm of pre-exponent (log A) for (RO),P(O)C(OH)(CF,), rearrangement R
E, (k cal/mol)
AS (cal/mol degree)
log A
factor were calculated in the usual way (see experimental) and activation parameters are summarized in Table 3. From the energies of activation it may be seen that the rate of rearrangement is favored in the order methyl > ethyl > n-butyl. On the other hand, the entropies of activation partly compensate for this trend since the negative entropies of activation are in the same order. The effect of the basic catalyst on the rearrangement was also studied. Typical basic catalysts such as pyridine, dimethylsulfoxide, aniline, N-methylaniline, N,N-dimethylaniline, piperidine, and diethylamine, which differ widely in base strength2 as measured by their pK,, were added to the alcohol (R = methyl) in dichloromethane and the rearrangement studied at 33". Rate constants k; and k2 were obtained and, as shown in Table 4, the data indicate that, with the exception of N,N-dimethylaniline, the stronger the base the greater the rate of rearrangement. 'The base strength with respect to the alcohol in dichloromethane solution is not known and pK, values -are an approximation only.
Mechanism of Rearrangement A reasonable mechanism for the alcohol to ester rearrangement must be consistent with second order kinetics and the observed activation parameters. If the mechanism was similar to that proposed by Brook, it would suggest that a common mechanism is possible for a variety of elements undergoing the "alcohol-alkoxide" rearrangement. Brook and co-workers (6) found that the base-catalyzed rearrangement of cr-silylcarbinol (alcohol) to silyl ether (alkoxide) is first order in alcohol and first order in base and involves processes having low energies of activation but very high negative entropies of activation, and a cyclic three-membered transition state was postulated. Pudovik and co-workers (1 1 ), who studied the thermal isomerization in the absence of solvent, of compounds 3, 4, and 5, postulated a mechanism in which the oxygen atom of the hydroxyl group attacks the phosphorus atom, which then
C A N A D I A N J O U R N A L OF <:HEMISTRY. VOL. 50, 1972
Can. J. Chem. Downloaded from www.nrcresearchpress.com by 148.251.232.83 on 05/03/18 For personal use only.
1208
- q?
- 42
- 3? AS1
- 27
- 12
-2
C A L I I/OL DEGREE
FIG. 4. Plot of E, us. AS* for ..alcohol to alkoxide" and rearrangement: this work, 0 ; Brook et a/. ( 6 ) , 0 ; Pudovik et al. (1 1), a.
leads to cleavage of 0-H and C-P bonds and and C-H bonds. formation of new C-0-P To test for an isokinetic relationship, the energy of activation was plotted us. the entropy of activation, as described by Leffler (13). As seen from Fig. 4, the results of this work and those of Brook for compounds 1 and 2 give an approximately linear plot, suggesting a common mechanism ofrearrangement. On the other hand, the results of Pudovik for compounds 3, 4, and 5 are displaced towards higher energies of activation and more negative entropies of activation and since his results were obtained in the absence of solvent and catalyst it is not surprising that a different mechanism is implied by Fig. 4. A reasonable mechanism that is consistent with the kinetics and activation parameters and, at the same time, is in agreement with the results obtained by Brook, is shown in eqs. 1-3. Step 1 involves polarization of the hydroxyl oxygen, either through hydrogen bonding, which is expected to be particularly strong for pyridine and perfluoroalcohols (14), or through ionization to give the pyridinium ion. Hydrogen bonding in the starting compound involving phosphoryl oxygen and hydroxyl would probably be of smaller magnitude (15). Step 1 probably involves rapid proton exchange and the effect of increasing base strength of catalyst must be to shift the equilibrium to the right, in the direction of increased ionization. Step 2 must then represent the rate determining step; in particular, it is postulated that the
slow step is the intramolecular nucleophilic attack ofthe hydroxyl oxygen on the phosphorus atom, thereby implying that any substituent on the phosphorus atom which releases electron density to the phosphorus atom will increase the energy barrier to the transition state. Since the inductive (electron release) effect of alkyl groups is generally assumed to be in the order butyl > ethyl >methyl the argument is consistent with the trend found for the activation energies. Step 3 involves collapse to final ester product, either through an intermediate carbanion followed by proton transfer, or directly via simultaneous cleavage of P-C bond and formation of C-H bond. The negative entropies of activation are consistent with a constrained intermediate, namely a cyclic three-membered complex as postulated by Brook. For R = n-butyl the entropy of activation is only - 3.0 e.u. and molecular models show that steric interactions are present between rotating n-butyl and trifluoromethyl groups. However, these steric interactions are not expected to differ greatly in the ground state and transition state and, in view of an isokinetic relationship, a cyclic three-membered transition state seems most reasonable. It is interesting to speculate whether eqs. 2 and 3 could, under different conditions, represent equilibrium processes and permit the reverse rearrangement i.e. ester to alcohol. Undoubtedly a very strong base would be u
1209
Can. J. Chem. Downloaded from www.nrcresearchpress.com by 148.251.232.83 on 05/03/18 For personal use only.
J A N Z E N A N D S M Y R L : ALCOHOL-ESTER R E A R R A N G E M E N T
required for carbanion formation, and we find that if n-butyllithium is added to ester (R = methyl) the characteristic 19F n.m.r. doublet of the ester is replaced by a singlet, as required for the carbanion. Further n.m.r. examination showed that a P-C bond was not formed and we conclude that the reverse reaction did not go past the carbanion stage. What other factors, if any, could drive the equilibrium past the carbanion to the alcohol stage are not known at present, although the relative stabilities of P-C us. P-0 and (RO),P(O)C(O-)(CF,), us. (RO),P(O)OC- (CF,), could be important (8).
approximately 30min, the ester reacted with pyridine to give a new product containing the grouping CH,OPOCH(CF,),, as determined by 'H and 19Fn.m.r. Approximately 50% of the ester had reacted after 24 h, and 66% after 48 h. Since this reaction was too slow to interfere with the kinetic results, it was not investigated further.
Kinetic Measurements A Varian A-56/60A n.m.r. spectrometer, equipped with a variable temperature probe, was used to determine rate constants. The 19F n.m.r. spectra showed well separated doublets due to trifluoromethyl groups of alcohol and isomeric ester (12). In a typical kinetic run, a solution of alcohol in Experimental dichloromethane, in a standard n.m.r. tube, was allowed to reach thermal equilibrium in the Conventional vacuum line techniques were used for handling volatile materials. Hexafluoroacetone, dialkyl sample probe of the n.m.r. spectrometer. The phosphonates, and bases were purified by vacuum distillan.m.r. tube was then withdrawn and a measured tion or trap-to-trap distillation and their purity checked by amount of base added and the mixture shaken i.r. and n.m.r. spectroscopy. Dichloromethane (B.D.H. to ensure thorough mixing. The n.m.r. tube Chemicals) was used without further purification. The n.m.r. spectra were obtained on a Varian A-56/60A specwas re-inserted and the 19F n.m.r. peaks due to trometer using 60 and 56.4 MHz for proton and fluorine, alcohol and ester were recorded at regular respectively. intervals of time until at least 90% rearrangePreparation of Dialkyl2-Hydroxy-1,1,1,3,3,3-hexafioroiso-ment had occurred. During the kinetic runs the propylpltosphonate, (RO),P(O)C(OH) (CF,),, in integral heights of alcohol and ester were High Yield summed to check that sums remained constant The procedure described previously (1 2) gave low yields i.e. to ensure that alcohol was converted only for R = methyl and ethyl. Changing reaction conditions as to ester. described below substantially increased the yield. Dimethyl phosphonate (2.40 g, 21.8 mmol) in dichloroPseudo first order rate constants k; were methane (10.0ml) was cooled to - 196' and hexafluoroobtained from a plot of the logarithm of alcohol acetone (4.29 g, 25.8 mmol) was added. The reaction tube peak area us. the time and the best slope calcuwas sealed under vacuum and placed in a Dry Ice-acetone lated by the method of least squares. Second bath (-78") which was allowed to evaporate and reach 25" over a period of 24 h. After this time the reaction tube was order rate constants were obtained by dividing opened and excess hexafluoroacetone removed. The products the pseudo first order rate constant by the were identified as dimethyl-2-hydroxy-l , 1,1,3,3,3-hexa- concentration of base i.e. k, = k;/(base). fluoroisopropylphosphonate ("alcohol"), (CH,O),P(O)The Arrhenius energy of activation E, was C(OH)(CF,), (5.06 g, 84%), and I, I, I ,3,3,3-hexafluoroobtained by repeating the kinetic runs at several isopropyldimethyl phosphate ("ester"), (CH,O),P(O)OCH(CF,), (0.96g, 16%) on the basis of 'H and 19F temperatures. An ethylene glycol temperature n.m.r. and i.r. examination (12). us. chemical shift chart was used to calibrate Diethyl phosphonate (3.01 g, 21.8 mmol) in dichlorothe temperature of the n.m.r. probe, and the methane (10.0 ml) and hexafluoroacetone (4.29 g, 25.8 temperatures are accurate to -t lo.The temperammol) were allowed to react as described above to give ture was checked before and after every run and (C,HsO)2P(0)C(OH)(CF3)2 (5.83 g, 88%) and (C2HsO),P(O)OCH(CF,), (0.80 g, 12%). if the temperature varied that particular run The compound (C4HgO)2P(0)C(OH)(CF3)2was prewas discarded. paredin yields ofat least 90%, the remainder being (C4HgO),The entropy of activation was calculated P(O)OCH(CF,),, using the procedure outlined above or as from the equation described previously (1 2). No rearrangement occurred when alcohols were kept at room temperature in the absence of basic catalyst. Furthermore, no rearrangement occurred when dialkyl phosphonates were added to the alcohols at room temperature. One of the esters, namely (CH,O),P(O)OCH(CF,),, was found to be unstable in the presence of bases such as pyridine and diethylamine. After an induction period of
and A from k = A e-E"IRT
(ref. 16, p. 53)
1210
C A N A D I A N J O U R N A L OF CHEMISTRY. VOL. 50. 1972
7. G . J. D. PEDDLEand J. E. H. WARD. J. Organomet. Chem. 13, 269 (1968). and A. WRIGHT. 8. R. WEST,R. LOWE,H. F. STEWART, J. Am. Chem. Soc. 93,282 (1971). 9. G. J. D. PEDDLEand J . E. H. WARD.J. Organomet. Chem. 14, 131 (1968). 10. W. F. BARTHEL, B. H. ALEXANDER, P. A. GIANG,and S. A. HALL. J. Am. Chem. Soc. 77, 2424 (1955); L. A. R. HALL,C. W. STEPHENS, and J. J. DRYSDALE. J. Am. Chem. Soc. 79, 1768 (1957); I. S. BENGELSDORF. J. Org. Chem. 21, 475 (1956); S. J. FITCHand K. MOEDRITZER.J. Am. Chem. Soc. 84, 1876 (1962); The financial assistance of the National Research Council V. S. ABRAMOV, N. I. D'YAKONOVA, and V. D. EFIMOVA. of Canada is gratefully acknowledged. Zh. Obschei Khim. 39, 1971 (1969). I. V. GUR'YANOVA, L. V. BANDEROVA, 11. A. N. PUDOVIK, 1. G . W. PARSHALL.Inorg. Chem. 4, 52 (1965). and G. V. ROMANOV.J. Gen. Chem. USSR, 38, 142 2. A. F. JANZENand C. J. WILLIS. Can. J. Chem. 43, (1968). 12. A. F . JANZENand R. POLLITT. Can. J. Chem. 48, 3063 (1965); Inorg. Chem. 6, 1900 (1967). 1987 (1970). 3. W. R. CULLENand G. E. STYAN. Inorg. Chem. 4, 13. J. E. LEFFLER.J. Org. Chem. 20, 1202 (1955). 1437 (1965). and R. V. LINDSEY, JR. J. Am. 14. W. J. MIDDLETON 4. C. G. KRESPANand W. J. MIDDLETON.Fluorine Chem. Soc. 86,4948 (1964). chemistry review. Vol. I. Editerl b! P. Tarrant. M. 15. L. J. BELLAMY.Advances in infrared group frequencies. Dekker. New York, 1967. Methuen, London, 1968. p. 204. 5. G . WITTIC and L. LOHMANN.Ann. 550, 260 (1942); Molecular rearrangements. Part I. 16. K. J. LAIDLER.Chemical kinetics. 2nd ed. McGrawH. E. ZIMMERMAN. Hill, New York, N.Y., 1965. p. 90. Edited by Paul de Mayo. Interscience, New York, N.Y., 17. C. A. STREULI.Anal. Chem. 30,997 (1958). Handbook 1963. p. 372-377. of chemistry and physics. 5Ist ed. Edited by R. C. West. 6. A. G. BROOK,G. E. LEGROW,and D. M. MACRAE. Chemical Rubber Co., Cleveland, 1970. p. D l 17. Can. J. Chem. 45, 239 (1967).
Can. J. Chem. Downloaded from www.nrcresearchpress.com by 148.251.232.83 on 05/03/18 For personal use only.
The rearrangement of (CH,O),P(O)C(OH)(CF,), was studied at 33" in dichloromethane as a function of base strength ofcatalyst. Dimethylsulfoxide, aniline, N-methylaniline, N,N-dimethylaniline, pyridine, piperidine, and diethylamine were used and, in each case, k; and k, were determined and the results are shown in Table 4.
