THE MECHANISM OF THE PERMANGANATE OXIDATION OF FLUORO ALCOHOLS IN AQUEOUS SOLUTION BY Ross STEWART AND R. VAN DER LINDEN Dept. of Chemistry, University of British Columbia, Vancouver, Canada Received 1 st February, 1960 The mechanism of the aqueous permanganate oxidation of a series of aromatic fluoro alcohols has been studied. The reaction rate is proportional to the concentration of alkoxide ion and this, together with the salt effect and the activation entropies, supports the idea that the reaction takes place between alkoxide and permanganate ions. A very large reduction in rate occurs for the oxidation of the 1-deutero compounds (16 : 1). The effect of aromatic ring substituents on the rate of oxidation is slight, assuming the reaction proceeds via the alkoxide ion. All the evidence except the effect of substituents on the rate is in agreement with a hydride transfer mechanism fram alkoxide ion to permanganate ion.Various other possible mechanisms are considered. Permanganate is a powerful oxidizing agent which, not surprisingly, shows a variety of reaction paths in its reactions with organic and inorganic compounds.l.2 Previous work in this laboratory 3 has shown that one of these, the oxidation of the secondary alcohol benzhydrol, PhzCHOH, to benzophenone by aqueous per manganate, proceeds cleanly with the following kinetics : V = k[Ph,CHOH][MnO,][OH-]. The following additional facts are known: a kinetic isotope effect of 6.6 exists for the oxidation of Ph2CDOH; the entropy of activation is large and negative; a positive kinetic salt effect exists, and there is no transfer of oxygen from per- manganate to the organic substrate during the reaction.These facts all suggest that the reaction path is the following with the rate-controlling step being a transfer of hydride ion from alkoxide ion to permanganate ion. Ph2CHOH + OH' +Ph,CHO- + H20 fast, HMnOi-+MnO; +OH--+2Mn0~-+H20 fast. The acidity of this alcohol is such, however, that it can be only slightly ionized under the reaction conditions. We have accordingly turned to fluorinated com- pounds in order to obtain alcohols of sufficiently high acidity that we can examine in more detail the first two steps above if, indeed, the validity of the overall mechanism appears to be c o n w e d by this further work. A series of aryl trifluoromethyl carbinols proved to be satisfactory since they are highly ionized in 0.1 N alkali and are oxidized cleanly to the corresponding ketones by permanganate : ArCHOHCF3+2Mn0, +20H-+ArCOCF3+2Mn0:- +2H20.Ph2CHO- + MnO,'+Ph,CO+HMnO~- slow, EXPERIMENTAL The preparation of the Auoro alcohols and ketones has been reported elsewhere.4.5 The reaction was followed by an iodometric procedure similar to that previously reported.% 6 Thc only important difference was that the precision of the method was improved by 21 1212 PERMANGANATE OXIDATION OF ALCOHOLS using separate solutions for each point on the rate plot rather than by removing aliquots from a single solution. For reactions in solutions more basic than pH 11.7 a 2 : 1 ratio of permanganate to alcohol was used.For more acidic solutions a 2 : 3 ratio was used in accordance with the altered stoichiometry. Phosphate buffers were used and potassium nitrate and potassium sulphate used to vary the ionic strength. The second-order rate constants k2 were determined using the following equation when a 2 : 1 ratio of reactants was used : where VO and V, are the volumes of thiosulphate required at time zero and at time t, and [alc]~ is the initial concentration of alcohol. When the 2 : 3 ratio is used the final term becomes ( VO - V{)/( Vt - {- VO). An examination of the ultra-violet spectra of the solution after oxidation of C~HSCHOHCF~ in 0.1 N NaOH under the same conditions as were used for the kinetic experiments showed that the ketone C6HsCOCF2 was present in the solution in greater than 95 % yield. The ketone was found to be resistant to further oxidation and to hydrolysis under the reaction conditions.A few experiments which were performed with the primary alcohol, HCFzCFzCHzOH, gave results very similar to those reported herein. The rate in 0.18 N NaOH was 2.2 1. mole-1 sec-1; AH* = 10.2 kcal/mole ; AS+ = -22 cal/mole deg. ; alcohol pK = 12.1. RESULTS AND DISCUSSION KINETICS At constant hydroxyl ion concentration the reaction is second order as can be seen in fig. 1, the second-order rate constant k2 being insensitive to variations in reactant concentration. Changing the hydroxyl ion concentration changes the rate in the manner shown in fig. 2. The pH values at the midpoints of these curves all correspond quite closely to the known pK values of the alcohols.4 These range from 12.2 for the p-CH30 compound to 11.2 for the m-NO2 compound. A positive salt effect is observed for the oxidation of m-N02Cfl&HOHCF3 in 0.01 N NaOH at 25".The rate increases from a value of 4.54 1. mole-1 sec-1 at ionic strength 0.0115 to a value of 5.90 1. mole-1 sec-1 at ionic strength 0-122. The above results are all consistent with the following reaction path : ROH+OH-~RO-+H,O, (2) (3) -d[MnO,]/dt = k[RO-][MnO;]. (4) -d[MnO,]/dt = k,[alc][MnO;], ( 5 ) kz[alc] = k[RO'], ( 6 ) k RO- + MnO; +products. The rate law corresponding to this mechanism is Since eqn. (1) was derived from the expression, where [alc] = [ROH]+ [RO-1, it follows that and under conditions of almost complete ionization of the alcohol the experimental rate constant k2 becomes identical with k, the rate constant of the rate-controlling step.It is interesting that the permanganate-formate reaction kinetics are exactly analogous.7-9I R . STEWART A N D R . VAN DER LINDEN i 21 3 4 8 12 t , min FIG. 1.-Second-order rate plots, [alc]~ = 5.1 x 10-4, [OH-] = 0.2, T = 25.2" ; 0 GH5CHOHCF3 ; c) C6HsCDOHCF3 (multiply units on both axes by four for latter). [OH-] FIG. 2.Variation of rate with basicity ; CsH5CHOHCF3 ; 0 m-NOz ; CD m-Br; A p-CH3; 63 C~H~CDOHCFJ.214 PERMANG ANATE OXIDATION OF ALCO I-IOLS A termolecular step involving ROH, MnOz and OH- is also consistent with the kinetics since the rate law - d[ Mn O,]/dt = I&[ ROH][ OH'] [MnO,]/[ H, O] (7) is equivalent to the one above. This possibility will be discussed in a later section. THERMODYNAMIC FUNCTIONS Fig.3 shows a plot of logl&/T) against l/Tfor the reaction in 0.1 N NaOH. At this basicity, k 2 ~ k and one can calculate the activation parameters for what we have assumed to be the rate-controlling step, i.e., the reaction between alkoxide ion, RO-, and permanganate ion. These values are listed in table 1 for the parent compound, its deutero analogue, and its m-nitro derivative. TABLE 1 .-HEATS AND ENTROPIES OF ACTIVATION 0.1 N NaOH, [alc]~ = 5.1 x 10-4 compound AH4, kcal/mole AS', cal/mole deg. CsH5CHOHCF3 9.1 & 0.3 -24.3f1 C~HSCDOHCF~ 11.3 - 22.4 rn-NO2C6H4CHOHCF3 9.6 -23.8 The heats of activation are similar to those observed in other permanganate oxidations 3 ~ 7 - 9 and the entropies of activation are reasonable for a reaction between two anions.10 1 I I I I 1 3.2 33 3.4 3.5 1000/T"K FIG.3.Variation of rate with temperature ; 0 C6HsCHOHCF3, y = 3 ; CsH5CDOHCF3, y = 4. ISOTOPE EFFECT A large kinetic isotope effect (16.1) is observed for the oxidation of the deuter- ated compounds, ArCHOHCF3. This immediately disposes of an electron transfer step from the alkoxide ion to permanganate ion. The following mechanism (hydride transfer) requires that there be an isotope effect but the values listed in table 2 are unusually large :R . STEWART AND R . VAN DER LINDEN 215 0 2 - ll +Ar-C-CF, I n I .4r-C-H CF3 transition state CF3 +HMnOi- (8) TABLE 2.-DEUTEKIUM ISOTOPE EFFECTS AT 25" kz, 1. mole sec-1 kH kD kHlkD P H C6H5CDOHCF3 13.3 13.0 12-0 11.0 7.0 * 1.0 * 13.0 13.0 m-BrC6H4CDOHCF3 13-3 P-CH~C~H~CDOHCF~ 13-3 7.5 7.6 3.8 0.18 -00394 40096 11.6 t 2.85 $ 7.6 8.4 0.47 *47 -23 -01 53 -000242 -00055 a88 j- 0155 $.-47 -52 16.0 16.2 16.5 12.0 16.0 1.8 13.2.1. 18.4 $ 16.2 16.1 * The reactant concentrations were considerably higher in these experiments. j- T = 38.0". T = 12-65", A high isotope effect is also observed in neutral solution even though the reaction is now very slow. In acid solution the reaction is still very slow but a low isotope effect is here observed. Normal deuterium isotope effects, which result from the loss of the C-H stretching mode in the transition state, are less than half this size at the same temperature.11 Four possible explanations for the anomalous effect present them- selves.(a) The C-H and C-D stretching frequencies might themselves be anomalous. This possibility can be rejected by an examination of the infra-red spectra of these compounds. For the three pairs of compounds studied CC-H is at 2895 cm-1 and i c - ~ is at 2150 cm-1, both normal values for carbon-hydrogen bonds. (b) Some unusual consecutive process involving possibly chain branching might produce a cumulative isotope effect. This, however, is difficult to reconcile with the " clean " kinetics and the similarity of the heats and entropies of activ- ation to those of other permanganate oxidations. (c) The reaction may involve loss of the bending modes in the transition state in addition to the loss of the stretching mode. This might be caused by the reactant molecules being widely separated in the transition state as a consequence, in the present case, of electro- static repulsion.The benzhydrol+ permanganate reaction 3 and the formate ion+ permanganate reaction 79 9 which resemble this reaction in many ways have, however, normal isotope effects, (d) Quantum-mechanical tunnelling which is more probable for protium than for deuterium may be occurring.13 Further work is required to see whether mechanistic or structural factors (presence of a CF3 group, for example) are responsible for the large effect reported here. Re- gardless of the cause, it is clear that the carbon-hydrogen bond is being broken in the rate-controlling step. It is of interest that the manganate oxidation of these compounds, although slaw, also gives large effects.216 PERMANGANATB OXJDATION OF ALCOHOLS The effect of deuterium substitution on the ionization of the alcohols is slight.The pK of C~HSCHOHCF~ and that of its deutero derivative are within 0.1 pK units of one another. EFFECT 01' KING SUBSTITUENTS The principal advantage of using fluoro alcohols to study the permanganate. alcohol reaction is that the first two steps of the suspected mechanism can be studied separately. The effect of ring substitution on the ionization step is similar to that for the benzoic acid ionization and shows a satisfactory Hammett linear free-energy relation 12 with a p value of 1.01.4 In 0.2 N NaOH the alcohols are almost completely ionized and k 2 z k A hydride transfer from the alkoxide ion to permanganate should result in a negativep value, i.e.a p-methoxy substituent should accelerate the process and a p-nitro group retard it. Fig. 2 shows that the reaction rate is, surprisingly, only slightly affected by nuclear substitution and that the small variations in rate which do occur (curve A) appear to be not linearly related to the Hammett substituent constant 0. (These rates are corrected for incomplete ionization of the alcohols in 0.2 N NaOH.) Several explanations for this situation can be considered. First, a simple hydride transfer may in fact be occurring with the electronic effect of a distant group in the molecule being unimportant. Secondly, two different processes with different electronic requirements may be occurring, one with a positive, the other with a negative, p.This would account for the curved plot but it is not certain that the small curvature is significant and this possibility will not be pursued here. These mechanisms can be written as follows : Thirdly, one of several termolecular reactions may be taking place. 0. (8) 1 Ar e n l n I H,O @-C-O- Mn0,-+H30++Ar-C-CF3+MnO:- - CF3 0. 1 n I A Ar-C-CF, + MnO,*'Sproducts. OH I Ar HO- @-C-OH MnO~-+H,O+Ar-C-CF, +MnO:- (5) I * * I . CF3 OH Ar-C-CF, + MnO,*%products, 0 Ar Ar f-31 n I HO- C--H MiiO,--+HO-C + HMnO?-. /\ HO CF3 /\ HO CF3 In reaction (8) the permanganate acts as an electron abstractor with a water moIecuIe acting simultaneously as a proton abstractor. This mechanism is inR . STEWART A N D R . VAN DER LINDEN 217 agreemcnt with the kinetic and isotopic evidence and since the alkoxide ion is losing both an electron and a proton in the transition state the effect of substituents should be slight.Reaction (9) is kinetically equivalent to (8) but hydroxyl ion removes the proton as the permanganate abstracts the electron from the hydroxyl group of the neutral substrate. (The permanganate cannot be removing a hydrogen atom or a hydride ion from the hydroxyl group since there is no appreciable change in rate when a 60 % D20+40 % H20 solvent is used.) It would be difficult to predict the p value for this reaction. It is of interest, however, that a reasonably linear relation is obtained when is now plotted against the logarithm of the rate constant (fig. 4, curve B). This treatment requires that the rate in 0.2 N NaOH be corrected for partial ionization rather than for incomplete ionization as before.I I 1 I I 1 -0.4 -0.2 0 0 . 2 0.4 0.6 0.8 U FIG. 4.-Hammett plot, T = 25*2", [OH-] = 0-2 ; 0 ki = k2[alcJ/[RO-] ; 0 ki = k2[alcl/lROH]. Eqn. (10) shows hydride transfer from the neutral alcohol being aided by dis- placement by hydroxyl ion to produce the stable hydrated ketone.4 Of the three termolecular mechanisms considered here, (8) does not require a termolecular solute collision and, in addition, it accommodates satisfactorily the rest of the evidence accumulated about this reaction. In summary, there are two unusual features of the permanganate oxidation of aryl trifluoromethyl carbinols, viz., the size of the deuterium isotope effect and the effect of aromatic substituents on the rate. If a simple hydride transfer from alkoxide ion to permanganate ion is rejected on the basis of the negligible effect of substituents one is forced to conclude that a termolecular process is occurring. The most attractive of these alternatives is that in which a permanganate ion and a water molecule remove an electron and a proton respectively from the alkoxide ion. The financial support of the National Research Council of Canada is gratefully acknowledged.21 8 PERMANCANATE OXIDATION OF ALCOHOLS 1 Ladbury and Cullis, Chem. Rev., 1958, 58, 403. 2 Waters, Quart. Rev., 1958, 277. 3 Stewart, J. Amer. Chem. SOC., 1957, 79, 3057. 4 Stewart and Van der Linden, Can. J. Chern., 1960, 38,399. 5 Stewart and Van der Linden, Tetrahedron Letters, 1960, OOOO. 6 Wiberg and Stewart, J. Amer. Chem. SOC., 1955, 77, 1786. 7 Wiberg and Stewart, J. Amer. Chem. SOC., 1956, 78, 1214. 8 Hill and Tompkins, Trans. Roy. SOC. S. Africa, 1943, 30, 59. 9 Taylor and Halpern, J. Amer. Chem. SOC., 1959, 81, 2933. 10 Frost and Pearson, Kinetics and Mechanism (Wiley, New York, 1953), p. 132. 11 Wiberg, Chem. Rev., 1955,55,713. 12 Hammett, Physical Organic Chemistry (McGraw-Hill, NewjYork, 1940). 13 Bell, Fendley and Hulett, Proc. Roy. SOC. .4.,!1956, 235, 453.