The mandelic acid keto-enol system in aqueous solution. Generation of the enol by hydration of phenylhydroxyketene and phenylcarboxycarbene

Article abstract of DOI:10.1021/ja971774r

Flash photolysis of phenyldiazoacetic acid and the methyl, n-butyl, and isobutyl esters of benzoylformic acid in aqueous solution generated a transient species that was identified, through product determination and the form of acid-base catalysis plus solvent isotope effects on its decay, as the enol of mandelic acid, 1. When benzoylformate esters were the flash photolysis substrates, the enol was formed by hydration of phenylhydroxyketene, 5, itself generated by Norrish type II photoelimination of the esters, and when phenyldiazoacetic acid was the substrate, the enol was formed by hydration of phenylcarboxycarbene, 6, produced by dediazotization of the diazo acid. Rates of enolization of mandelic acid were also determined, by monitoring the incorporation of deuterium into its a-position from a D₂O solvent, and combination of these with rates of ketonization gave the keto-enol equilibrium constant, pK(E) = 16.19. The acidity constant of the enol ionizing as an oxygen acid was determined as well, pQ(a)(E) = 6.39, and combination of that with K(E) led to the ionization constant of the keto form of mandelic acid ionizing as a carbon acid, pQ(a)(K) = 22.57. (These acidity constants are concentration quotients, applicable at ionic strength = 0.10 M.) These results are compared with other keto-enol systems, and their bearing on the enzymatic racemization of mandelic acid is discussed.

Full text of DOI:10.1021/ja971774r

J. Am. Chem. Soc. 1997, 119, 10203-10212
10203
The Mandelic Acid Keto-Enol System in Aqueous Solution.
Generation of the Enol by Hydration of Phenylhydroxyketene
and Phenylcarboxycarbene
Y. Chiang, A. J. Kresge,* V. V. Popik, and N. P. Schepp
Contribution from the Department of Chemistry, UniVersity of Toronto,
Toronto, Ontario M5S 3H6, Canada
ReceiVed May 30, 1997X
Abstract: Flash photolysis of phenyldiazoacetic acid and the methyl, n-butyl, and isobutyl esters of benzoylformic
acid in aqueous solution generated a transient species that was identified, through product determination and the
form of acid-base catalysis plus solvent isotope effects on its decay, as the enol of mandelic acid, 1. When
benzoylformate esters were the flash photolysis substrates, the enol was formed by hydration of phenylhydroxyketene,
5, itself generated by Norrish type II photoelimination of the esters, and when phenyldiazoacetic acid was the substrate,
the enol was formed by hydration of phenylcarboxycarbene, 6, produced by dediazotization of the diazo acid. Rates
of enolization of mandelic acid were also determined, by monitoring the incorporation of deuterium into its R-position
from a D2O solvent, and combination of these with rates of ketonization gave the keto-enol equilibrium constant,
pKE ) 16.19. The acidity constant of the enol ionizing as an oxygen acid was determined as well, pQE ) 6.39, and
a
combination of that with KE led to the ionization constant of the keto form of mandelic acid ionizing as a carbon
acid, pQKa ) 22.57. (These acidity constants are concentration quotients, applicable at ionic strength ) 0.10 M.)
These results are compared with other keto-enol systems, and their bearing on the enzymatic racemization of mandelic
acid is discussed.
Although the enol isomers of most simple aldehydes and
ketones are quite unstable, both thermodynamically and kineti-
cally, there has recently been a remarkable increase in our
knowledge of their chemistry, enabled largely by the develop-
ment of methods for generating these substances in solution in
greater than equilibrium amount under conditions where they
can be observed directly and their reactions can be studied in
detail.1 Application of these techniques to the enol isomers of
simple carboxylic acids, however, is much more difficult,
because carboxylic acid enols are very much more unstable.
For example, the equilibrium constant for keto-enol isomerism
of acetic acid in aqueous solution, eq 1, though not yet
determined experimentally, has been reliably estimated at pKE
) 20;2 this is 14 orders of magnitude less than the equilibrium
acid enol, CH2dC(OMe)2, and the fact that methyl enol ethers
undergo hydrogen-ion catalyzed hydrolysis one to two orders
of magnitude more slowly than ketonization of the correspond-
ing enols.5 The result, kH+ ) 108-109 M-1 s-1, is 7-8 orders
of magnitude greater than kH+ ) 33 M-1 s-1 for the ketonization
of acetaldehyde enol.3
Daunting realities such as this have directed the so far quite
limited studies of carboxylic acid enols to systems where the
enol is stabilized, either sterically6 or electronically.7 Carbon-
carbon double-bond-stabilizing substituents such as phenyl and
hydroxyl8 can be expected to be effective in this respect, and
we have found that the enol, 1, of mandelic acid, 2, which
contains both of these groups, though still quite reactive, is
amenable to study.
constant for the keto-enol transformation of acetaldehyde, eq
2, for which pKE ) 6.23.3 A rate constant for the hydrogen-
ion catalyzed ketonization of acetic acid enol may also be
(5) Keeffe, J. R.; Kresge, A. J. In The Chemistry of Enols; Rappoport,
Z., Ed.; Wiley: New York, 1990; Chapter 7.
(6) Hegarty, A. F.; O’Neill, P. J. Chem. Soc., Chem. Commun. 1987,
744-745. Allen, B. M.; Hegarty, A. F.; O’Neill, P.; Nguyen, M. T. J. Chem.
Soc., Perkin Trans. 2 1992, 927-934. Frey, J.; Rappoport, Z. J. Am. Chem.
Soc. 1995, 117, 1161-1162; 1996, 118, 5169-5182; 5182-5191.
(7) (a) Urwyler, B.; Wirz, J. Angew. Chem., Int. Ed. Engl. 1990, 29,
790-792. (b) Chiang, Y.; Kresge, A. J.; Pruszynski, P.; Schepp, N. P.;
Wirz, J. Angew. Chem., Int. Ed. Engl. 1990, 29, 792-794. (c) Barra, M.;
Fisher, T. A.; Cernigliaro, G. J.; Sinta, R.; Scaiano, J. C. J. Am. Chem.
Soc. 1992, 114, 2630-2634. (d) Andraos, J.; Chiang, Y.; Huang, C-G.;
Kresge, A. J.; Scaiano, J. C. J. Am. Chem. Soc. 1993, 115, 10605-10610.
(e) Andraos, J.; Chiang, Y.; Kresge, A. J.; Pojarlieff, I. G.; Schepp, N. P.;
Wirz, J. J. Am. Chem. Soc. 1994,116, 73-81. (f) Almstead, J.-I. K.; Urwyler,
B.; Wirz, J. J. Am. Chem. Soc. 1994, 116, 954-956. (g) Andraos, J.; Kresge,
A. J.; Popik, V. V. J. Am. Chem. Soc. 1994, 116, 961-967. (h) Andraos,
J.; Chiang, Y.; Kresge, A. J.; Popik, V. V. J. Am. Chem. Soc. In press.
(8) (a) Hine, J. Structural Effects on Equilibria in Organic Chemistry;
Wiley-Interscience: New York, 1975; p 273. (b) Keeffe, J. R.; Kresge, A.
J. J. Phys. Org. Chem. 1992, 5, 575-580.
+
estimated, using kH ) 7 × 105 M-1 s-1 for the hydrolysis of
ketene dimethyl acetal,4 which is the dimethyl ether of acetic
X Abstract published in AdVance ACS Abstracts, October 1, 1997.
(1) For a review, see: The Chemistry of Enols; Rappoport, Z., Ed.;
Wiley: New York, 1990.
(2) Guthrie, J. P. Can. J. Chem. 1993, 71, 2123-2128. Guthrie, J. P.;
Liu, Z. Can. J. Chem. 1995, 73, 1395-1398.
(3) Chiang, Y.; Hojatti, M.; Keeffe, J. R.; Kresge, A. J.; Schepp, N. P.;
Wirz, J. J. Am. Chem. Soc. 1987, 109, 4000-4009.
(4) Kresge, A. J.; Leibovitch, M. J. Am. Chem. Soc. 1992, 114, 3099-
3102.
S0002-7863(97)01774-5 CCC: $14.00 © 1997 American Chemical Society

10204 J. Am. Chem. Soc., Vol. 119, No. 42, 1997
Chiang et al.
We generated the enol of mandelic acid flash photolytically
using two different photochemical reactions: (1) Norrish type
II cleavage of benzoylformic acid esters, 3, eq 4, and elimination
of nitrogen from phenyldiazoacetic acid, 4, eq 5. The first of
these reactions had been studied before in nonaqueous solvents
Experimental Section
Materials. n-Butyl and isobutyl benzoylformate were prepared by
treating benzoylformyl chloride with the corresponding alcohol,9 and
phenyldiazoacetic acid was prepared by saponification of methyl
phenyldiazoacetate, itself obtained by lead tetraacetate oxidation of the
hydrazone of methyl benzoylformate.12 The diazoacid undergoes rapid
acid-catalyzed hydrolysis,13 and it was therefore not isolated as such.
Aqueous stock solutions of the acid salt were used instead; these were
obtained by hydrolysis of methyl phenyldiazoacetate with an equimolar
amount of 0.1 M aqueous sodium hydroxide and then washing the
resulting solution with methylene chloride to remove any unreacted
ester.
Acetylmandelic acid was prepared by treating mandelic acid with
acetyl chloride.14
All other materials were best available commercial grades.
Kinetics. Flash photolysis was performed using a conventional flash
lamp system and an excimer-laser system that have already been
described.3,7d The conventional system produced a 50-µs excitation
pulse and the laser system, operating at λ ) 248 nm, produced a 20-ns
excitation pulse. In both systems the temperature of the reacting
solutions was controlled at 25.0 ( 0.05 °C.
Rates of hydrogen exchange were measured using 1H NMR to
monitor incorporation of deuterium from a D2O solvent into the
R-position of mandelic acid. The area of the R-proton signal was
determined by comparison with that of the aromatic protons; further
comparison of the latter with the signal from tetramethylammonium
chloride, used as an internal frequency standard, showed that the
aromatic hydrogens did not exchange under the rather drastic experi-
mental conditions needed to effect R-hydrogen exchange. Rate
measurements were made in concentrated DCl solutions (0.5-4 M)
over the temperature range 130-150 °C. Corrections for changes in
acid concentration between the temperature at which the solutions were
prepared (25 °C) and those at which the reactions were run were made
using polynomial relationships between density and temperature.15
These relationships covered H2O up to 150 °C, but they covered D2O
only up to 100 °C; they showed, however, that the ratio of the density
of H2O to that of D2O remains virtually constant above 50 °C, and the
temperature coefficient of H2O density was therefore used to calculate
D2O densities at 130-150 °C, on the assumption that this constancy
extended to this range. All exchange reactions were monitored for
several half-lives. The data obtained fit the first-order rate law well,
and observed first-order rate constants were calculated by least-squares
fitting of an exponential function.
where phenylhydroxyketene, 5, and substances derived from it
were found to be produced;9 in aqueous solution this ketene
would be expected to be hydrated to mandelic acid enol, as
shown in eq 4. In a preliminary publication of some of the
presently described work,7b we postulated that photolysis of
phenyldiazoacetic acid, 4, also produced phenylhydroxyketene
by a photo-Wolff reaction. We now know that this photo-Wolff
reaction is only a minor route to mandelic enol from this
substrate, and we believe instead that mandelic acid enol is
generated from this substance principally by hydration of
phenylcarboxycarbene, 6, as shown in eq 5. These reaction
paths are supported, as will be described below, by the form of
acid-base catalysis and solvent isotope effects shown by
transient absorbance changes produced by the flash photolysis
of benzoylformic acid esters, 3, and phenyldiazoacetic acid, 4,
as well as by isotopic tracer studies.
Rates of racemization were measured using (R)-(-)-mandelic acid
as the substrate. High temperatures and concentrated acids, comparable
to those used for hydrogen exchange measurements, were again required
to effect reaction. These data also fit the first-order rate law well, and
observed first-order rate constants were once again obtained by least
squares fitting of an exponential function.
We have also measured the very slow rates of enolization of
mandelic acid in order to evaluate the keto-enol equilibrium
constant for this system, KE, by combining the enolization rate
constant, kE, with the flash-photolytically determined ketoniza-
tion rate constant, kK, according to the relationship KE ) kE/kK.
We orginally7b believed that acid-catalyzed racemization of
mandelic acid occurred by an enolization pathway, but we now
know that this is not so, and we have consequently used
hydrogen exchange to monitor enolization.
Results
Product Studies. Product studies were carried out by
subjecting wholly aqueous solutions of substrate, at concentra-
tions comparable to those used for the flash photolytic rate
measurements (ca. 10-5 M), to a single flash from our
conventional flash photolysis system and then analyzing the
resulting solutions by HPLC. Identities of substances appearing
in the chromatograms were established by comparing retention
times and UV spectra and also by spiking with authentic
samples. Analyses were performed for methyl benzoylformate
as the substrate in 0.1 and 0.001 M perchloric acid solutions,
in an acetic acid buffer of pH ) 5 and in a biphosphate buffer
of pH ) 7, for n-butyl benzoylformate as the substrate in 1 M
The results that we have obtained constitute one of the first
sets of detailed information on the chemistry of a simple,
nonsterically hindered carboxylic acid enol. They also bear
upon a much studied enzymatic reaction catalyzed by mandelate
racemase10 by providing a quantitative measure of the thermo-
dynamic barrier that this enzyme must overcome; two different
hypotheses have recently been advanced concerning the mech-
anism that the enzyme uses to overcome this barrier.11
(9) Encinas, M. V.; Lissi, E. A.; Zanocco, A.; Stewart, L. C.; Scaiano,
J. C. Can. J. Chem. 1984, 62, 386-391, and references to earlier work
cited therein.
(10) For a recent review see Kenyon, G. L.; Gerlt, J. A.; Petsko, G. A.;
Kozarich, J. W. Acc. Chem. Res. 1995, 28, 178-186.
(11) (a) Gerlt, J. A.; Gassman, P. G. J. Am. Chem. Soc. 1993, 115,
11552-11568. (b) Guthrie, J. P.; Kluger, R. J. Am. Chem. Soc. 1993, 115,
11569-11572.
(12) Ciganek, E. J. Org. Chem. 1970, 35, 862-864.
(13) Kresge, A. J.; Mathew, L.; Popik, V. V. J. Phys. Org. Chem. 1995,
8, 552-558.
(14) Thayer, F. K. Organic Syntheses; Wiley: New York, 1932; Collect.
Vol. I, pp 12-13.
(15) Eisenberg, D.; Kauzmann, W. The Structure and Properties of
Water; Oxford University Press: New York, 1969; pp 182-187.

The Mandelic Acid Keto-Enol System in Aqueous Solution
J. Am. Chem. Soc., Vol. 119, No. 42, 1997 10205
hydrochloric acid and in 0.01 M perchloric acid, and for
phenyldiazoacetic acid16 as the substrate in acetic acid buffers
of pH = 5 and in 0.02 M sodium hydroxide. Mandelic acid
was found to be the principal product in all cases; no phenyl-
glyoxal was formed from the benzoylformate ester substrates,
and no acetylmandelic acid was formed from phenyldiazoacetic
acid in acetic acid buffers.
Flash Photolysis: Enol Generation. Flash photolysis of
either the methyl or butyl esters of benzoylformic acid (3) or
phenyldiazoacetic acid (4) in aqueous solution produced a
transient species that absorbed strongly at λ ) 270-290 nm in
acidic solutions and at λ ) 320-340 nm in basic solutions.
The styrene-type chromophore of mandelic acid enol, 1, can
be expected to have an absorption based in this regionsthe
Figure 1. Rate profiles for the generation (upper lines) and ketonization
(lower line) of mandelic acid enol, using benzoylformic acid esters
(O) and phenyldiazoacetate ion (4) as flash photolysis substrates.
closely related â,â-dimethoxystyrene, 7, for example absorbs
at λ ) 268 nm17sand a shift to longer wavelength in going
from acidic to basic solution, in which the enol will be ionized
to enolate ion (Vide infra), is characteristic of the conversion
of enols to enolate ions.18 That, plus the fact that flash
photolysis of benzoylformate esters is known to produce
phenylhydroxyketene,9 and ketenes are hydrated rapidly to
carboxylic acid enols,19 suggests that this transient is the enol
of mandel acid. This suggestion is confirmed by the form of
acid-base catalysis and the solvent isotope effects shown by
the decay of this transition (Vide infra).
Both the appearance and the decay of this transient species
conformed to the first-order rate law well. When the appearance
and decay were widely separated in time, the data for appearance
and decay were analyzed separately, using least-squares fitting
of single exponential functions, and when the two rates were
similar, fitting was done using a double exponential expression.
Rate measurements were made in aqueous solution over the
range of acidity [H+] ) 0.1-10-11 M, using perchloric acid
and sodium hydroxide solutions as well as formic acid, acetic
acid, biphosphate ion, hydrogen tert-butylphosphonate ion, and
tris(hydroxymethyl)methylammonium ion buffers. The ionic
strength of the reaction solutions was maintained at 0.10 M.
The benzoylformate ester substrates underwent thermal
hydrolysis under strongly basic conditions and were conse-
quently not used above pH ) 11, and the phenyldiazoacetic
acid substrate underwent thermal hydrolysis in strongly acidic
solutions and was not used below pH ) 3.
Figure 2. Comparison of rates of formation of mandelic acid enol in
acetic acid buffer solutions, buffer ratio ) 1, using n-butyl benozyl-
formate (O) and phenyldiazoacetic acid (4) as the flash photolysis
substrates.
hydroxide solutions, are displayed as the upper rate profiles of
Figure 1. Hydrogen ion concentrations of the buffer solutions
needed to construct this profile were obtained by calculation,
using literature acidity constants of the buffer acids and acitivity
coefficients recommended by Bates.21
It may be seen that the rate profile obtained using benozyl-
formic acid esters as flash photolysis substrates (Figure 1,
circles) is hardly different from that obtained using phenyl-
diazoacetic acid as the substrate (Figure 1, triangles) in neutral
and basic solutions. There is, however, a decided difference
in acidic solutions. Least squares analysis of the data gives
the following results, for the ester and diazoacid substrates
respectively, kH+ ) (5.45 ( 0.53) × 107 M-1 s-1 and (9.70 (
1.85) × 109 M-1 s-1, ko ) (5.51 ( 0.43) × 105 s-1 and (6.06
-
( 0.44) × 105 s-1, and kHO ) (2.33 ( 0.43) × 109 M-1 s-1
The rate data for appearance of the enol transient are
summarized in Tables S1-S6.20 The measurements in buffers
were performed in series of solutions of constant buffer ratio
but varying buffer concentration; hydrogen ion concentrations
therefore remained constant within a given solution series.
Observed first-order rate constants increased linearly with buffer
concentration, and extrapolation of the data to zero buffer
concentration gave rate constants that represent reaction through
catalysis by solvent-derived species only. These rate constants,
together with those measured in perchloric acid and sodium
and (4.00 ( 0.21) × 109 M-1 s-1
.
A further difference between the two kinds of substrate is
shown by their behavior in buffer solutions. As is illustrated
in Figure 2, whereas generation of the enol using phenyldiazo-
acetic acid as the flash photolysis substrate is strongly catalyzed
by acetic acid buffers, that using benzoylformate ester substrates
is hardly affected at all by changes in buffer concentration. Least
squares analysis of the data shown in Figure 2 gives kobs/s-1
)
(9.00 ( 4.24) × 105 + (6.67 ( 0.62) × 108 [buffer] for
phenyldiazoacetic acid as the substrate and kobs/s-1 ) (6.64 (
0.14) × 105 + (8.06 ( 2.66) × 105 [buffer] for n-butyl
benzoylformate as the substrate.
(16) The pKa of this acid is 3.70,13 and it was consequently present as
the phenyldiazoacetate ion in most of the studies performed here.
(17) Okuyama, T.; Kawao, S.; Fueno, T. J. Org. Chem. 1981, 46, 4372-
4375.
These differences in behavior indicate that the two different
kinds of substrate, though they both produce mandelic acid enol
(Vide infra), do this by different reactions through different enol
(18) Haspra, P.; Sutter, A.; Wirz, J. Angew. Chem., Int. Ed. Engl. 1979,
18, 617-619.
(19) Tidwell, T. T. Ketenes; Wiley Interscience: New York, 1995; pp
576-585.
(20) Supporting Information: see paragraph at the end of this paper
regarding availability.
(21) Bates, R. G. Determination of pH Theory and Practice; Wiley: New
York, 1973; p 49.

10206 J. Am. Chem. Soc., Vol. 119, No. 42, 1997
Chiang et al.
precursors. The rate profile given by the benzoylformic acid
ester substrates, showing a broad region of uncatalyzed reaction
and only weak acid and base catalysis, is typical of ketene
hydrations.7a,d,f,h,22 Ketenes, moreover, react only weakly with
buffers such as those used here,23 and they also give only weak
solvent isotope effects such as we found by comparing rates of
enol formation from n-butyl benzoylformate in H2O and D2O
solution (Table S220): kH O/kD O ) 1.49 ( 0.22.7d,22b,24 Fur-
Oxygen-18 Tracer Study. A photo-Wolff reaction29 of
phenyldiazoacetic acid would produce phenylhydroxyketene,
whose hydration in oxygen-18 labeled water would give
mandelic acid with the isotopic label in its carboxylic acid group,
eq 7. On the other hand, hydration of phenylcarboxycarbene,
2
2
thermore, the intensity of the transient signal given by the enol
dropped markedly in the region of acid-catalyzed ketene
hydration, consistent with the fact that, whereas uncatalyzed
and hydroxide-ion catalyzed ketene hydrations occur by a
mechanism involving enol intermediates,19 the acid-catalyzed
reaction takes place by a different route that avoids enols.22d,24
This behavior, plus the fact that flash photolysis of benzoyl-
formate esters is known to produce phenylhydroxyketene,9
makes it safe to conclude that mandelic acid enol was generated
from this ketene in the present study by the hydration reaction
shown in eq 4.
It seems likely, on the other hand, that the different enol
precursor formed when phenyldiazoacetic acid is the flash
photolysis substrate is phenylcarboxycarbene, 6, as shown in
eq 5. It is well-known that irradiation of aliphatic diazo
compounds with light produces carbenes.25 The overall conver-
sion of phenylcarboxycarbene into mandelic acid, moreover,
corresponds to insertion of a carbene into an O-H bond of
water, which is also a well-known reaction.26 We have, in fact,
observed a process similar to that of eq 5 upon flash photolysis
of methyl phenyldiazoacetate, eq 6,27 and also of its cyclic
generated by photoinduced loss of nitrogen from phenyldiazo-
acetic acid in oxygen-18 labeled water, would give mandelic
acid with the isotopic label in its hydroxyl group, eq 8.
We investigated this difference by photolyzing (Rayonet, λ
) 254 nm) 0.01 M phenyldiazoacetic acid in oxygen-18 labeled
water to which an equal volume of tetrahydrofuran had been
added in order to produce a homogeneous solution. Isotopic
analysis of the mandelic acid produced was performed by mass
spectroscopy, taking advantage of the fact that mandelic acid
fragments readily upon electron impact producing the phenyl-
hydroxymethyl cation as the principal species in its mass
spectrum, eq 9.30 Results obtained in 0.01 M hydrochloric acid
solution showed 96% of the mandelic acid obtained to be
isotopically labeled in the hydroxyl group and 4% to be
unlabeled at that position, and results obtained in 0.01 M sodium
hydroxide solution showed 80% of the mandelic acid labeled
in the hydroxyl group and 20% unlabeled at that position.
These results support the formation of phenylcarboxycarbene
as the enol precursor produced by flash photolysis of phenyl-
diazoacetic acid, and they show that this carbene is the principal
precursor produced from this substrate. They suggest as well,
however, that a minor amount of phenylhydoxyketene is also
formed and that some Wolff-rearrangement of phenyldiazoacetic
acid does in fact take place.
Flash Photolysis: Enol Ketonization. Rates of ketonization
of mandelic acid enol were measured in aqueous perchloric acid
and sodium hydroxide solutions and in formic acid, acetic acid,
biphosphate ion, hydrogen tert-butylphosphonate ion, and tris-
(hydroxymethyl)methylammonium ion buffers. The data so
obtained are summarized in Tables S7-S9.20
The measurements in buffers were once again performed in
series of solutions of constant buffer ratio and constant ionic
strength (0.10 M) but varying buffer concentration, and observed
first-order rate constants were again linearly proportional to
buffer concentration. Extrapolation of the data to zero buffer
concentration then gave rate constants which, together with the
values measured in perchloric acid and sodium hydroxde
solution, were used to construct the lower rate profile shown in
Figure 1.
The circles in this rate profile represent rate constants obtained
using benzoylformic acid esters (methyl, n-butyl, and isobutyl)
as flash photolysis substrates, and the triangles represent those
obtained using phenyldiazoacetic acid as the substrate. It may
be seen that there is no difference between the two sets of data
analog, 4-diazo-3-isochromanone.28 The fact that we observe
enol intermediates in these reactions shows that they do not
simply involve water and the carbenic carbon atom but that they
might be more accurately described as conjugate additon of
water across the entire carbonylcarbene function.
These differences in kinetic behavior require that most of
the mandelic acid enol formed by flash photolysis of phenyl-
diazoacetic acid is generated from an intermediate that is not
phenylhydroxyketene. This is supported by the oxygen-18 tracer
study described below, but the tracer study also suggests that a
minor amount of enol obtained from that substrate does arise
via the ketene and that some Wolff-rearrangement does in fact
take place.
(22) (a) Bothe, E.; Dessouki, A. M.; Schulte-Frohlinde, D. J. Phys. Chem.
1980, 84, 3270-3272. (b) Allen, A. D.; Kresge, A. J.; Schepp, N. P.;
Tidwell, T. T. Can. J. Chem. 1987, 65, 1719-1723. (c) Allen, A. D.;
Tidwell, T. T. J. Am. Chem. Soc. 1987, 109, 2774-2780. (d) Allen, A. D.;
Stevenson, A.; Tidwell, T. T. J. Org. Chem. 1989, 54, 2843-2848. (e)
Chiang, Y.; Kresge, A. J.; Popik, V. V. J. Am. Chem. Soc. 1995, 117, 9165-
9171. Chiang, Y.; Guo, H.-X.; Kresge, A. J.; Tee, O. S. J. Am. Chem. Soc.
1996, 118, 3386-3391.
(23) Andraos, J.; Kresge, A. J. J. Am. Chem. Soc. 1992, 114, 5643-
5646.
(24) Andraos, J.; Kresge, A. J. J. Photochem. Photobiol. A 1991, 57,
165-173.
(25) Regitz, M.; Maas, G. Diazo Compounds Properties and Synthesis;
Academic Press: New York, 1986; Chapter 4.
(26) Kirmse, W. Carbene Chemistry, 2nd ed.; Academic Press: New
York, 1971; pp 423-430.
(27) Chiang, Y.; Kresge, A. J.; Pruszynski, P.; Schepp, N. P.; Wirz, J.
Angew. Chem., Int. Ed. Engl. 1991, 30, 1366-1368.
(28) Jefferson, E. A.; Kresge, A. J. Unpublished work.
(29) Regitz, M.; Maas, G. Diazo Compounds Properties and Synthesis;
Academic Press: New York, 1986; pp 185-195.
(30) Sharp, R. T. Org. Mass Spec. 1980, 15, 381-382.

The Mandelic Acid Keto-Enol System in Aqueous Solution
J. Am. Chem. Soc., Vol. 119, No. 42, 1997 10207
and that both kinds of substrate are producing the same transient
species. This similarity is corroborated by the buffer-dependent
portions of observed rate constants, as is illustrated in Figure
3. This figure shows that there is no significant difference
between rate constants measured in biphosphate buffers using
methyl benzoylformate, isobutylbenzoylformate, and phenyl-
diazoacetic acid as the flash photolysis substrates, which stands
in marked contrast to the different behavior shown by ester and
diazoacid substrates illustrated in Figure 2. Least squares
analysis of the data shown in Figure 3 gives kobs/s-1 ) (1.61 (
0.04) × 105 + (3.21 ( 0.22) × 106 [buffer] for methyl
benzoylformate as the substrate, kobs/s-1 ) (1.55 ( 0.04) ×
105 + (3.80 ( 0.20) × 106 [buffer] for isobutyl benzoylformate
as the substrate, and kobs/s-1 ) (1.43 ( 0.02) × 105 + (3.68 (
0.10) × 106 [buffer] for phenyldiazoacetic acid as the substrate.
This similarity, of course, is the behavior expected on the basis
of the reaction schemes of eqs 4 and 5.
The lower rate profile of Figure 1 is typical of enol
ketonization reactions,5 and it may be interpreted in the usual
way in terms of reaction through rate-determining â-carbon
protonation of either enol or enolate ion and either hydronium
ion or water as the proton source, as shown in eq, 10.31 The
diagonal region of acid catalysis at the high acidity end of this
rate profile then represents ketonization of un-ionized enol with
Figure 3. Comparison of rates of ketonization of mandelic acid enol
in biphosphate ion buffer solutions, buffer ratio ) 1/3, using methyl
benzoylformate (O), isobutyl benzoylformate (]), and phenyldiazo-
acetic acid (4) as the flash photolysis substrates.
reaction at [H+] ) 10-2-10-3 M, which, because of the
ambiguity regarding reaction mechanism in this region, is simply
designated kuc. Least squares fitting of the data using this
expression produced the following results: kH+ ) (9.03 ( 0.27)
× 103 M-1 s-1, kuc ) (3.89 ( 0.08) × 102 s-1, k′o ) (1.70 (
0.04) × 105 s-1, and QaE ) (4.12 ( 0.18) × 10-7 M, pQaE )
6.39 ( 0.02.33
As was pointed out above, the process represented by kuc in
the rate law of eq 11 could be either carbon protonation of un-
ionized enol by water or ionization of the enol to enolate ion
followed by carbon protonation of enolate by hydronium ion.
In the latter case, this part of the rate law becomes k′H+QaE with
k′H+ being the rate constant for carbon protonation of the enolate
+
hydronium ion acting as the protonating agent. The short
ion; least squares analysis then gives k′H ) (8.62 ( 0.47) ×
108 M-1 s-1, with kH+, k′o, and QEa as listed above. This value
of k′H+, though large, does not exceed the diffusion-controlled
limit, and it consequently cannot be used to rule out this reaction
path.
Support for interpretation of the process we observe as the
ketonization of mandelic acid enol according to the reaction
scheme of eq 10 comes from its behavior in buffer solutions.
As Figure 3 illustrates, rates of reaction increase with increasing
buffer concentration. The nature of this buffer catalysis was
investigated with the aid of eq 12, in which kcat is the slope of
a buffer dilution plot (∆kobs/∆[buffer]),34 kB and kHA are general
portion of apparently uncatalyzed reaction at [H+] ) 10-2
-
10-3 M immediately following this diagonal region could
represent either un-ionized enol reacting with water as the proton
source or ionization of the enol to the much more reactive
enolate ion followed by carbon protonation of that by hydronium
ion. It is sometimes possible to exclude the second of these
two possibilities because the data require a rate constant for
the carbon protonation step that is impossibly large, but that is
not the case here (Vide infra). In the second diagonal region of
the rate profile following this uncatalyzed reaction, ketonization
still begins with un-ionized enol as the initial state, but ionization
to the more reactive enolate ion is followed by carbon
protonation with water as the proton source; since the concen-
tration of enolate ion under these conditions is inversely
proportional to [H+] and hydroniuum ion is not involved in the
rate-determining step, this produces an apparent hydroxide-ion
catalysis. At sufficiently low acidities, where the enol ionization
equilibrium shifts over to enolate ion, this hydroxide-ion
catalysis becomes saturated, and the second region of uncata-
lyzed reaction above [H+] ) 10-7 M results. This saturation
of hydroxide-ion catalysis produces a bend in the rate profile
that corresponds to the acidity constant of the enol.
kcat ) kB + (kHA - kB)fA
(12)
base and general acid catalytic coefficients respectively, and fA
is the fraction of buffer present in the acid form. Application
of this equation showed general base catalysis in formic and
acetic acid buffers and general acid catalysis in biphosphate
ion buffers. This behavior is consistent with the fact that the
measurements in formic and acetic acid buffers were made in
the region of the rate profile showing base catalysis, where
reaction is expected to take place by conversion of the enol to
the more reactive enolate ion followed by rate-determining
carbon protonation of the latter, eq 13.
The rate law that applies to the reaction scheme of eq 10 is
shown as eq 11. The rate constants in this expression are as
The rate-determining carbon protonation step of such a
reaction sequence will, of course, show general acid catalysis,
kobs ) kH+[H+] + kuc + k′oQEa /(QEa + [H+])
(11)
(32) Chiang, Y.; Kresge, A. J.; Walsh, P. A.; Yin, Y. J. Chem. Soc.,
Chem. Commun. 1989, 869-871.
defined in eq 10, except for that representing the uncatalyzed
(31) We have written the ionization of mandelic acid enol as involving
the hydroxyl groups trans to phenyl because the trans enol of phenylac-
etaldehyde is a stronger acid than the cis isomer.32 Ionization of the hydroxyl
group geminal to phenyl can be ruled out because ketonization would then
produce phenylglyoxal, which is not found as a reaction product.
(33) This is a concentration dissociation constant, applicable at ionic
strength ) 0.10 M.
(34) The slopes of buffer dilution plots for some acetic acid and
biphosphate ion buffers were adjusted to compensate for the fact that the
enol existed in both ionized and un-ionized forms in these solutions.

10208 J. Am. Chem. Soc., Vol. 119, No. 42, 1997
Chiang et al.
Table 2. Summary of Rate and Equilibrium Constantsa
Table 1. General Acid Catalytic Coefficients for Rate-Determining
Carbon Protonation of Mandelic Acid Enolate Iona
acid
HCO2H
pKa
k′HA/106 M-1 s-1
3.75
4.76
7.20
8.07
8.71
58.6
33.5
14.7
8.96
4.80
CH3CO2H
-
H2PO4
+
(CH2OH)3CNH3
t-BuPO3H-
a Ionic strength ) 0.10 M.
but this will be converted into general base catalysis for the
overall process by the action of base upon the enol in the prior
equilibrium. The biphosphate ion buffers, on the other hand,
were used in a region of the rate profile beyond the acidity
constant of the enol where the enol was converted to enolate
ion. Reaction here is expected to take place by simple carbon
protonation of the enolate substrate, i.e., by the last step of the
sequence shown in eq 13, which will give general acid catalysis.
Thus, both the fact that buffer catalysis is observed and the
nature of this catalysis are wholly consistent with the identifica-
tion of this reaciton as ketonziation of mandelic acid enol.
Rate measurements in hydrogen tert-butylphosphonic acid and
tris(hydroxymethyl)methyl ammonium ion buffers were per-
formed at only one buffer ratio and eq 12 could therefore not
be applied. In these solutions, however, as in the biphosphate
ion buffers, the enol was converted to enolate ion, and it seems
safe to assume that general acid catalysis was operating here as
well. The data were therefore treated accordingly.
The general base catalytic coefficients for formic and acetic
acids could be converted into general acid coefficients for rate-
determining carbon protonation of the enolate ion by using the
relationship shown in eq 14, in which QHa A is the acidity
constant of the catalyzing acid.33 The
kB ) k′HAQEa /QHa A
(14)
results so obtained, together with the directly determined values
for the other buffer acids, are listed in Table 1. It may be seen
that these catalytic coefficients form a sensible series, decreasing
in value with decreasing acid strength of the catalyst. They
do, in fact, give a reasonably good Brønsted relation, shown in
Figure 4, despite the fact that the catalysts they represent vary
considerably in charge type. The exponents of Brønsted
relations are believed to provide a measure of transition state
structure,35 and the low value produced in the present case, R
) 0.24 ( 0.01, implies an early, reactant-like transition state,
which is consistent with the fact that the present reactions are
quite rapid.36
Further support for the identification of the present reaction
as the ketonization of mandelic acid enol comes from isotope
effects based upon rate measurements made in D2O solutions
of perchloric acid and sodium hydroxide. The data are
summarized in Tables S7 and S8, and the results in perchloric
acid are compared with those for H2O solutions of this acid in
Figure 5.
a 25 °C; ionic strength ) 0.10 M.
This figure shows that rates of reaction are slower in D2O
than in H2O throughout the range of acidity investigated but
that the difference is greater in the region of apparent hydroxide
ion catalysis at low acidity than in the region of acid catalysis
at higher acidity. This is the behavior expected on the basis of
the reaction scheme of eq 10, for, whereas the isotope effect in
the region of acid catalysis is on a rate constant alone (kH+),
that in the region of base catalysis is on the product of a different
rate constnt (k′o) and an equilibrium constant (QE). Least
squares analysis of the D2O data gave results that, when
a
combined with H2O values, produced the following isotope
effects: kH+/kD ) 3.25 ( 0.12 and (k′oQE)H O/(k′oQE)D O
)
+
2
2
a
a
30.8 ( 2.9. The latter could be separated into its constituent
parts with the aid of (k′o)H O/(k′o)D O ) 6.90 ( 0.54 based upon
2
2
(35) Kresge, A. J. In Proton Transfer Reactions; Caldin, E. F., Gold,
V., Eds.; Chapman and Hall: London, 1975; Chapter 7.
(36) Hammond, G. S. J. Am. Chem. Soc. 1955, 77, 334-338.
the measurements made in sodium hydroxide solutions; the
result is (QE)H O/(QE)D O ) 4.47 ( 0.55.
2
2
a
a

The Mandelic Acid Keto-Enol System in Aqueous Solution
J. Am. Chem. Soc., Vol. 119, No. 42, 1997 10209
more reliable but operationally more difficult measure of
enolization, takes place several times more slowly than race-
mization. Since the acid-catalyzed enolization of carbonyl
compounds gives inverse solvent isotope effects and is conse-
quently faster in D2O than in H2O,5 deuterium incorporation
should have been faster rather than slower than racemization,
if both reactions were indeed taking place by an enolization
pathway. Racemization of mandelic acid, of course, could occur
by another pathway, e.g., by protonation of its R-hydroxyl group
followed by displacement of that by a water molecule. It is
difficult, on the other hand, to conceive of a reasonable
nonenolization pathway for R-hydrogen exchange. We con-
clude, therefore, that acid-catalyzed racemization does not occur
via enolization, and we now use deuterium incorporation to
monitor enolization.
Deuterium incorporation into the R-position of mandelic acid
is a very slow process, and concentrated acid solutions and high
temperatures had to be used to obtain reasonable reaction rates.
Rate measurements were made in 4 M DCl/D2O solutions at
five temperatures in the range 130-150 °C, and at two of these
temperatures (140 and 150 °C) measurements were also made
over a range of acidities, from 4 M down to 0.5 M. These data
are summarized in Table S11.20
Figure 4. Brønsted plot for the ketonization of mandelic acid enol.
As is commonly the case for reactions conducted in concen-
trated acid solutions, observed first-order rate constants increased
with acidity more rapidly than in direct proportion to acid
concentration; an acidity function was therefore used to
extrapolate the data down to dilute solution, in order to obtain
the second-order hydronium-ion catalytic coefficient needed for
calculating a keto-enol equilibrium constant. The Cox-Yates
technique using X functions39 appears to be the best method
currently available for this purpose.40 The X scale has not been
determined for DCl/D2O solutions, but the values of other acidity
functions have been found not to change in going from H2O to
D2O when the comparison is made at the same acid molarity,41
and the X scale for HCl/H2O42 was consequently used instead.
Values of X at the temperatures of the kinetic measurements
were calculated from 25 °C values using the relationship given
in eq 15.43 The correlation function employed was the standard
Figure 5. Comparison of rates of ketonization of mandelic acid enol
in H2O (O) and D2O (4) solutions of perchloric acid.
The first of these isotope effects is a typcial value for rate-
determining hydron transfer from the hydronium ion,37 and
similar results have been obtained for ketonization of other
enols.5 This kind of isotope effect actually consists of a primary
component in the normal direction, kH/kD > 1, plus an offsetting
secondary component in the inverse direction, kH/kD < 1,
produced by changes in the nonreacting bonds of the hydronium
ion.37 Isotope effects on hydron transfer from a water molecule
also contain a secondary component in addition to a normal
primary component, but now the secondary component is in
the normal direction and consequently augments the primary
effect. It is significant, therefore, that the isotope effect observed
here on k′o, which according to the scheme of eq 10 represents
protonation of enolate ion by a water molecule, is greater than
that on kH+. The isotope effect on QEa is as expected as well:
isotope effect theory requires oxgyen acids to be more com-
pletely ionized in H2O than in D2O,37b and the value observed
here is typical for acids of the strength of the present enol.38
Enolization. In the early stages of the present investigation,
we measured rates of acid-catalyzed racemization of mandelic
acid, and, in the belief that racemization was occurring by an
enolization pathway, we used these rate constants to determine
the keto-enol equilibrium constant reported in a preliminary
publication of this work.7b These data are summarized in Table
S10.20
+
Cox-Yates expression, shown as eq 16, in which kD is the
desired dilute solution rate constant and m is a slope parameter.
When X is adjusted for temperature differences according to
XT ) X25[298.15/(273.15 + T°C)]
log(kobs/[D+]) ) log kH+ + mXT
(15)
(16)
eq 15, m is expected to be temperature invariant.43 This was
found to be the case in the present work for the two temperatures
at which rates were measured over a range of acid concentra-
tions: m ) 0.61 ( 0.03 at 139.6 °C and m ) 0.69 ( 0.03 at
151.2 °C. The average of these two values was therefore used
to extrapolate down to dilute solution from the rate constants
measured in 4 M acid at all temperatures.
The dilute solution hydronium-ion catalytic coefficients
obtained in this way gave a good linear Eyring plot, shown in
Figure 6. Least squares analysis produced the activation
parameters ∆Hq ) 30.9 ( 0.3 kcal mol-1 and ∆Sq ) -9.4 (
We later discovered, however, that acid-catalyzed deuterium
incorporation from D2O into the R-position of mandelic acid, a
(39) Cox, R. A.; Yates, K. J. Am. Chem. Soc. 1978, 100, 3861-3867.
(40) Kresge, A. J.; Chen, H. J.; Capen, G. L.; Powell, M. F. Can. J.
Chem. 1983, 61, 249-256.
(37) (a) Kresge, A. J.; Sagatys, D. S.; Chen, H. L. J. Am. Chem. Soc.
1977, 99, 7228-7233. (b) Kresge, A. J.; More O’Ferrall, R. A.; Powell,
M. F. In Isotopes in Organic Chemistry; Buncel, E., Lee, C. C., Eds.;
Elsevier: Amsterdam, 1987; Vol. 7, pp 177-273.
(38) Laughton, P. M.; Robertson, R. E. In Solute-SolVent Interactions;
Coetzee, J. F., Ritchie, C. D., Eds.; Dekker: New York, 1969; Chapter 7.
(41) Ho¨gfeldt, E.; Bigeleisen, J. Am. Chem. Soc. 1960, 82, 15-20. Cox,
R. A.; Lam, S.-O.; McClelland, R. A.; Tidwell, T. T. J. Chem. Soc., Perkin
Trans. II 1979, 272-275.
(42) Cox, R. A.; Yates, K. Can. J. Chem. 1981, 59, 2116-2124.
(43) Cox, R. A.; Goldman, M. F.; Yates, K. Can. J. Chem. 1979, 57,
2960-2966.

10210 J. Am. Chem. Soc., Vol. 119, No. 42, 1997
Chiang et al.
equilibrium protonation on carbonyl carbon followed by rate-
determining removal of R-hydrogen by a water molecule, eq
17. Conversion of the relatively loose bonds of the hydronium
ion into the tighter bonds of a water molecule in the equilibrium
step of this reaction sequence produces an inverse solvent
isotope effect whose magnitude is usually, ca. 0.5; for example,
+
+
kH+/kD ) 0.57 for the enolization of acetaldehyde,3 kH+/kD
Figure 6. Eyring plot for the acid-catalyzed incorporation of deuterium
into the R-position of mandelic acid.
) 0.46-0.54 for enolization of acetone,46 and kH+/kD+ ) 0.56
for enolization of isobutyrophenone.47 No determination of this
isotope effect for the enolization of a carboxylic acid appears
to have been made, and, in the absence of more definite
information, we shall use the nominal value kH+/kD ) 0.5 to
+
convert the rate constant for deuterium incorporation in D2O
into a rate constant for enolization in H2O: kEH+ ) 0.5 (1.17 (
0.18) × 10-12 ) (5.85 ( 0.91) × 10-13 M-1 s-1. Combination
of this result with the rate constant for hydronium-ion-catalyzed
ketonization then leads to the keto-enol equilibrium constant
KE ) kEH+/kHK+ ) (5.85 ( 0.91) × 10-13/(9.07 ( 0.31) × 103 )
(6.48 ( 1.02) × 10-17, pKE ) 16.19 ( 0.07. The error limit
cited here is a statistical uncertainty (standard deviation) which
does not include possible systematic errors; allowance for that,
plus the uncertainty in the estimated solvent isotope effect, leads
to the probably more realistic assessment pKE ) 16.2 ( 0.2.
The latter estimate is different, by a factor of ca. 6, from the
result, pKE ) 15.4, reported earlier based upon the misguided
assumption that acid-catalyzed racemization of mandelic acid
occurs by an enolization pathway.7b
As the thermodynamic cycle of eq 18 shows, this equilibrium
constant connecting un-ionized keto and enol isomers is related
Figure 7. Free energy diagram comparing the mandelic acid keto-
enol system to racemization catalzyed by mandelate racemase.
0.8 cal K-1 mol-1, and extrapolation down to 25 °C gave the
rate constant k25 ) (1.17 ( 0.18) × 10-12 M-1 s-1
.
The rate constant kD+ ) (3 ( 1) × 10-5 M-1 s-1 has recently
been reported for acid-catalyzed deuterium incorporation into
mandelic acid at 170 °C.44 Extrapolation of the present data to
to the keto-enol equilibrium constant for ionized species by
+
this temperature gives kD ) (4.6 ( 0.1) × 10-5 M-1 s-1
,
two acidity constants, that for the carboxyl group of mandelic
consistent with this reported value.
aicd, Qa, and that for conversion of enol to enolate ion, QE.
a
(45) The rate constant for ketonization in D2O with D3O+ as the catalyst
is also available from the present work. However, comparison of that with
the enolization rate constant in D2O, in order to avoid making an estimate
of the solvent isotope effect, involves an unknown primary isotope effect:
enolization, i.e., deuterium incorporation, was necessarily monitored using
R-protiomandelic acid as the substrate; H was consequently the hydrogen
in flight in the rate-determining step in these experiments, whereas D was
in flight inthe D3O+/D2O ketonization measurements. Primary isotope effects
are usually greater and more variable than solvent isotope effects and are
consequently more difficult to estimate.
Discussion
Keto-Enol and Related Equilibria. An estimate of the
keto-enol equilibrium constant for mandelic acid may be made
by combining the rate constants for hydronium-ion-catalyzed
enolization and ketonization determined here. Allowance must
be made, however, for the fact that the enolization rates were
measured in D2O, where D3O+ was the catalyst, whereas the
ketonization rates were measured in H2O, where H3O+ was the
catalyst.45 The well established mechanism for hydronium-ion-
catalyzed enolization of carbonyl compounds5 consists of
(46) Reitz, O. Naturwissenschaften 1936, 24, 814-814. Albery, W. J.;
Gelles, J. S. J. Chem. Soc., Faraday Trans. 1 1982, 78, 1569-1578. Shelly,
K. P.; Venimadhavan, S.; Nagaranjan, K.; Stewart, R. Can. J. Chem. 1989,
67, 1274-1282.
(44) Bearne, S. L.; Woflenden, R. Biochemistry 1997, 36, 1646-1656.
(47) Keeffe, J. R.; Kresge, A. J. Can. J. Chem. 1996, 74, 2481-2486.

The Mandelic Acid Keto-Enol System in Aqueous Solution
J. Am. Chem. Soc., Vol. 119, No. 42, 1997 10211
The latter constant has been determined in the present work,
and the former may be estimated by applying activity coef-
ficients21 to a literature value of the thermodynamic acidity
constant of mandelic acid.48 Use of these values then leads to
the result K′E ) (4.55 ( 0.75) × 10-20, pK′E ) 19.34 ( 0.07.
The reaction scheme of eq 18 also allows evaluation of the
acidity constant of mandelic acid ionizing as a carbon acid,
QKa ) (2.67 ( 0.44) × 10-23, pQKa ) 22.57 ( 0.07.33
for the enol of acetaldehyde3 and pQaE ) l0.94 for the enol of
acetone.50 Comparison of the value for acetaldedhye enol, 14,
with pQEa ) 9.46 for the trans enol of phenylacetaldehyde,
Comparison with Other Systems. The keto-enol equilib-
rium constant for mandelic acid determined here, pKE ) 16.2,
is many orders of magnitude less than those for simple aldehydes
and ketones, e.g., pKE ) 6.23 for acetaldehyde3 and pKE ) 8.33
for acetone.49 This striking difference may be attributed to
resonance stabilization of the keto form in the carboxylic acid
system, produced by interaction of its hydroxyl and carbonyl
moieties; this interaction is lost in the enol, and it consequently
serves to increase the energy difference between keto and enol
forms, thus making KE as small as it is.
15,32 suggests that about one pK unit of the increased acidity
of mandelic acid enol is due to the phenyl substituent. The
â-hydroxy group of mandelic acid enol, on the other hand,
probably has a slight acid-weakening effect, as evidenced by
pQEa ) 10.40 for acetophenone enol, 1651 and pQEa ) 10.86 for
the enol of R-methoxyacetophenone, 17.52 It seems likely that
the major factor influencing the acidity of mandelic acid enol
is the presence of a second hydroxyl group geminal to the first:
gem-diols in saturated systems are ca. 2 pK units more acidic
than the corresponding alcohols,53 and this effect of the second
hydroxyl might well be transmitted more efficiently through
an unsaturated carbon atom and thus be stronger in gem-
enediols.
The value for mandelic acid, on the other hand, is four orders
of magnitude greater than the estimte for acetic acid, pKE )
20.2 It is instructive to compare this difference with the effect
of a â-phenyl group on keto-enol equilibrium constants, as
assessed by the difference between pKE ) 6.23 for acetaldehyde,
8, and pKE ) 2.88 for phenylacetaldehyde, 9.32 This difference,
The acid strength of mandelic acid enol is similar to that of
the enol of methyl mandelate, for which pQEa ) 6.62,27
showing that the geminal effects of hydroxyl and methoxyl are
similar. The acidity of the enols of cyclopentadiene carboxylic
acid and its benzo analogs, 11-13, however, is considerably
greater: pQEa ) 1.3-2.0 for these substances7a,g,i and ) 0.4-
1.3 for the enols of some of the ring-sulfonated derivatives of
12.7d The remarkably strong acidity of these substances may
be attributed to delocalization of the negative charge of their
enolate ions into the cyclopentadiene ring, which creates an
aromatic system. The enol of cyanophenylacetic acid, 10, is
also very acidic: pQaE ) 0.99 for this substance.7e The
difference between this value and pQEa ) 6.35 for mandelic
acid enol attests to the well-known very strong acidifying effect
of the cyano group.
Mandelate Racemase. We have found in the present study
that racemization of mandelic acid and deuterium incorporation
into its R-position occur at different rates in acid solution and
that acid-catalyzed racemization therefore cannot be taking place
by an enolization pathway. The situaton is different in neutral
and basic solution: in these media racemization and deuterium
incorporation do occur at the same rate,44 which implies that
the base-catalyzed racemization that takes place under these
conditions does occur by an enolization pathway.
∆pKE ) 3.35, is not unlike the four pK difference between
mandelic and acetic acids, which suggests that most of the
increased enol content of mandelic acid is due to the phenyl
group and that the â-hydroxyl substituent makes little contribu-
tion.
The effect of a â-cyano group, on the other hand, is very
much stronger: pKE ) 7.22 for cyanophenylacetic acid, 10.7e
This very powerful influence of the cyano group is much
stronger than its simple double bond stabilizing effect, D )
2.3 kcal/ mol-1 8a
, and must be due in large part to an additional
synergistic interaction of the electronegative cyano group and
electropositive enol hydroxyl groups, through the enol double
bond. The corresponding interaction of a hydroxyl group in
place of cyano is, of course, antagonistic because the groups
involved here are all electropositive, and that works against the
quite appreciable simple double-bond stabilizing effect of
hydroxyl, D ) 5.4 kcal mol-1 8b
.
Cyclopentadiene carboxylic acid, 11, and its monobenzo, 12,
and dibenzo, 13, analogs also have keto-enol equilibrium
constants that are very much greater than the estimate pKE )
20 for acetic acid: for these substances, pKE lies in the range
The racemization of mandelic acid in neutral solution is also
catalyzed very efficiently by the enzyme mandelate racemase,
and recent detailed mechanistic studies10 have shown that the
enzymatic reaction occurs by an enolization mechanism as well.
Our results provide a quantitative measure of the thermodynamic
barrier that this enzyme must overcome in catalyzing this
reaction.
8.4-9.7.7a,g,i Formation of the enol in these systems places a
double bond exocyclic to a cyclopentadiene ring, eq 19, which
creates fulvenoid resonance.
The free energy diagram given in Figure 7 presents our results
in graphical form. The scale on the left side of this diagram
The acidity constant for mandelic acid enol determined here,
pQEa ) 6.39 makes this substance considerably more acidic
than enols of simple aldehydes and ketones, e.g., pQEa ) 10.50
(50) Chiang, Y.; Kresge, A. J.; Tang,Y. S.; Wirz, J. J. Am. Chem. Soc.
1984, 106, 460-462.
(51) Jefferson, E. A.; Keeffe, J. R.; Kresge, A. J. J. Chem. Soc., Perkin
Trans. 2 1995, 2041-2046.
(48) Banks, W. H.; Davies, C. W. J. Chem. Soc. 1938, 73-78.
(49) Chiang, Y.; Kresge, A. J.; Schepp, N. P. J. Am. Chem. Soc. 1988,
111, 3977-3980.
(52) Kresge, A. J.; Yin, Y. Unpublished work.
(53) Stewart, R. The Proton: Applications to Organic Chemistry;
Academic Press: New York, 1985; p 44.

10212 J. Am. Chem. Soc., Vol. 119, No. 42, 1997
Chiang et al.
shows that mandelic acid enol and enolate ion lie 22.1 and 30.8
kcal mol-1, respectively, above un-ionized mandelic acid. These
values refer to the commonly employed standard state [H+] )
1.0 M. The enzymatic reaction, however, takes place in neutral
solution, and switching over to the standard state [H+] ) 10-7
M lowers the free energies of the ionized species by RT ln(107)
) 9.5 kcal mol-1. This is shown by the scale on the right side
of Figure 7, which also reveals that in neutral solution mandelic
acid enol and enolate ion have closely similar free energies and
therefore differ little in stability; either one of these species could
consequently be involved in the enzymatic reaction with equal
facility. These species lie an average of 26.8 kcal mol-1 above
the mandelate ion, which is the form in which the substrate
will exist in the neutral solution of the enzymatic reaction. The
thermodynamic barrier that the enzyme must overcome in
forming the enol or enolate ion intermediates involved in its
enzymatic reaction some 18 kcal mol-1 below that of the enol
or enolate ion. The enzyme must therefore stabilize its reaction
intermediate by this amount if its transition state lies at the same
free energy as the enol or enolate ion intermediate and by
somewhat more if, as seems likely, the intermediate lies below
the transition state. Two different hypotheses have recently been
advanced to account for this stabilization: one attributes it to
formation of a strong “low-barrier” hydrogen bond between the
intermediate and an essential glutamic acid residue at the active
site of the enzyme11a,55 and another attributes it to electrostatic
stabilization of the intermediate by the divalent metal ion
cofactor needed by the enzymatic reaction.11b Currently avail-
able evidence would seem to favor the latter explanation.56
Acknowledgment. We are grateful to the Natural Sciences
and Engineering Research Council of Canada and the United
States National Institutes of Health for financial support of this
work.
reaction is thus some 27 kcal mol-1
This thermodynamic barrier is much greater than the free
.
Supporting Information Available: Tables S1-S11 of rate
data (24 pages). See any current masthead page for ordering
and Internet access instructions.
energy of activation of the enzymatic reaction. Using kcat
)
1070 s-1 and Km ) 0.63 mM54 gives a rate constant, referenced
to free enzyme and mandelate ion as the initial state, that
corresponds to the free energy of activation ∆Gq ) 9.0 kcal
mol-1. This puts the free energy of the transition state of the
JA971774R
(55) Mitra, B.; Kallarakal, A. T.; Kozarich, J. W.; Gerlt, J. A.; Clifton,
J. G.; Petsko, G. A.; Kenyon, G. L. Biochemistry 1995, 34, 2777-2787.
(56) Kresge, A. J. CHEMTRACTS Org. Chem. 1996, 9, 201-203.
(54) Kenyon, G. L.; Hegeman, G. D. AdV. Enzymol. 1979, 50, 325-
360.