Carbocations in the β-lactam and β-thiolactam series

Article abstract of DOI:10.1021/ja9624329

Solvolyses of mesylate and trifluoroacetate derivatives of 3-aryl-3-hydroxy-substituted β-thiolactams and β-lactams proceeds to give substitution products in a process which exhibits a large solvent effect as well as large substituent effects. A common io

Full text of DOI:10.1021/ja9624329

J. Am. Chem. Soc. 1996, 118, 12331-12338
12331
Carbocations in the â-Lactam and â-Thiolactam Series
Xavier Creary,* Chen Zhu, and Ziqi Jiang
Contribution from the Department of Chemistry and Biochemistry, UniVersity of Notre Dame,
Notre Dame, Indiana, 46556
ReceiVed July 15, 1996X
Abstract: Solvolyses of mesylate and trifluoroacetate derivatives of 3-aryl-3-hydroxy-substituted â-thiolactams and
â-lactams proceeds to give substitution products in a process which exhibits a large solvent effect as well as large
substituent effects. A common ion rate suppression as well as largely racemic products from solvolyses of an optically
active substrate all point to the involvement of â-lactam derived carbocationic intermediates. Unlike acyclic analogs,
these cations are captured by solvent to give simple substitution products with no competing proton loss. Computational
studies suggest that proton loss from these â-lactam derived cationic intermediates is an unfavorable process due to
antiaromatic character in the potential elimination product. Computational studies also suggest that CdS and CdO
conjugative stabilization of these cations is minimal or non-existent. Substituent effect studies show that the major
mode by which these cations derive stabilization is by aryl group charge delocalization. Azide ion in DMSO or
DMF reacts with N-methyl-3-mesyloxy-3-phenylazetidine-2-thione via a bimolecular substitution mechanism to give
the corresponding 3-azido substitution product, which can be converted to N-methyl-3-amino-3-phenylazetidine-2-
thione.
Solvolytic reactions that proceed via R-carbonyl and R-thio-
carbonyl cations have continued to hold our interest.1 This
interest grows out of the high reactivity of substrates such as
trifluoroacetates 1 and 3 relative to the R-H analog 5. Our
previous studies2 have shown that solvolysis reactions of 1 and
3 in acetic acid give exclusively elimination products. These
reactions proceed via the electronegatively substituted cations
2 and 4 at rates that exceed that of the trifluoroacetate 5 despite
the electron-withdrawing amide and thioamide groups attached
directly to the cationic centers of the intermediates 2 and 4.
Conjugation of the developing cationic center with the CdX
bond has been proposed to lead to the relatively high reactivity
of substrates such as 1 and 3.
developing methods for conversion of the hydroxyl groups of
7 and 8 to other functionality. Incorporation of a leaving group
into the 3-position of these substrates should potentially permit
the generation of cationic intermediates and hence the introduc-
tion of other functionality. These systems should also allow
one to compare the behavior R-carbonyl and R-thiocarbonyl
cations in â-lactam systems. Reported here are the results of
these studies.
Results and Discussion
(a) Solvent Effect Studies. The alcohols 7 and 8 were
converted to mesylate derivatives or to other derivatives with
appropriate leaving group reactivities. In the case of the
p-methoxyphenyl derivative, the trifluoroacetate or chloro
leaving groups have convenient reactivity, while in the case of
the 3,5-bis(trifluoromethyl)phenyl derivative, the triflate leaving
group was appropriate. The appropriate derivatives 9 and 10
were reacted in a variety of solvents and the simple substitution
products were formed exclusively. Rate data are given in Table
1. Solvent effect studies on 9a (Ar ) Ph) and 10a (Ar ) Ph)
reveal large increases in rates with increasing solvent ionizing
power. Rates increase by factors of 1.3 × 104 and 4.9 × 103
respectively as solvent is changed from the relatively poorly
ionizing solvent ethanol to the highly ionizing hexafluoroiso-
propyl alcohol solvent.4 Winstein-Grunwald m values5 are 0.89
Recently we have developed a novel and simple synthesis of
â-thiolactams 7 and â-lactams 8 substituted in the 3-position
with the hydroxyl group.3 We were therefore interested in
X Abstract published in AdVance ACS Abstracts, November 15, 1996.
(1) Creary, X. Chem. ReV. 1991, 91, 1625.
(2) (a) Creary, X.; Hatoum, H. N.; Barton, A.; Aldridge, T. E. J. Org.
Chem. 1992, 57, 1887. (b) Creary, X.; Aldridge, T. J. Org. Chem. 1988,
53, 3888.
(4) Bentley, T. W.; Bowen, C. T.; Morten, D. H.; Schleyer, P. v. R. J.
Am. Chem. Soc. 1981, 103, 5466 and references therein.
(3) Creary, X.; Zhu, C. J. Am. Chem. Soc. 1996, 117, 5859.
S0002-7863(96)02432-8 CCC: $12.00 © 1996 American Chemical Society

12332 J. Am. Chem. Soc., Vol. 118, No. 49, 1996
Creary et al.
Table 1. Solvolysis Rates of Substrates in Various Solvents at 25
°C
a
substrate
9a (p-H)
solvent
k (s-1
)
kCdS/kCdO
CH3CH2OH
HOAc
CH3OH
CF3CH2OH
HCO2H
(CF3)2CHOH
HOAc
1.49 × 10-5
2.57 × 10-5
8.55 × 10-5
4.84 × 10-3
3.82 × 10-2
1.95 × 10-1
1.88 × 10-3
8.17 × 10-7 b
1.18 × 10-4
7.37 × 10-5
4.48 × 10-5
3.91 × 10-5
2.11 × 10-2
1.23 × 10-3
8.52 × 10-2
3.36 × 10-2
8.91 × 10-1
2.43 × 10-6
2.55 × 10-6
1.51 × 10-5
2.44 × 10-4
5.86 × 10-3
1.19 × 10-2
9.68 × 10-2
3.04 × 10-8 b
3.15 × 10-5
5.23 × 10-3
1.77 × 10-4
2.10 × 10-2
1.03 × 10-3
3.26 × 10-5
10.1
9b (p-CH3)
9c (p-CF3)
9d (p-OCH3)
6.47
26.8
3.76
CF3CH2OH
HOAc
HOAcc
HOAcd
HOAce
9f (3,5-(CF3)2)
10a (p-H)
CH3CH2OH
HOAc
CH3OH
CF3CH2OH
HCO2H
CH3CH2OH
HOAc
CH3OH
CF3CH2OH
HCO2H
(CF3)2CHOH
CF3CH2OH
CF3CH2OH
HOAc
CH3CH2OH
HOAc
6.94
Figure 1. A plot of log k for solvolyses of mesylates 9 and 10 in
trifluoroethanol at 25.0 °C vs σ+.
10b (p-CH3)
10c (p-CF3)
10d (p-OCH3)
10f (3,5-(CF3)2)
The effect of solvent on the rate behavior of triflates 9f and
10f contrasts with that of mesylates 9a and 10a. While
solvolyses of these triflates gives simple substitution products
in all solvents, rate data are not in line with a limiting kC process
in all solvents. Solvolysis rates are actually faster in methanol
than in trifluoroethanol despite the fact that trifluoroethanol is
much more highly ionizing. This points to the importance of
solvent nucleophilicity in solvolyses of 9f and 10f. At the same
time, solvent ionizing power is also of importance as evidenced
by the HCO2H/CH3CO2H rate ratio. The rate of 9f is 726 times
faster in the more highly ionizing solvent formic acid than in
acetic acid, despite the fact that these solvents have comparable
nucleophilicities. This behavior is reminiscent of the behavior
of 2-propyl tosylate, where both solvent nucleophilicity and
solvent ionizing power are important in determining rate.6 The
behavior of 9f and 10f can therefore best be characterized as
“borderline”,7 i.e., there is evidence for significant solvent
involvement in a transition state with developing positive charge
at the benzylic carbon atom.
CH3OH
CF3CH2OH
HOAc
1-phenylcyclobutyl
trifluoroacetate
a Maximum standard deviations in duplicate runs were (1.5%. See
the Experimental Section for the kinetic method. b Extrapolated from
data at higher temperature. c 0.025 M CF3CO2Na added. d 0.050 M
CF3CO2Na added. e 0.075 M CF3CO2Na added.
and 0.83 respectively for 9a (Ar ) Ph) and 10a (Ar ) Ph).
(b) Substituent Effect Studies. A substituent effect study
on 9 and 10 shows large increases in rate as the aromatic ring
is substituted with increasing electron-donating groups. Me-
sylate derivatives containing the p-anisyl group were too reactive
to allow preparation. Therefore rates of the p-methoxy substi-
tuted mesylates were estimated from those of the corresponding
trifluoroacetates 9d and 10d. The Hammett-Brown F+ values8
in trifluoroethanol (Figure 1) for 9 and 10 are -6.6 and -7.3,
respectively. These F+ values are consistent with cationic
intermediates 11 and 12 that have a large demand for stabiliza-
tion by the aryl substituent. The slightly larger F+ value for
â-lactams 10 is indicative of a slightly larger demand for aryl
group stabilization in the developing cationic intermediates 12.
These data are consistent with the involvement of cationic
intermediates 11 and 12 under solvolytic conditions.
(6) (a) Bentley, T. W.; Schleyer, P. v. R. J. Am. Chem. Soc. 1976, 98,
7658. (b) Schadt, F. L.; Bentley, T. W.; Schleyer, P. v. R. J. Am. Chem.
Soc. 1976, 98, 7667.
(7) The detailed mechanism for reaction of so called “borderline”
substrates has been a controversial area. See: (a) Allen, A. D.; Kanagasa-
bapathy, V. M.; Tidwell, T. T. J. Am. Chem. Soc. 1985, 107, 4513. (b)
Bentley, T. W.; Schleyer, P. v. R. AdV. Phys. Org. Chem. 1977, 14, 1. (c)
Richard, J. P.; Jencks, W. P. J. Am. Chem. Soc. 1984, 106, 1383. (d) Jencks,
W. P. Acc. Chem. Res. 1980, 13, 161. (e) Sneen, R. A. Acc. Chem. Res.
1973, 6, 46.
(5) Y values based on t-BuCl were used in this correlation. See: (a)
Grunwald, E.; Winstein, S. J. Am. Chem. Soc. 1948, 70, 846. (b) Winstein,
S.; Grunwald, E.; Jones, H. W. J. Am. Chem. Soc. 1951, 73, 2700.
(8) Brown, H. C.; Okamoto, Y. J. Am. Chem. Soc. 1958, 80, 4979.

Carbocations in the â-Lactam and â-Thiolactam Series
J. Am. Chem. Soc., Vol. 118, No. 49, 1996 12333
The relatively large F+ value for the cyclic system 9 contrasts
with the behavior of the acyclic system 3 (and substituted phenyl
analogs). These acyclic thioamide systems give a nonlinear
Hammett plot as well as relatively small substituent effects. For
example the rate of the p-methoxy analog of 3 is only 118 times
faster than that of the unsubstituted system. By way of contrast,
the estimated rate increase due to p-methoxy substitution in 9
is a factor of 2.34 × 105. The reasons for small aryl group
substituent effects in the acyclic systems have been discussed
and are due in part to extensive transition state charge delocal-
ization utilizing the thiocarbonyl group. This decreases the
demand for aryl group stabilization of the acyclic carbocationic
intermediates.
Table 2. Azide Products 16 in Solvolyses of Substrates 9 and 10
in 0.2 M NaN3 in MeOH
substrate
% azide
kazide/kMeOH (M-1
)
9a (p-H)
15
12
55a
80
32
39
72
0.86
0.69
6.12
20.0
2.35
3.23
12.7
9b (p-CH3)
9c (p-CF3)
9e (p-OCH3)
10a (p-H)
10b (p-CH3)
10e (p-OCH3)
a After 37% reaction.
too strained to be formed via k∆ processes (which could also
(c) Chiral Systems. The optically active mesylate (+)-9a
can be prepared from the corresponding optically active alcohol.9
Solvolysis in 80% aqueous acetone leads to an alcohol product
7 (Ar ) Ph) which is 98% racemized but also has a 2% ee of
the inVerted alcohol. A control experiment showed that
racemization of optically active alcohol does not occur under
the solvolysis conditions. This type of solvolysis which pro-
ceeds to form a slight excess of the inverted product is often
observed.10 It has been rationalized in terms of a cationic inter-
mediate which has a slight preference for capture of solvent
from the rear of the departing leaving group due to a slightly
greater shielding of one side of the ion-pair intermediate by
the departed mesylate ion. By way of contrast, solvolysis in
acetic acid leads to an acetate product which has a 3% ee of
the retained acetate (which is optically stable under the reaction
conditions). This study is also consistent with the involvement
of a cationic intermediate. The 3% net retention is believed to
be a result of acetate being delivered preferentially to the same
side of the cationic intermediate from which the leaving group
departed due to hydrogen bonding of acetic acid with the
departing mesylate leaving group (as in 13). Such retentive
processes have been observed in the past.11 Cyclized intermedi-
ates such as 14 and 15 are probably not involved since they are
lead in principle to net retention).
(d) Common Ion Effect. When 0.04 M trifluoroacetate (9d,
Ar ) 4-MeOC6H4) is solvolyzed in acetic acid containing 0.05
M sodium acetate, the slope of a first order plot decreases with
time. Closer examination shows that 9d (Ar ) 4-MeOC6H4)
in acetic acid is subject to a common ion rate suppression. As
shown in Table 1, addition of external trifluoroacetate ion slows
the rate appreciably. The observation of a common ion effect
is classic evidence for the involvement of a relatively stable
cationic intermediate 11 (Ar ) 4-MeOC6H4).12 This cationic
intermediate survives past the ion-pair stage to the free ion stage,
where it can react with solvent or return to starting trifluoro-
acetate by reaction with externally added trifluoroacetate ion.
A quantitative treatment12b of rate data for 9d in acetic acid
using a kinetic model involving competitive solvent capture and
trifluoroacetate ion capture of the cationic intermediate gives
ktrifluoroacetate/(kHOAc[HOAc]) ) 28 M-1
.
(e) Azide Ion Studies. Azide ion is recognized as an
effective carbocation trap. Low concentrations of azide ion can
often compete with solvent (HOS) to form the corresponding
azido derivatives in solvolysis reactions. Since the rate of
reaction of azide ion with cations is extremely rapid (close to
diffusion controlled),13 values of kazide/kHOS also allow an
estimation of carbocation lifetimes in solution.14 Additionally,
many â-lactams of therapeutic value have a nitrogen atom in
the 3-position.15 Methanolysis reactions of 9 and 10 in the
presence of azide ion16 were therefore carried out in an attempt
to form the 3-azido derivatives 16. Values of kazide/kMeOH were
determined from the relative amounts of 16 and the methanol
displacement product 17 and are listed in Table 2. Reaction of
(11) (a) Okamoto, K.; Kinoshita, T.; Oshida, T.; Yamamoto, T.; Ito, Y.;
Dohi, M. J. Chem. Soc., Perkin Trans. 2 1976, 1617. (b) Kinoshita, T.;
Komatsu, K.; Ikai, K.; Kashimura, T.; Tanikawa, S.; Hatanaka, A.; Okamoto,
K. J. Chem. Soc., Perkin Trans. 2 1988, 1875. (c) Goering, H. L.; Hopf, H.
J. Am. Cherm. Soc. 1971, 93, 1224. (d) Bone, J. A.; Pritt, J. R.; Whiting,
M. C. J. Chem. Soc., Perkin Trans. 2 1975, 1447. For further leading
references see: Okamoto, K. In AdVances in Carbocation Chemistry; Creary,
X., Ed.; JAI Press Inc.: Greenwich, CT 1989; Vol. 1, p 171.
(12) (a) March, J. AdVanced Organic Chemistry, 3rd ed.; John Wiley
and Sons, Inc.: New York, 1985; p 261. (b) Hine, J. Physical Organic
Chemistry, 2nd ed.; McGraw-Hill Book Co., Inc.: New York, 1962; pp
131-5.
(9) This alcohol was prepared by using the chiral base i to promote the
cyclization of PhCOCSNMe2. Full details will be reported elsewhere.
(13) (a) McClelland, R. A.; Banait, N.; Steenken, S. J. Am. Chem. Soc.
1986, 108, 7023. (b) Richard, J. P.; Jencks, W. P. J. Am. Chem. Soc. 1984,
106, 1361.
(14) (a) Richard, J. P.; Jencks, W. P. J. Am. Chem. Soc. 1982, 104, 4689.
(b) Richard, J. P.; Jencks, W. P. J. Am. Chem. Soc. 1984, 106, 1383. See
also: (c) Richard, J. P. In AdVances in Carbocation Chemistry; Creary, X.,
Ed.; JAI Press Inc.: Greenwich, CT, 1989; Vol. 1, p 121.
(15) For leading references see: Guzzo, P. R.; Miller, M. J. In AdVances
in Nitrogen Heterocycles; Moody, C. J., Ed.; JAI Press Inc.: Greenwich,
CT, 1996; Vol. 2, p 1.
(16) Control experiments show that the rate of methanolysis of 9b (with
ionic strength maintained at 0.1 M with sodium triflate) is independent of
the amount of added sodium azide.
(10) For typical examples for our laboratory, see: (a) Creary, X.; Geiger,
C. C. J. Am. Chem. Soc. 1982, 104, 4151. (b) Creary, X.; Geiger, C. C.;
Hilton, K. J. Am. Chem. Soc. 1983, 105, 2851. (c) Creary, X.; McDonald,
S. R.; Eggers, M. Tetrahedron Lett. 1985, 26, 811. For earlier examples,
see: (d) Doering, W. v. E.; Zeiss, H. H. J. Am. Chem. Soc. 1953, 75, 4733.
(e) Winstein, S.; Morse, B. K. J. Am. Chem. Soc. 1952, 74, 1133. (f)
Steigman, J.; Hammett, L. P. J. Am. Chem. Soc. 1937, 59, 2536.

12334 J. Am. Chem. Soc., Vol. 118, No. 49, 1996
Creary et al.
the chloride 9e (Ar ) 4-MeOC6H4) in methanol containing 0.2
M NaN3 gave 80% of the azido derivative 16 (Ar ) 4-MeOC6H4;
X ) S), along with 20% of the solvent substitution product.
This corresponds to a kazide/kMeOH ratio of 20 M-1, i.e., azide
ion is a relatively effective trap of the cationic intermediate.
However, in the case of 9a (Ar ) Ph; X ) S) and 9b (Ar )
4-CH3C6H4; X ) S) the azide products are only 15% and 12%,
respectively, of the total with the major product being the
methanol capture product. The kazide/kMeOH ratios of 0.86 and
0.69 M-1 show that azide ion is relatively ineffective at capture
of these two cationic intermediates.
analogs 21 which gave exclusive elimination products 22.2 The
formation of elimination product from R-thiocarbonyl cations
is the usual fate of such cations which contain â-hydrogen
atoms. Although R-carbonyl cations with â-hydrogens can give
significant amounts of substitution products, the R-carbonyl
cation 2 also gives exclusively the elimination product in acetic
acid.17 We were therefore interested in the contrasting products
derived from 9 and 21 (as well as their carbonyl analogs).
Ab initio computational studies18 have been used to gain
insight into the lack of elimination products from 9 and 10.
The reactions of cations 23 and 24 (where the vinyl group has
been used as a phenyl group analog) with water has been
analyzed at the HF/6-31G** level. In both cases, reaction with
water as a nucleophile to give 25 is an exothermic process with
∆E being -11.8 and -10.9 kcal/mol, respectively.
The amount of azide product increases again (to 55%) in the
case of the more electron-withdrawing p-CF3-substituted system
10d. This type of behavior, where selectivity for azide ion
decreases, passes through a minimum, and then increases as
the substituent becomes more electron-withdrawing, has been
observed before.14 It has been interpreted by Jencks and Richard
as being due to a mechanistic change to the so called “enforced
SN2” mechanism.7d,14 The transition state for this process has
positive charge development on the benzylic carbon, but there
is no discrete cationic intermediate with a finite lifetime.
In view of the relatively low yields of azide adducts 16
produced in solvolysis reactions containing added sodium azide,
attention was next turned to reaction of mesylates 9 and 10 with
azide ion in polar aprotic solvents. The mesylate 9a reacts with
sodium azide in DMF at 70 °C to give a good yield of the azide
18. This process proceeds with clean inversion, as revealed by
examination of the optically active mesylate (+)-9a. Subsequent
reaction of (-)-18 with triphenylphosphine and hydrolysis of
the intermediate iminophosphorane led to the optically active
amine (-)-19.
However, reaction with water as a base leading to the
elimination product 26 is a highly endothermic process com-
putationally. This study suggests that capture of cations 11 and
12 with nucleophilic solvents should be much more preferable
than proton loss to form alkene products.
(17) Creary, X.; Casingal, V. P.; Leahy, C. E. J. Am. Chem. Soc. 1993,
115, 1734.
Reaction of the carbonyl analog 10a with sodium azide in
DMSO at 45 °C also proceeds in good yield to give the azido-
substituted product in a second order process (first order in azide
ion). These processes in DMF and DMSO proceed, in all
likelihood, via an SN2-type mechanism.
(18) Frisch, M, J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.;
Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; Keith, T.; Petersson, G.
A.; Montgomery, J. A.; Raghavachari, K.; Al-Laham, M. A.; Zakrzewski,
V. G.; Ortiz, J. V.; Foresman, J. B.; Peng,C. Y.; Ayala, P. Y.; Chen, W.;
Wong, M. W.; Andres, J. L.; Replogle, E. S.; Gomperts, R.; Martin, R. L.;
Fox, D. J.; Binkley, J. S.; Defrees, D. J.; Baker, J.; Stewart, J. P.; Head-
Gordon, M.; Gonzalez, C.; Pople, J. A. Gaussian 94, Revision B.3; Gaussian,
Inc.: Pittsburgh, PA, 1995.
(f) Substitution vs Elimination. Computational Studies.
The solvolytic behavior of 9 contrasts with that of the acyclic

Carbocations in the â-Lactam and â-Thiolactam Series
J. Am. Chem. Soc., Vol. 118, No. 49, 1996 12335
(g) Rate Comparisons. An analysis of rates of solvolyses
of 9 and 10 is quite informative. Previous studies have shown
that R-thiocarbonyl derivatives are much more reactive than
R-carbonyl analogs under solvolytic conditions.2 For example,
the thioamide 3 is 1.9 × 105 times more reactive than the amide
1. It is even more reactive than the tertiary analog, cumyl
trifluoroacetate, 33. By way of contrast, this present study
shows that the â-thiolactam 9a is only 10 times more reactive
than the â-lactam 10a. In fact, the entire series of â-thiolactams
9 are only slightly more reactive than the â-lactams 10 (Table
1). The mesylate 9a is significantly less reactive than the tertiary
analog 34 (estimated from the trifluoroacetate rate). The
mesylate 9a is even 1.7 × 103 times less reactiVe than the
acyclic trifluoroacetate 3 in acetic acid. This result is even more
striking in view of the fact that mesylates can be up to 105 times
more reactive than the analgous trifluoroacetate derivative.19
These data point to cationic intermediates 11 and 12 which are
not significantly stabilized by thiocarbonyl or carbonyl conjuga-
tion.
These computational results contrast with those on the next
higher homologs 27 and 28. Proton loss from 27 to water to
give the alkene 28 is endothermic (+20.2 kcal/mol), but by a
much smaller factor than for formation of 26 (+33.6 kcal/mol).
Additionally, the alkene 28 exhibits a shortened C-N bond
(1.332 Å) relative to 26, indicative of a substantial amide
resonance interaction in 28. These results suggest that, relative
to 28, the alkenes 26 are subject to a destabilizing feature.
Insight into the highly endothermic reaction which forms 26
can also be gained by an additional computational study (HF/
6-31G**). The endocyclic-exocyclic double bond preference
in 29 and 30 has been evaluated and the endocyclic double bond
in 27 is quite unfavorable (∆E ) -15.8 and -11.4 kcal/mol)
relative to the exocyclic double bond of 28. These values are
significant but smaller than that calculated for the methylcy-
clobutadiene-methylenecyclobutene systems 29 and 30. The
amides 29 (as well as amides 26) are not planar molecules, i.e.,
the nitrogen atom is slightly removed from the plane defined
by the three ring carbons. Additionally, the nitrogen is highly
pyramidalized and the N4-C1 bonds of 29 are unusually long
(1.428 and 1.444 Å, respectively) for thioamides and amides.
These features all point to a lack of π-interaction of the nitrogen
of 29 with the adjacent CdO or CdS bond. In valence bond
terms, stabilization of 29 by amide resonance as in 29a is
minimal due to the antiaromatic character in forms represented
by 29a. This accounts for the highly endothermic deprotonation
reactions of 23 and 24 with water. In terms of solvolysis
reactions of 9 and 10, cationic intermediates such as 11 and 12
do not undergo proton loss since the potential alkene products
have low stability.
The relatively large F+ value for the cyclic system 9 contrasts
with the behavior of the acyclic system 3 (and substituted phenyl
(19) (a) Noyce, D. S.; Virgilio, J. A. J. Org. Chem. 1972, 37, 2643. For
further examples of this large mesylate/trifluoroacetate rate ratio, see also:
(b) Creary, X. J. Org. Chem. 1979, 44, 3938. (c) Creary, X.; Jiang, Z. J.
Org. Chem. 1996, 61, 3482.

12336 J. Am. Chem. Soc., Vol. 118, No. 49, 1996
Creary et al.
analogs). These acyclic thioamide systems give a non-linear
Hammett plot as well as relatively small substituent effects.2a
For example, the rate of the p-methoxy analog of 3 is only 118
times faster than that of the unsubstituted system. By way of
contrast, the estimated rate increase due to p-methoxy substitu-
tion in 9 is a factor of 2.34 × 105. The reasons for small aryl
group substituent effects in the acyclic systems have been
discussed2a and are due in part to extensive transition state
charge delocalization utilizing the thiocarbonyl group. This
decreases the demand for aryl group stabilization of the acyclic
carbocationic intermediates. The much larger substituent effects
in 9 and 10 are indicative of cationic intermediates 11 and 12
devoid of thiocarbonyl or carbonyl conjugation.
in previous studies on R-thiocarbonyl cations. Carbonyl
stabilization of 24 is non-existent. On the other hand, allyl
stabilization of carbonyl cation 24 is slightly larger than allyl
stabilization of thiocarbonyl cation 23. In other words, the sum
of allyl stabilization and the small amount of thiocarbonyl
stabilization in 23 is quite comparable to the allyl stabilization
in 24. Cations 23 and 24 therefore have comparable stabilities.
(h) Computational Studies on Cationic Intermediates. In
order to shed further light on the cationic intermediates 11 and
12, the hydride transfer reaction of 24 with 35 has been
evaluated computationally. This isodesmic reaction suggests
that the R-thiocarbonyl cation 23 and the R-carbonyl cation 24
are comparable in stability. Previous computational studies have
all suggested that R-thiocarbonyl cations are substantially
stabilized relative to R-carbonyl analogs.20 This has been
attributed to extensive charge delocalization onto the sulfur atom
of R-thiocarbonyl cations. However, the computational study
on 23 and 24 is completely in line with our solvolytic rate study,
which suggests that 11 and 12 are also comparable in stability.
These suggestions on the basis of computational studies are
in line with the observed Hammett F+ values. The slightly
larger F+ value observed in solvolyses of the â-lactam systems
10 are in line with a slightly larger demand for stabilization of
the cationic intermediate by the aryl group. Specifically, the
â-thiolactam 9d is 3.8 times more reactive than the carbonyl
analog 10d. This reactivity difference increases as the sub-
stituent becomes more electron-withdrawing and the CdS/CdO
rate ratio is 26.8 for the p-CF3 substituted systems 9c and 10c.
This is a reflection of increased charge delocalization onto the
aryl group of the carbonyl cations 12 (relative to the thiocarbonyl
cations 11). The same phenomenon is observed computationally
in the allyl analog 24, where conjugation with the allyl group
is slightly greater than in 23.
Closer examination of bond lengths gives further insight. The
CdS bond length is actually shorter in cation 23 (1.612 Å) than
in the neutral molecule 35 (1.640 Å). The C-C bond length
in 23 from the cationic center to the adjacent thiocarbonyl carbon
atom (1.515 Å) is only slightly shorter than the analogous C-C
bond in the neutral 35 (1.532 Å). These features point to only
minimal stabilization of 23 due to thiocarbonyl conjugation. As
suggested by bond lengths, the major stabilizing feature of cation
23 is allylic conjugation. Analogous bond length data for the
R-carbonyl cation 24 and the neutral 36 suggests that cation 24
receives no stabilization from carbonyl conjugation. On the
other hand, cation 24 is also extensively stabilized by allyl group
conjugation and the extent of allyl stabilization is slightly greater
than stabilization of the thiocarbonyl cation 23. The comparable
stabilities of 23 and 24 can therefore be attributed to offsetting
factors. Thiocarbonyl conjugative stabilization of 23 is quite
small and does not approach the magnitude that we have seen
Of related interest is a computational study on cations 38
and 39, where R-carbonyl and R-thiocarbonyl conjugative
stabilization is precluded by geometric constraints. As implied
by the isodesmic reaction below, these cations also possess
comparable stabilities. This study verifies computationally the
suggestion that R-carbonyl and R-thiocarbonyl cations will have
comparable stabilities in the absence of conjugative effects.
Takeuchi has recently verified this suggestion experimentally
(20) (a) Lien, M. H.; Hopkinson, A. C. J. Am. Chem. Soc. 1988, 110,
3788. (b) Creary, X.; Wang, Y.-X.; Jiang, Z. J. Am. Chem. Soc. 1995, 117,
3044.

Carbocations in the â-Lactam and â-Thiolactam Series
J. Am. Chem. Soc., Vol. 118, No. 49, 1996 12337
Preparation of Trifluoroacetates 9d and 10d. A solution of 114
mg of 7 (Ar ) 4-CH3OC6H4) (0.511 mmol) and 83 mg of 2,6-lutidine
(0.776 mmol) in 1.0 mL of CH2Cl2 was cooled to 0 °C and 141 mg of
trifluoroacetic anhydride (0.671 mmol) in 0.5 mL of CH2Cl2 was added
dropwise. The stirred mixture was warmed to room temperature and
after 10 min 5 mL of ether was added. The mixture was transferred to
a separatory funnel, washed rapidly in succession with cold water, cold
dilute HCl solution, saturated NaCl solution, and then dried over
MgSO4. Solvent removal using a rotary evaporator left 160 mg (98%
yield) of the trifluoroacetate 9d, mp 84-85 °C. 1H NMR (CDCl3) δ
7.690 and 6.945 (AA′BB′ quartet, 4 H), 4.620 (d, J ) 8.1 Hz, 1 H),
4.415 (d, J ) 8.1 Hz, 1 H), 3.821 (s, 3 H), 3.208 (s, 3 H). 13C NMR
(CDCl3) δ 197.15, 160.84, 156.99 (q, J ) 65 Hz), 128.84, 124.55,
114.21, 114.08 (q, J ) 284 Hz), 86.60, 60.73, 55.38, 31.76. Anal.
Calcd for C13H12F3NO3S: C, 48.90; H, 3.79. Found: C, 48.79; H,
3.79.
in a study of triflates where conjugation is systematically varied
by geometric constraints.21
The preparations of trifluoroacetate 10d and 1-phenylcyclobutyl
trifluoroacetate were completely analogous to the above procedure.
Preparation of Triflates 9f and 10f. General Procedure.
A
Conclusions. Mesylate and trifluoroacetate derivatives of
3-aryl-3-hydroxy-substituted â-thiolactams and â-lactams (9 and
10) can undergo solvolytic reactions via carbocationic inter-
mediates 11 and 12. Solvent effect studies, substituent effect
studies, studies on chiral systems, and the observation of a
common ion rate suppression all point to the involvement of
carbocationic intermediates. In contrast to open chain analogs,
these cationic intermediates do not suffer proton loss, but instead
are captured by solvent to give simple substitution products.
Computational studies suggest that proton loss from cationic
intermediates 11 and 12 will be an unfavorable process due to
antiaromatic character in the potential elimination product. In
contrast to acyclic analogs, rate studies as well as computational
studies suggest that cations such as 11 and 12 have comparable
stabilities. CdS and CdO conjugative stabilization of these
cations is therefore minimal or non-existent. As revealed by
substituent effect studies, the major mode by which cations 11
and 12 derive stabilization is by aryl group charge delocalization.
Finally, sodium azide in polar aprotic solvents reacts with
mesylates 9 and 10 via a bimolecular substitution mechanism
to give the corresponding azido-substitution product.
solution of alcohol 7 or 8 (Ar ) 3,5-(CF3)2C6H3) (1 equiv) and 2,6-
lutidine (2.0 equiv) in CH2Cl2 was cooled to -50 °C and triflic
anhydride (1.5 equiv) was added dropwise. The mixture was warmed
to 0 °C and ether was added. The mixture was then rapidly washed
with cold dilute HCl solution, NaHCO3 solution, and saturated NaCl
solution. The solution was dried over MgSO4 and the solvent was
removed using a rotary evaporator. The residue contained the triflate
product as well as a significant amount of the adduct derived from
reaction of the triflate with 2,6-lutidine. This byproduct was removed
by dissolving the residue in pentane, and washing with dilute HCl
solution and saturated NaCl solution, and drying over MgSO4. Removal
of the pentane solvent using a rotary evaporator left the triflates 9f or
10f as unstable oils. These triflates are best stored in the freezer in
ether solution. The following is representative.
Reaction of 150 mg of 8 (Ar ) 3,5-(CF3)2C6H3) and 73 mg of 2,6-
lutidine in 4 mL of CH2Cl2 with 142 mg of triflic anhydride gave,
after a workup as described above, 99 mg (66%) of triflate 10f. 1H
NMR (CDCl3) δ 8.148 (br s, 2 H), 8.020 (br s, 1 H), 4.174 (d, J ) 7.3
Hz, 1 H), 4.137 (d, J ) 7.3 Hz, 1 H), 3.029 (s, 3 H). 13C NMR (CDCl3)
δ 160.54, 134.04, 132.79 (q, J ) 34 Hz), 128.54 (q, J ) 3 Hz), 124.89
(heptet, J ) 4 Hz), 122.68 (q, J ) 273 Hz), 117.88 (q, J ) 320 Hz),
93.68, 54.68, 29.23.
Preparation of Chlorides 9e and 10e. A suspension of 130 mg of
7 (Ar ) 4-MeOC6H4) and 330 mg of Na2CO3 in 5 mL of ether was
stirred at room temperature as 100 mg of SOCl2 was added. The
mixture was stirred for 40 min at room temperature and then the solution
was filtered through a small amount of Na2CO3 . The ether solvent
was removed using a rotary evaporator leaving 138 mg (98%) of
chloride 9e as an oil. 1H NMR (CDCl3) δ 7.683 and 6.907 (AA′BB′
quartet, 4 H), 4.405 (d, J ) 7.2 Hz, 1 H), 4.368 (d, J ) 7.2 Hz, 1 H),
3.811 (s, 3 H), 3.197 (s, 3 H). 13C NMR (CDCl3) δ 200.30, 160.05,
128.73, 128.39, 113.99, 69.80, 67.23, 55.34, 31.92. Chloride 9e
decomposes on standing at room temperature and is best stored as a
solution in ether. The preparation of 10e from 8 (Ar ) 4-MeOC6H4)
was completely analogous.
Experimental Section
Preparation of Mesylates 9 and 10. General Procedure.
A
solution of the appropriate alcohol (1 equiv) and CH3SO2Cl (1.5 equiv)
in methylene chloride was cooled to -50 °C and Et3N (1.8 equiv) was
added dropwise. The stirred mixture was slowly warmed to 0 °C. The
mixture was transferred to a separatory funnel with ether, washed
rapidly in succession with cold water, cold dilute HCl solution, saturated
NaCl solution, and then dried over MgSO4. Solvent removal using a
rotary evaporator left the corresponding mesylate derivative. The
following are representative.
Addition of 488 mg of Et3N to a solution of 518 mg of 7 (Ar ) Ph)
and 461 mg of CH3SO2Cl in 8 mL of CH2Cl2 gave, after workup, 666
mg (92%) of mesylate 9a, mp 116-118 °C. 1H NMR (CDCl3) δ 7.92-
7.83 (m, 2 H), 7.52-7.42 (m, 3 H), 4.689 (d, J ) 8.1 Hz, 1 H), 4.542
(d, J ) 8.1 Hz, 1 H), 3.199 (s, 3 H), 2.792 (s, 3 H). 13C NMR (CDCl3)
δ 197.14, 132.84, 130.27, 128.76, 128.26, 87.60, 62.38, 41.13, 31.79.
Anal. Calcd for C11H13NO3S2: C, 48.69; H, 4.83. Found: C, 48.50;
H, 4.87.
The mesylate (+)-9a was prepared from (+)-N-methyl-3-hydroxy-
3-phenylazetidine-2-thione (74% ee)9 using an identical procedure.
In similar fashion, mesylate 10a, mp 95-97 °C, was produced from
8 (Ar ) Ph). 1H NMR (CDCl3) δ 7.73-7.64 (m, 2 H), 7.50-7.42 (m,
3 H), 4.119 (d, J ) 6.5 Hz, 1 H), 3.928 (d, J ) 6.5 Hz, 1 H), 2.948 (s,
3 H), 2.864 (s, 3 H). 13C NMR (CDCl3) δ 163.96, 132.96, 130.18,
128.93, 127.83, 91.94, 55.73, 40.45, 28.73. Anal. Calcd for C11H13-
NO4S: C, 51.75; H, 5.13. Found: C, 51.81; H, 5.11.
Solvolyses of Mesylates 9 and 10 in Alcohol Solvents. General
Procedure. A solution of the mesylate in the appropriate alcohol
solvent containing 1.2 equiv of 2,6-lutidine was kept at room temper-
ature for 10 half-lives. The alcohol solvent was then removed using a
rotary evaporator and the residue was taken up into ether. The ether
was washed with dilute HCl solution followed by saturated NaCl
solution. The ether extract was then dried over MgSO4 and the solvent
was removed using a rotary evaporator leaving the substitution product.
The following is representative.
A solution of 147 mg of 9a in 13 mL of methanol (0.05 M 2,6-
lutidine) was kept at room temperature for 40 h. A workup as described
above gave 102 mg (91%) of N-methyl-3-methoxy-3-phenylazetidine-
2-thione, mp 74-75 °C. 1H NMR (CDCl3) δ 7.69-7.60 (m, 2 H),
7.45-7.31 (m, 3 H), 4.152 (d, J ) 6.9 Hz, 1 H), 3.970 (d, J ) 6.9 Hz,
1 H), 3.484 (s, 3 H), 3.183 (s, 3 H). 13C NMR (CDCl3) δ 202.14,
136.22, 128.71, 128.54, 127.01, 87.14, 62.84, 53.16, 31.44. Anal. Calcd
for C11H13NOS: C, 63.74; H, 6.32. Found: C, 63.22; H, 6.34.
(21) Tokunaga, K.; Ohga, Y.; Takeuchi, K. Tetrahedron Lett. 1996, 37,
2241.

12338 J. Am. Chem. Soc., Vol. 118, No. 49, 1996
Creary et al.
Solvolysis of Mesylate 9a in Acetic Acid. A solution of 169 mg
of mesylate 9a in 16 mL of acetic acid (0.05 M in NaOAc) was kept
at room temperature for 3.5 days. The mixture was then taken up into
ether and water and the acetic acid was neutralized by slow addition
of solid K2CO3. The ether extract was then dried over MgSO4 and
filtered, and the solvent was removed using a rotary evaporator, leaving
141 mg (96%) of N-methyl-3-acetoxy-3-phenylazetidine-2-thione, mp
105-106 °C. 1H NMR (CDCl3) δ 7.67-7.60 (m, 2 H), 7.44-7.31
(m, 3 H), 4.581 (d, J ) 8.1 Hz, 1 H), 4.385 (d, J ) 8.1 Hz, 1 H), 3.201
(s, 3 H), 2.153 (s, 3 H). 13C NMR (CDCl3) δ 200.04, 169.74, 134.68,
128.96, 128.54, 126.44, 84.68, 61.81, 31.69, 21.35. Anal. Calcd for
C12H13NO2S: C, 61.25; H, 5.57. Found: C, 61.27; H, 5.60.
Preparation of (+)-N-Methyl-3-acetoxy-3-phenylazetidine-2-
thione. A solution of 51 mg of (+)-N-methyl-3-hydroxy-3-phenyl-
azetidine-2-thione, [R]D ) + 131°, and 68 mg of acetic anhydride in
2 mL of pyridine was stirred as 5 mg of 4-dimethylaminopyridine was
added. After 18 h at room temperature, the mixture was taken up into
ether and the ether was washed with water, dilute HCl solution, and
saturated NaCl solution and dried over MgSO4. Solvent removal using
a rotary evaporator gave 58 mg (94%) of (+)-N-methyl-3-acetoxy-3-
phenylazetidine-2-thione, [R]D ) + 101.6° (acetone).
Solvolysis of Mesylate (+)-9a in Acetic Acid. A solution of 60
mg of (+)-9a in 6 mL of 0.05 M NaOAc in acetic acid was kept at
room temperature for 3 days. A standard workup as described for
acetolysis of recemic 9a gave 51 mg of N-methyl-3-acetoxy-3-
phenylazetidine-2-thione, [R]D ) +3.1° (acetone).
Solvolysis of Mesylate (+)-9a in Acetone-Water. A solution of
61.5 mg of (+)-9a in 5.5 mL of 4:1 acetone/water (by volume)
containing 30 mg of 2,6-lutidine was kept at room temperature for 8
days. The actone was removed using a rotary evaporator and the
residue was taken up into ether. The ether extract was washed with
dilute HCl, water, and saturated NaCl solution and dried over MgSO4.
Solvent removal using a rotary evaporator left 38 mg of N-methyl-3-
hydroxy-3-phenylazetidine-2-thione, [R]D ) -2.6° (acetone).
In a similar fashion, reaction of 590 mg of (+)-9a with 708 mg of
NaN3 in 15 mL of DMF for 8 h at 70 °C gave 360 mg (85%) of (-)-
18, [R]D ) -221° (CH3OH).
Preparation of (-)-N-Methyl-3-amino-3-phenylazetidine-2-thione,
(-)-19. A solution of 335 mg of (-)-18 and 403 mg of triphenylphos-
phine in 6 mL of CH2Cl2 was stirred at room temperature for 2 days.
The solvent was then removed using a rotary evaporator and the crude
iminophosphorane was dissolved in 12 mL of tetrahydrofuran. Six
milliliters of 5% aqueous NaOH solution was then added and the stirred
mixture was heated in an oil bath maintained at 65 °C for 3.5 days.
The THF was then removed using a rotary evaporator and the residue
was acidified with 10% HCl solution. After extraction with two
portions of ether, the aqueous phase was treated with excess Na2CO3.
The mixture was then extracted with three portions of ether and the
combined ether extracts were washed with saturated NaCl and dried
over MgSO4. The solvent was removed using a rotary evaporator and
the solid residue was washed with ether. After drying under vacuum,
138 mg (47%) of (-)-N-methyl-3-amino-3-phenylazetidine-2-thione,
(-)-19, remained, mp 110-111 °C; [R]D ) -141° (CH3OH). This
1
material contained an 80% ee of (-)-19 as determined by H NMR
using the chiral shift reagent tris(3-heptafluoropropylhydroxymeth-
ylene)((+)-camphorato)europium(III). The ether wash was concentrated
using a rotary evaporator and the residue was chromatographed on 6 g
of silica gel using ether as an eluent. An additional 71 mg (24%) of
amine 19 was obtained which contained a 60% ee of (-)-19. 1H NMR
(CDCl3) δ 7.68-7.60 (m, 2 H), 7.42-7.26 (m, 3 H), 4.065 (d, J ) 7.0
Hz, 1 H), 3.956 (d, J ) 7.0 Hz, 1 H), 3.198 (s, 3 H), 2.120 (br s, 2 H).
13C NMR (CDCl3) δ 208.80, 139.69, 128.56, 128.12, 125.99, 69.77,
67.01, 31.68. Anal. Calcd for C10H12N2S: C, 62.47; H, 6.29.
Found: C, 62.37; H, 6.34.
Kinetics Procedures. Rates of reaction of the substrates described
1
in this paper were determined by either H NMR, 19F NMR spectros-
copy or UV spectroscopy. Rates of solvolyses of 9a and 10a in
methanol and ethanol, as well as 9c and 10c in trifluoroethanol, were
1
Determination of kazide/kMeOH Ratios. General Procedure. The
appropriate substrate 9 or 10 (approximately 10 mg) was dissolved in
3.5 mL of a 0.20 M solution of NaN3 in methanol. The mixture was
kept at room temperature for approximately 10 half lives and then the
methanol was removed using a rotary evaporator. The residue was
taken up into ether and water and the ether extract was dried over
MgSO4. After filtration, the solvent was removed using a rotary
evaporator and the residue was analyzed for relative amounts of 16
monitored by H NMR spectroscopy by monitoring the changing of
the chemical shift of the methyl groups of the buffering base
2,6-lutidine.22 Rates of solvolyses of trifluoroacetates 9d and 10d,
triflate 10f, and 1-phenylcyclobutyl trifluoroacetate in acetic acid
(containing 0.05 M sodium acetate) were monitored by 19F NMR
spectroscopy.23 Solvolyses of 9d and 10d in acetic acid (0.05 M sodium
acetate), which are subject to a substantial common ion rate suppression,
were carried out at low substrate concentrations (0.005 M), where
changes in trifluoroacetate ion concentration were minimal over the
course of the reaction. Rates of solvolysis of 10a in acetic acid were
monitored by 1H NMR integration. Rates of solvolvses of 9a and 10a
in (CF3)2CHOH (containing 10-3 M Et3N) were monitored by UV
spectroscopy at 264 and 225 nm, respectively, using previously
described methods.24 The remainder of the solvolysis rates given in
Table 1 were also monitored by UV spectroscopy at wavelengths
between 258 and 271 nm. The alcohol solvents all contained 2 × 10-4
M 2,6-lutidine, which becomes protonated as the reaction proceeds.
Computational Studies. Ab initio molecular orbital calculations
were performed using either the Gaussian 92 or Gaussian 94 series of
programs.18 All structures were characterized as true minima via
frequency calculations which showed no negative frequencies.
1
and 17 by H NMR. The following is representative.
Reaction of 10.1 mg of mesylate 9a in 3.5 mL of 0.20 M NaN3 in
methanol for 18 h at 24 °C gave a mixture of 16 and 17 in a 14.7:85.3
ratio as determined by NMR integration of the doublets at δ 4.054 and
4.156 due to these respective products.
In the case of the relatively unreactive mesylate 9c, the reaction
was worked up after 13 days at room temperature. Most of 9c remained
unreacted and the product ratio reported is based on the 37% of 9c
that had reacted.
Reaction of Mesylate 9a with Sodium Azide. A mixture of 260
mg of NaN3 and 210 mg of mesylate 9a in 15 mL of dimethylform-
amide was heated with stirring at 70 °C for 9 h. The mixture was
diluted with 40 mL of water and extracted with two portions of ether.
The combined ether extracts were washed with water and saturated
NaCl solution and dried over MgSO4. After solvent removal using a
rotary evaporator, the residue was chromatographed on silica gel. Azide
18 (148 mg; 88% yield) eluted with 30% ether in hexanes. IR (CCl4)
Acknowledgment is made to the National Science Founda-
tion for support of this research.
2090, 2120 cm-1 1H NMR (CDCl3) δ 7.71-7.74 (m, 2 H), 7.49-
.
JA9624329
7.36 (m, 3 H), 4.050 (d, J ) 6.9 Hz, 1 H), 4.012 (d, J ) 6.9 Hz, 1 H),
3.192 (s, 3 H). 13C NMR (CDCl3) δ 199.55, 134.60, 129.11, 128.79,
126.70, 71.85, 63.62, 31.53. Anal. Calcd for C10H10N4S: C, 55.03;
H, 4.62. Found: C, 55.20; H, 4.54.
(22) Creary, X.; Jiang, Z. J. Org. Chem. 1994, 59, 5106.
(23) Creary, X.; Wang, Y.-X. J. Org. Chem. 1992, 57, 4761.
(24) Creary, X.; Hatoum, H.; Barton, A.; Aldridge, T. E. J. Org. Chem.
1992, 57, 1887.