Kinetics and mechanism of oxirane formation by darzens condensation of ketones: Quantification of the electrophilicities of ketones

Article abstract of DOI:10.1021/jacs.8b01657

The kinetics of epoxide formation by Darzens condensation of aliphatic ketones 1 with arylsulfonyl-substituted chloromethyl anions 2 (ArSO₂CHCl^-) have been determined photometrically in DMSO solution at 20 °C. The reactions proceed v

Full text of DOI:10.1021/jacs.8b01657

Article
pubs.acs.org/JACS
Cite This: J. Am. Chem. Soc. XXXX, XXX, XXX−XXX
Kinetics and Mechanism of Oxirane Formation by Darzens
Condensation of Ketones: Quantification of the Electrophilicities of
Ketones
Zhen Li, Harish Jangra, Quan Chen, Peter Mayer, Armin R. Ofial,* Hendrik Zipse,*
and Herbert Mayr*
Department Chemie, Ludwig-Maximilians-Universitat Munchen, Butenandtstrasse 5-13, 81377 Munchen, Germany
̈
̈
̈
S
* Supporting Information
ABSTRACT: The kinetics of epoxide formation by Darzens
condensation of aliphatic ketones 1 with arylsulfonyl-substituted
chloromethyl anions 2 (ArSO2CHCl−) have been determined
photometrically in DMSO solution at 20 °C. The reactions
proceed via nucleophilic attack of the carbanions at the carbonyl
group to give intermediate halohydrin anions 4, which
subsequently cyclize with formation of the oxiranes 3. Protonation
of the reaction mixture obtained in THF solution at low
temperature allowed the intermediates to be trapped and the
corresponding halohydrins 4-H to be isolated. Crossover experi-
ments, i.e., deprotonation of the halohydrins 4-H in the presence of
a trapping reagent for the regenerated arylsulfonyl-substituted
chloromethyl anions 2, provided the relative rates of backward (k−CC) and ring closure (krc) reactions of the intermediates.
Combination of the kinetic data (k2exptl) with the splitting ratio (k−CC/krc) gave the second-order rate constants kCC for the attack
of the carbanions 2 at the ketones 1. These kCC values and the previously reported reactivity parameters N and sN for the
arylsulfonyl-substituted chloromethyl anions 2 allowed us to use the linear free energy relationship log k2(20 °C) = sN(N + E) for
deriving the electrophilicity parameters E of the ketones 1 and thus predict potential nucleophilic reaction partners. Density
functional theory calculations of the intrinsic reaction pathways showed that the reactions of the ketones 1 with the chloromethyl
anions 2 yield two rotational isomers of the intermediate halohydrin anions 4, only one of which can cyclize while the other
undergoes retroaddition because the barrier for rotation is higher than that for reversal to the reactants 1 and 2. The
electrophilicity parameters E correlate moderately with the lowest unoccupied molecular orbital energies of the carbonyl groups,
very poorly with Parr’s electrophilicity indices, and best with the methyl anion affinities calculated for DMSO solution.
solvents were reported by H. C. Brown.4 Geneste and
INTRODUCTION
■
associates studied the kinetics of the reactions of ketones
Combinations of electrophiles with nucleophiles are the most
important reactions in organic synthesis. To predict the scope
and selectivities of such reactions, we have developed scales of
nucleophilicity and electrophilicity on the basis of eq 1, which
characterizes electrophiles by one parameter, E (electro-
philicity), and nucleophiles by two solvent-dependent param-
eters, N (nucleophilicity) and sN (susceptibility).1
5a
5b
5b,c
2−
5e
with BH4 , CN−, SO3
,
NH2OH,5b,d and RS− in
−
water and reported linear correlations5f between the different
sets of data. Thermodynamics accounts for the fact that
ordinary acceptor-stabilized carbanions (e.g., malonate anions),
which have previously been used as reference nucleophiles for
the quantification of electrophilic reactivities,1b,6 are not
suitable for the determination of the E parameters of carbonyl
compounds in aprotic solvents: Due to the high basicity of the
initially formed alkoxide anions (pKaH = 29.0 for MeO− in
DMSO),7 additions of weakly basic carbanions (pKaH ≈ 16 for
dimethyl malonate in DMSO)8 to ordinary ketones and
aldehydes are highly endergonic in aprotic solvents and only
proceed in the presence of a suitable proton source. For that
reason, reference nucleophiles are needed, which yield
intermediates that undergo fast subsequent irreversible
log k2(20°C) = sN(N + E)
(1)
Though carbonyl compounds belong to the most frequently
employed electrophiles in organic synthesis, there has been
only one previous attempt to integrate aldehydes in these
scales.2 The major problem for the quantitative determination
of the electrophilic reactivities of carbonyl compounds is the
fact that the nucleophilic attack at the carbonyl group is often a
reversible process, which is followed by an irreversible rate-
determining step.3
In the late 1950s, kinetic investigations of the reactions of
ketones and aldehydes with sodium borohydride in protic
Received: February 9, 2018
© XXXX American Chemical Society
A
DOI: 10.1021/jacs.8b01657
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
reactions to form stable products. One possibility is to use
carbanions carrying a leaving group (LG) in the α-position,
since the resulting intermediates may undergo cyclization with
formation of epoxides (Scheme 1).
RESULTS
■
Product Study. The reactions of the ketones 1a−l with the
carbanions 2a,b in anhydrous DMSO proceeded smoothly at
room temperature (Scheme 2) and gave the epoxides 3 in good
yields. The asymmetric ketones 1i−l generally reacted with low
diastereoselectivity. Only in the reaction of 1l with 2b, the
formation of the diastereomer with ArSO2 and CF3 trans to
each other is highly preferred (de = 88%) (Scheme 2). The
different stereoselectivities of 2a and 2b in reactions with 1l
have been observed in numerous experiments where 1l was
used as a trapping reagent in crossover experiments (see
below). As shown in the Supporting Information, 2a always
gave 2/1 mixtures of two diastereomers, while 2b gave one
diastereomer almost exclusively, possibly because 1l reacts with
2a, but not with 2b, under diffusion control.17
Scheme 1. Epoxides from Carbonyl Compounds
For LG = Hal, the reaction depicted in Scheme 1
corresponds to the Darzens condensation,9,10 which has
mechanistically been investigated by Ballester11 and others.3a,12
Whereas early work has preferentially been performed with α-
halogen-substituted esters, ketones, and aldehydes (Acc =
CO2R or COR), Vogt and Tavares reported that α-halo-
substituted sulfones (Acc = ArSO2) also undergo the reaction
sequence shown in Scheme 1 to give sulfonyl-substituted
epoxides.3b For LG = R2S+, the nucleophile in Scheme 1 is a
sulfonium ylide, and the sequence depicted in Scheme 1 then
corresponds to the Corey−Chaykovsky epoxidation.13,14
In previous work, we have determined the nucleophile-
specific reactivity parameters for acceptor-substituted sulfonium
ylides15 and for arylsulfonyl-substituted chloromethyl anions.16
Since acceptor-substituted sulfonium ylides are not sufficiently
nucleophilic to react with typical ketones, we have employed
anions 2 (Scheme 2) as reference nucleophiles to quantify the
electrophilicities of ketones.
Kinetic Investigations. All kinetic investigations were
performed in anhydrous DMSO solution at 20 °C by following
the disappearance of the UV/vis absorptions of the carbanion
2a (320 nm) or 2b (405 nm) under pseudo-first-order
conditions ([1]0/[2]0 > 10). As the carbanions 2a,b decompose
on the minute time scale at 20 °C (depending on the method
of preparation), they were generated by treatment of their
conjugate CH acids with 1.00−1.05 equiv of t-BuOK in dry
THF at −78 °C. Small amounts of these solutions were
dissolved in DMSO at 20 °C immediately before the ketones 1
were added. The first-order rate constants kobsd were obtained
by least-squares fitting of the exponential function A = A0
exp(−kobsdt) + C to the observed time-dependent absorbances
A of 2 (Figure 1a). The slopes of the linear correlations
between kobsd and the different concentrations of 1a−j (Figure
exptl
Scheme 2. Reactions of Carbanions 2 with Ketones 1 and Corresponding Gross Second-Order Rate Constants k2
a
b
Counterion: K+ for kinetics, Na+ for product studies. Carbanions 2a,b generated by treatment of (2a,b)-H with t-BuOK, as described in the section
c
d
“Kinetic Investigations”. Isolated yield obtained after chromatographic purification. Yield determined by 1H NMR spectroscopy using m-xylene as
an internal standard. Too fast to be measured by the stopped-flow technique.
e
B
DOI: 10.1021/jacs.8b01657
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 3. Mechanism of the Reactions of Arylsulfonyl-
Substituted Chloromethyl Anions with Ketones
d[4]/dt = kCC[1][2] − k−CC[4] − krc[4]
d[3]/dt = krc[4]
(3)
(4)
As the intermediate β-chloroalkoxide anion 4 is formed as a
short-lived species, the Bodenstein approximation holds (d[4]/
dt = 0), and the concentration of 4 is given by eq 5.
Substitution into eq 4 yields eq 6, and k2exptl is a function of kCC
k−CC, and krc as shown by eq 7.
,
[4] = kCC[1][2]/(k−CC + krc)
(5)
⇒d[3]/dt = −d[2]/dt = kCCkrc[1][2]/(k−CC + krc)
(6)
(7)
exptl
⇒k2
= kCC/(k−CC/krc + 1)
According to eq 7, the rate of the attack of 2 at the carbonyl
group (kCC) can be derived from the measured rate constant
k2exptl (Scheme 2) if the ratio k−CC/krc is known. To determine
k−CC/krc, we have developed an independent access to the
intermediate 4.
Synthesis of the Halohydrins 4-H. Whereas treatment of
2-H with base in the presence of ketones 1 at ambient
temperature led to the formation of the epoxides 3 (Scheme 2),
the reactions of 2a-H with BuLi or of 2b-H with lithium
diisopropylamide (LDA) in THF at −78 °C, followed by
addition of the ketones 1a−j, and subsequent acidification at
low temperature yielded the halohydrins 4-H in good yields
(Scheme 4).19 In the case of the acyclic ketones 1i and 1j, two
diastereomeric compounds were formed, which were separated
by column chromatography on silica gel and fully characterized.
The structure of 4ca-H was confirmed by single-crystal X-ray
crystallography and showed a conformer with chlorine gauche
to the hydroxyl group (Figure 2).
Examination of the Reversibility of the Attack of 2 at
the Ketones 1. To examine whether the intermediates 4,
generated by treatment of the halohydrins 4-H with base,
undergo ring closure with formation of 3 (krc, Scheme 3) or
retroaddition with regeneration of 1 and 2 (k−CC, Scheme 3), it
was necessary to find a trapping reagent which rapidly
intercepts 2 after its generation from 4. In view of their high
reaction rates (Scheme 2), ketones 1g, 1j, 1k, and 1l were
considered to be suitable trapping agents. Ketone 1j was then
eliminated from this series because the resulting oxirane 3jb
turned out not to be stable at 20 °C.
Figure 1. (a) Monoexponential decay of the absorbance A of 2b (at
405 nm) during the reaction of 1a (3.91 × 10−3 mol L−1) with 2b
(2.50 × 10−4 mol L−1) in DMSO at 20 °C (the remaining absorbance
is due to products generated by degradation of carbanion 2b). (b) kobsd
for the reaction of 1a with 2b versus the concentration of 1a.
exptl
1b) correspond to the second-order rate constants k2
listed
in Scheme 2.
Table 1 shows the role of counterions on the reaction
kinetics. Neither addition of 18-crown-6 ether to the potassium
salts of 2a or 2b nor exchange of t-BuOK by Schwesinger’s base
P4-tBu18 for the generation of 2b from its conjugate acid had a
exptl
significant effect on the second-order rate constants k2
.
Determination of the Rate-Limiting Step. As shown in
Scheme 3, nucleophilic attack of 2 at the ketone 1 yields the
intermediate alkoxide anion 4, which either cyclizes with
formation of the epoxide 3 or undergoes retroaddition with
regeneration of ketone 1 and carbanion 2.
The time-dependent concentrations of 2, 4, and 3 can be
expressed by eqs 2−4.
d[2]/dt = −kCC[1][2] + k−CC[4]
(2)
(M−1 s−1) for the Reactions of the Ketones 1 with Carbanions 2 under Various
exptl
Table 1. Second-Order Rate Constants k2
Conditions
a
b
k2exptl(P4-tBu)
exptl,
ketone
nucleophile
k2
k2exptl(18-crown-6)
1e
1f
2a
2b
2b
2b
(1.77 0.13) × 104
(7.62 0.35) × 103
(1.81 0.07) × 104
(3.21 0.55) × 104
(1.79 0.11) × 104
(7.65 0.47) × 103
(1.71 0.04) × 104
(3.59 0.33) × 104
1g
1j
(1.66 0.05) × 104
a
b
Data from Scheme 2. 18-Crown-6 (2.0−2.5 equiv) was added to the potassium salts of 2a,b.
C
DOI: 10.1021/jacs.8b01657
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 4. Synthesis of the Halohydrins 4-H
Scheme 5. Competition Reactions To Examine the
Suitability of Ketones 1g,k,l as Trapping Agents
Scheme 6. Crossover Reaction of the Cyclohexanone Adduct
4ca-H
Figure 2. ORTEP drawing of the crystal structure of 4ca-H (the
ellipsoid probability level is 50%).
concentration than 1c, any regenerated carbanion 2a will
exclusively be converted into the crossover product 3ga, and
the ratio [3ga]/[3ca] equals the ratio k−CC/krc.
When 1/1 mixtures of 1g on one side and of 1c, 1d, or 1i on
the other were combined with 0.5 equiv of the carbanion 2a,
the oxiranes derived from 1g (i.e., 3ga) were formed exclusively
(Scheme 5). Since 1g is only 6 times more reactive than 1h, 3
equiv of 1g was employed to obtain 3ga exclusively from a
mixture of 1h and 1g.
The oxiranes 3la and 3lb were the only products obtained
from the reactions of 3/1 mixtures of 1l and 1a−h with 2 (1
equiv with respect to 1a−h). Since the product obtained by
treatment of a mixture of 1j and 1l with 2b was difficult to
analyze, 1k was used as a trapping agent, and treatment of a 7/1
mixture of 1k and 1j with 1 equiv of 2b gave the oxirane 3kb
exclusively.
The principle of the crossover experiments is illustrated in
Scheme 6. When the independently synthesized halohydrin
4ca-H is treated with NaOH in the presence of the highly
reactive ketone 1g, the generated intermediate 4ca has the
choice of undergoing either ring closure with formation of the
epoxide 3ca or retroaddition with regeneration of 1c and 2a. As
1g is considerably more reactive and present in higher
Scheme 7 shows that in all crossover experiments at least 3
equiv of trapping agents was employed to ensure that they will
quantitatively intercept the regenerated carbanions 2. In
Scheme 7, one can furthermore see that, in most cases
investigated, ring closure (krc) is up to 8 times faster than
retroaddition. Entries 5−7 show, however, that the inter-
mediates generated from cycloheptanone (1d) undergo
retroaddition 3−4 times faster than ring closure. Comparison
of entry 5 with entry 7 and of entry 13 with entry 14 indicates
that almost the same k−CC/krc ratio is obtained with different
trapping agents, and entries 5/6 and 8/9 show that the nature
of the counterion (K+ vs Na+) has only a small influence on this
ratio. The similarity of k−CC/krc in entries 3/4, 8/10, and 13/15
implies that the ratio of retroaddition vs ring closure is almost
independent of the substituents at the arylsulfonyl groups.
Entries 16/17 as well as 18/19 show that the two
diastereomeric halohydrins obtained from the asymmetric
ketones 1i and 1j react with significantly different k−CC/krc
ratios.
D
DOI: 10.1021/jacs.8b01657
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
the 1H NMR spectrum of 3ga, the product formed by trapping
the regenerated carbanion 2a with the ketone 1g. Treatment of
ketone 1i with anion 2a yielded a mixture of the diastereomeric
epoxides 3ia′ and 3ia″ (Figure 3b). Since the ring protons A′
and A″ of the epoxides 3ia′ and 3ia″ have similar chemical
Scheme 7. Crossover Reactions of 4-H
1
shifts, their ratio was derived from the H NMR signals of the
methyl groups B′/B″ and C′/C″. Nuclear Overhauser effect
(NOE) experiments show that the methyl resonances at lower
field (B′ and C″) arise from the groups cis to the phenylsulfonyl
substituent.20 Parts c and d of Figure 3 reveal that the epoxides
3ia′ and 3ia″ are formed stereospecifically from the
diastereomeric halohydrins 4ia′-H and 4ia″-H, respectively.
When 4ia′-H is treated with NaOH in the presence of 1g,
epoxides 3ia′ and 3ga are formed in the ratio 1/2.2, as derived
from the integrals of protons A′ and A in Figure 3c. Since there
are no peaks at δ 1.32 and 0.93, the chemical shifts of the
methyl protons (B″ and C″) of the diastereomer 3ia″, we can
conclude that 4ia′ either cyclizes with formation of 3ia′ or
fragments with formation of 1i and 2a, the latter of which is
subsequently trapped by 1g to give 3ga.
Analogously, treatment of the other diastereomer (4ia″-H)
with NaOH in the presence of 1g yields the epoxides 3ia″ and
3ga in a ratio of 1/6.9 (from integrals A″ and A, Figure 3d).
The stereospecificity of this cyclization, i.e., the exclusive
formation of 3ia″ from 4ia″-H, can be derived from the
absence of 3ia′ in the product mixture, which would be
a
Determined by 1H NMR spectroscopic analysis of the crude product.
b
Product yields determined by using 1,3,5-trimethoxybenzene as an
c
internal standard (Supporting Information). KOH was used as the
base instead of NaOH.
1
detectable by a H NMR signal for the methyl group B′ at δ
1.63 and less clearly by the methyl triplet at δ 0.80 for C′.
Combination of the k−CC/krc ratios from Scheme 7 with
k2exptl from Scheme 2 according to eq 7 yields the rate constants
for nucleophilic attack of 2 at the ketones 1 (kCC), which are
listed in Table 2. While this procedure is straightforward for the
reactions with symmetrical ketones, the situation is more
complex for unsymmetrical ketones because their reactions
with the carbanions 2 yield mixtures of diastereomeric
halohydrins 4-H, as specified for 1i and 1j in the last two
entries of Table 2.
With the product ratio 3ia′/3ia″ = 2.6 given in Scheme 2, we
exptl
can split the measured gross second-order rate constant k2
=
74.2 M−1 s−1 for the reaction of 1i with 2a (Scheme 2) into the
partial rate constants k2′(exptl) = 53.6 M−1 s−1 and k2″(exptl) = 20.6
M−1 s−1 for the formation of 3ia′ and 3ia″, respectively. The
ratio of these partial rate constants corresponds to the observed
product ratio (eq 8) given in Scheme 2, and their sum
corresponds to the measured rate constants (k2exptl in Scheme 2,
eq 9).
The stereospecifity of ring closure has exemplarily been
studied for the reaction of 2a with pentan-2-one (1i). The
diastereomeric halohydrins 4ia′-H and 4ia″-H undergo either
stereospecific ring closure with formation of epoxides or
retroaddition with regeneration of 1i and 2a. Figure 3a shows
k2′(exptl)/k2″(exptl) = 2.6
(8)
k2′(exptl) + k2″(exptl) = k2
= 74.2 M−1 s−1
exptl
(9)
exptl
Table 2. Determination of Second-Order Rate Constants kCC from Measured Rate Constants k2
and Ratios k−CC/krc
a
b
(M−1 s−1
)
(k−CC/krc) + 1
kCC (M−1 s−1
)
E
kcalcd/kCC
exptl,
ketone
nucleophile
k2
1a
1b
1c
2b
2b
2a
2b
2a
2a
2b
2b
2b
2a
2b
2a
7.05 × 103
1.31 × 102
2.98 × 103
2.61 × 102
8.49 × 101
1.77 × 104
1.82 × 103
7.62 × 103
1.81 × 104
2.12 × 104
3.00 × 103
7.42 × 101
1.89
1.63
1.60
1.86
4.5
1.33 × 104
2.14 × 102
4.77 × 103
4.85 × 102
3.8 × 102
−17.5
−21.0
−19.9
identical
identical
0.69
c
c
1.6
1d
1e
−22.1
−18.4
identical
0.58
1.37
1.33
1.13
1.57
1.19
1.21
2.42 × 104
2.42 × 103
8.61 × 103
2.84 × 104
2.52 × 104
3.63 × 103
1.9
1f
−17.9
−16.9
−18.2
identical
identical
0.67
1g
1h
c
1.6
d
e
1i
1j
3.2
3.3 × 102
−22.3
−15.6
identical
d
7.9
d
e
2b
3.21 × 104
3.2
1.3 × 105
identical
identical
d
6.1
a
b
c
d
From Scheme 2. From Scheme 7. Calculated by averaging the individual E parameters. Ratios (k−CC/krc) for the individual halohydrin
e
diastereoisomers. kCC = k′CC + k″CC (see the text for the calculation).
E
DOI: 10.1021/jacs.8b01657
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
1
Figure 3. Examination of the stereospecifity of ring closure of the diastereomeric halohydrins 4ia′-H and 4ia″-H by H NMR spectroscopy: (a)
independently synthesized trapping product of regenerated 2a, (b) mixture of 3ia′ and 3ia″ obtained from the reaction of 2a with 1i, (c) exclusive
formation of 3ia′ and 3ga (1/2.2 = k′rc/k′−CC) by treatment of 4ia′-H with NaOH in the presence of 1g, (d) exclusive formation of 3ia″ and 3ga (1/
6.9 = k″rc/k″−CC) by treatment of 4ia″-H with NaOH in the presence of 1g.
Figure 4. Gibbs energy profile (kJ mol−1) for the reaction of carbanion 2a with pentan-2-one (1i) at 20 °C in DMSO derived from rate
measurements (Scheme 2) and crossover experiments (Figure 3, Table 2).
With application of eq 7 to the two parallel reactions with
k′−CC/k′rc = 2.2 (Figure 3c and Scheme 7, entry 16) and k″−CC
the ketones 1. In cases where the electrophilicity parameters E
are derived from reactions with 2a and 2b, both rate constants
should ideally give the same value of E. As this is not the case,
the E values derived from different reactions were averaged and
listed in Table 2. The last column of Table 2, which compares
the rate constants calculated by eq 1 with the directly
determined rate constants, shows that, in this series, eq 1
reproduces the rate constants kCC within a factor of 2.
A confirmation for the ketone reactivities derived in this way
was obtained by competition experiments. When a mixture of
diethyl maleate (5) (E = −19.49) and cycloalkanone 1b or 1c
was treated with 2a (in situ generated from 2a-H and NaOH),
mixtures of the epoxides 3 and the cyclopropane 6 were
obtained. Their ratio was determined by 1H NMR spectroscopy
/
=
k″rc = 6.9 (Figure 3d and Scheme 7, entry 17), we obtain k′CC
(2.2 + 1) × 53.6 M−1 s−1 = 1.7 × 102 M−1 s−1 and k″CC = (6.9 +
1) × 20.6 M−1 s−1 = 1.6 × 102 M−1 s−1, i.e., both halohydrins
are formed with similar rates, and the stereoselectivity
originates from the different rates of cyclization as illustrated
in Figure 4. In contrast, in THF at −78 °C, 4ia′-Li is formed
1.8 times faster than 4ia″-Li (Scheme 4).
An analogous calculation gave k′CC = 7.1 × 104 M−1 s−1 and
k″CC = 6.1 × 104 M−1 s−1 for the reaction of 2b with the
unsymmetrical ketone 1j.
Substitution of kCC and the published parameters N and sN
for 2a,b into eq 1 yielded the electrophilicity parameters E of
F
DOI: 10.1021/jacs.8b01657
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
and used to calculate the ratio k2exptl(1)/k2exptl(5) given in
Scheme 8. The ratios of direct rate measurements, k2(1)/k2(5),
Scheme 8. Examination of the E Parameters of Ketones 1
Figure 5. Gibbs energy surface (25 °C) for the reaction of 1c with 2a
[at the PCM(DMSO,UA0)/B2PLYP-D3/def2TZVPP//PCM-
(DMSO,UA0)/B3LYP-D3/6-31+G(d,p) level, kJ mol−1].
a
Equation 1 gives kCC = 1.13 × 103 M−1 s−1, which was corrected for
b
reversibility by applying eq 7 with k−CC/krc = 0.63. This work (Table
S20, Supporting Information). For the calculation, see the Supporting
Information. From Scheme 2.
c
d
rotation is higher (ΔG⧧ = +58.5 kJ mol−1 relative to separate
reactants) than the barrier for the reversal to the separate
reactants 2a and 1c. The comparable heights of the barriers for
initial nucleophilic addition and conformational reorientation
are in line with the results derived from the crossover
experiments with halohydrin 4ca-H described in Scheme 7.
Guided by the conformational analysis of halohydrin 4ca-H and
its deprotonated form 4ca, we assume that 4ca-H exists as a
mixture of conformers in solution, from which only that with
the C−Cl and C−O bonds in gauche conformation had
crystallized (see Figure 2 and Figure S15 in the Supporting
Information). Deprotonation of 4ca-H will give both adduct
conformers shown in Figure 5. While the adduct 4ca with
gauche C−Cl and C−O bonds will revert back to reactants, the
conformer with anti C−Cl/C−O orientation will cyclize to
epoxide 3ca.
The reaction of anion 2a with the more reactive ketone 1g (E
= −16.9) has been studied analogously. The resulting Gibbs
energy surface in Figure 6 shows that the nucleophilic addition
can also lead to intermediates 4ga with C−Cl/C−O gauche or
anti orientation, and the rotational barrier for their
interconversion (ΔG⧧ = +41.4 kJ mol−1 relative to separate
reactants) is again comparable to the barriers of the reverse
reaction. The barrier for epoxide ring closure is, in comparison,
lower at ΔG⧧ = +30.0 kJ mol−1, which again implies that
adducts with C−Cl/C−O anti orientation will move forward to
epoxide product rather than revert to separate reactants. In
agreement with the larger E value of ketone 1g as compared to
1c (−16.9 vs −19.9), the calculated overall Gibbs energy barrier
for reaction with anion 2a is much lower for 1g than for 1c
(+41.6 vs +58.5 kJ mol−1).
agree with product ratios from competition experiments, κ =
[3]/[6], within a factor of 2. This suggests that diethyl maleate
(5) and cyclohexanone (1c) have similar electrophilicities E,
one order of magnitude greater than the electrophilic reactivity
of cyclopentanone (1b).
Intrinsic Reaction Pathway Calculations. The mecha-
nistic picture derived from the kinetic studies was subsequently
complemented by reaction path calculations. Geometry
optimizations and calculations of intrinsic reaction pathways
have been performed at the B3LYP21-D322/6-31+G(d,p)23
level of theory in combination with the polarizable continuum
model (PCM)24 for DMSO as the solvent and UA0 radii.
Improved energies for ground and transition states have been
calculated at the PCM(DMSO,UA0)/B2PLYP25-D3/
def2TZVPP26 level. Combination of energies with thermo-
chemical corrections obtained at a lower level then yields the
reaction Gibbs energies reported in the Supporting Information
and summarized in Figures 5−7.
For the reaction of carbanion 2a with cyclohexanone (1c) (a
ketone of intermediate electrophilicity, E = −19.9), the Gibbs
energy surface is shown in Figure 5. Two distinct pathways
have been identified for the addition of anion 2a to the CO
double bond in ketone 1c, which differ by the relative
orientation of the two reactants. The energies shown are
those of the energetically best conformers for each pathway (for
full details, see the Supporting Information). The blue,
energetically less favorable reaction pathway (ΔG⧧ = +61.5 kJ
mol−1) directly yields an adduct with the C−Cl bond anti to
the C−O bond. Chloride expulsion through epoxide ring
closure is possible from this adduct with a barrier of +44.8 kJ
mol−1 (relative to separate reactants). A second, red reaction
pathway (ΔG⧧ = +47.4 kJ mol−1) leads to a primary adduct
where the C−Cl bond assumes a gauche orientation relative to
the C−O bond. Epoxide ring closure from this adduct is not
immediately possible, but requires rotation around the newly
formed C−C bond such that the C−Cl and C−O bonds attain
the anti orientation required for cyclization. The barrier for this
To test whether the mechanistic picture obtained for the
ketone addition reaction is comparable to that for addition to
electron-poor alkenes (Michael acceptors), the reaction of
anion 2a with dimethyl maleate (5*), as a model for the
experimentally studied diethyl maleate (5), was also treated
computationally (Figure 7). While the same sequence of initial
nucleophilic addition, gauche/anti reorientation, and ring
closure was also found for this system, the relative barriers
G
DOI: 10.1021/jacs.8b01657
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(1m; E = −19.5),2 in contrast to common experience that
aldehydes are generally more electrophilic than ordinary
ketones. How can this discrepancy be explained?
On the basis of Aggarwal’s report that the independently
synthesized anti-betaine 8a, formed from sulfonium ylide 7a
and benzaldehyde (1m), does not undergo retroaddition but
rapidly cyclizes with formation of trans-stilbene oxide (9a)
(Scheme 9),3c we had extrapolated that the same was true for
the betaine generated from benzaldehyde (1m) and p-cyano-
substituted sulfonium ylide 7b.2
Scheme 9. Aggarwal’s Mechanism Accounting for the High
trans Selectivity in the Epoxidation Reaction of
Benzaldehyde with Semistabilized Sulfur Ylide 7a
Figure 6. Gibbs energy surface (25 °C) for the reaction of 1g with 2a
[at the PCM(DMSO,UA0)/B2PLYP-D3/def2TZVPP//PCM-
(DMSO,UA0)/B3LYP-D3/6-31+G(d,p) level, kJ mol−1].
for the individual steps differ significantly from those found for
the ketones: While the barrier for the initial addition step (ΔG⧧
= +46.2 kJ mol−1) is similar to that for ketone 1c (in line with
the kinetic measurements described above), the barriers for
rotation and cyclopropane ring closure are much lower than the
barrier for the reverse reaction. This implies that the initial
addition step of anion 2a to alkene 5* is practically irreversible,
irrespective of the gauche/anti orientation in the initial addition
step.
Electrophilicities of Benzaldehydes. According to Table
2, most of the ketones characterized in this work have E
parameters similar to that previously reported for benzaldehyde
Figure 7. Gibbs energy surface (25 °C) for the reaction of 5* with 2a [at the PCM(DMSO,UA0)/B2PLYP-D3/def2TZVPP//PCM(DMSO,UA0)/
B3LYP-D3/6-31+G(d,p) level, kJ mol−1]. Dimethyl maleate (5*) is used as a model substrate for diethyl maleate (5).
H
DOI: 10.1021/jacs.8b01657
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
This conclusion was obviously incorrect as shown by the
following experiments. Treatment of the benzyl thioether 10
with LDA and benzaldehyde (1m) yielded the β-thio-
substituted alcohol 11, which was converted into the sulfonium
tetrafluoroborate 12 by treatment with trimethyloxonium
tetrafluoroborate (Scheme 10a). As shown in Scheme 10b,
the intermediates formed from sulfur ylides? In line with the
expected higher electrophilic reactivity of benzaldehyde (1m),
its reactions with the carbanions 2a,b were found to be too fast
for direct measurements with our stopped-flow techniques. We
succeeded, however, to measure the rate of the reaction of 2b
with the less electrophilic p-methoxybenzaldehyde (1o) in the
same way as described above for the corresponding reactions
exptl
Scheme 10. Synthesis of Sulfonium Tetrafluoroborate 12 and
Crossover Experiment To Elucidate the Rate-Determining
Step in the Epoxidation of Benzaldehyde with the Sulfonium
Ylide 7b
with ketones and obtained the second-order rate constant k2
= 2.69 × 104 M−1 s−1, which will be used in Table 3.
Subsequently, the relative reactivities of benzaldehyde (1m)
and p-methoxybenzaldehyde (1o) toward 2a were determined
by the competition experiment illustrated in Scheme 11.27
Scheme 11. Competition Experiments for Determining the
Relative Reactivities of Benzaldehyde (1m) and p-
Methoxybenzaldehyde (1o) toward 2a
From the composition of the reaction mixtures given in
Scheme 11, we derived the competition constant κ using eq
10,28 which is applicable when the competing reagents are not
used in high excess and the ratio of their concentrations
changes during the reaction.
the sulfonium ylide generated by treatment of the sulfonium
ion 13 with NaOH in the presence of p-nitrobenzaldehyde
(1n) and benzaldehyde (1m) reacts exclusively with the former
to yield the epoxide 3n. As expected, p-nitrobenzaldehyde (1n)
is much more reactive than the parent benzaldehyde (1m).
The crossover experiment in Scheme 10c shows the exclusive
formation of the epoxide 3n when 12 is treated with base in the
presence of p-nitrobenzaldehyde (1n). This observation implies
that the betaine 8c, which is formed by deprotonation of 12,
does not cyclize, but rather undergoes retroaddition with
regeneration of benzaldehyde (1m) and the sulfonium ylide 7b,
which is quantitatively intercepted by p-nitrobenzaldehyde
(1n). The rate-determining step for the reaction of the
sulfonium ylide 7b with benzaldehyde (1m) thus is the
cyclization and not the nucleophilic attack of the ylide at the
carbonyl group, as assumed for the derivation of the
electrophilic reactivity of benzaldehyde (1m) in ref 2. The
electrophilicity parameters of aldehydes reported in ref 2 thus
do not refer to the initial attack of nucleophiles at the carbonyl
group but describe the gross rate constants for the reactions of
carbonyl groups with the sulfonium ylide 7b.
[1m]o
[1m]t
[3ma]t
[1m]t
log
log 1 +
k2exptl(1m)
k2exptl(1o)
(
)
=
(
)
)
κ =
=
[1o]o
[3oa]t
[1o]t
log
log 1 +
([1o] )
(
(10)
t
From the directly measured rate constant for the reaction of
2b with 1o and the competition constant κ (Scheme 11), one
can calculate the rate constant k2exptl for the reaction of 2b with
benzaldehyde (1m) according to eq 11.
k2exptl(1m) = κk2exptl(1o)
(11)
As described in Scheme 3 and eqs 2−7, the rate constants
exptl
k2
thus obtained refer to the rates of the overall reactions.
To derive the rate of attack of the anions 2 at the carbonyl
groups of the aldehydes 1m and 1o (kCC), we must know the
degree of reversibility of the initial addition step, which again
was determined by crossover experiments. For their design, it
was necessary to identify trapping agents which can
quantitatively intercept the carbanions 2 generated by reverse
addition of the halohydrin anions. As shown in Scheme 12, the
epoxides 3la′ and 3la″ are formed exclusively when a 1/1
mixture of benzaldehyde (1m) and trifluoroacetophenone (1l)
is treated with 0.5 equiv of 2a, indicating that the fluorinated
ketone 1l is much more electrophilic than benzaldehyde (1m).
How can the rate of the initial nucleophilic attack at
aldehydes be determined? Are the chloro-substituted carban-
ions 2a,b suitable reference nucleophiles, because the
corresponding intermediates cyclize with lower barriers than
I
DOI: 10.1021/jacs.8b01657
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Equation 7 was then used to calculate the rate constants kCC
Scheme 12. Competition Experiment To Demonstrate the
Much Higher Reactivity of Ketone 1l Compared to
Benzaldehyde (1m)
for the nucleophilic attack of 2b at the aldehydes 1o,m from the
exptl
gross rate constants k2
listed in Table 3 and the k−CC/krc
ratios from Scheme 13. Substitution into eq 1 with the N and sN
parameters of 2b eventually yielded the electrophilicity
parameters E of the aldehydes 1m and 1o in the last column
of Table 3. It should be admitted, however, that there are two
uncertainties in this derivation. First, the mixtures of
diastereomers of the chlorohydrins 4ma-H and 4oa-H, which
are used for the crossover experiments in Scheme 13, are
formed by the reactions with the lithiated nucleophiles 2a-Li in
THF at −78 °C and may differ somewhat from the
diastereomeric ratios of the halohydrins generated under the
conditions of the kinetic experiments. Second, we had to use
the k−CC/krc ratios for the adducts of 2a for the calculation of
kCC in Table 3 because the epoxides obtained from 2b were not
stable. The plausibility of this assumption is based on the
comparison of entries 3/4, 8/10, and 13/15 in Scheme 7, which
In analogy to the procedure described in Scheme 4, the
chlorohydrins 4ma-H and 4oa-H (Scheme 13) were synthe-
indicated that the ratio of retroaddition vs ring closure (k−CC
/
krc) is almost independent of the substituents at the arylsulfonyl
groups.
Scheme 13. Crossover Reactions of Aldehydes 1m and 1o
These uncertainties prompted us to examine the E value for
benzaldehyde thus derived by an independent experiment.
Substitution of the E values for 1m and 14 and of the N and sN
parameters for 2a into eq 1 gives kCC, the rate constant for the
initial nucleophilic attack of 2a at these electrophiles. Since the
attack of 2a at 1m (in contrast to the attack at 14) is reversible,
k
CC was corrected by the splitting ratio k−CC/krc according to eq
7 to obtain the overall rate constant k2exptl for the formation of
the epoxide 3ma.
In the competition experiment described in Scheme 14,
which compares the electrophilic reactivity of benzaldehyde
Scheme 14. Competition Experiment To Determine the
Relative Reactivities of Benzaldehyde (1m) and Imine 14
sized in THF at −78 °C by the reaction of 2a-Li with the
aldehydes 1m and 1o, respectively, and subsequent acid-
ification. As illustrated in Scheme 13, treatment of a 1/5
mixture of 4ma-H and 1l gave the crossover product 3la in
addition to 3ma, the cyclization product of 4ma, in a ratio of
7.9/1. Since 1l is much more electrophilic than 1m (Scheme
12), we can conclude that 3ma is exclusively formed by direct
cyclization of the deprotonated chlorohydrin 4ma, whereas the
carbanion 2a, which is formed by reversal of 4ma, is
quantitatively converted into 3la. The ratio [3la]/[3ma] =
7.9 (Scheme 13a) thus reflects the ratio k−CC/krc given in Table
3.
Since the competition experiment in Scheme 11 showed 1m
to be 12 times more reactive than 1o, benzaldehyde (1m) could
be used as a trapping reagent for the crossover experiment in
Scheme 13b, and the ratio [3ma]/[3oa] = 4.7 (Scheme 13b)
again reflects the ratio k−CC/krc given in Table 3.
(1m) with that of the N-tosyl imine 14, we observed the ratios
[3ma]/[1m] = 0.22/3.31 and [15]/[14] = 1.08/2.04 from
which the reactivity ratio κ = k(14)/k(1m) = 6.6 was derived by
Table 3. Derivation of the Rate Constants kCC for Nucleophilic Attack of 2b at the Aldehydes 1o and 1m and the Resulting E
Parameters
a
b
(M−1 s−1
)
k−cc/krc
kCC (M−1 s−1
)
approximate E
exptl
aldehyde
nucleophile
k2
c
1m
1o
2b
2b
3.36 × 105
7.9
4.7
3.0 × 106
1.5 × 105
−12.9
−15.4
d
2.69 × 104
a
b
c
d
From Scheme 13. From eq 7. From eq 11 using the averaged κ from Scheme 11, k2exptl = 12.5 × (2.69 × 104 M−1 s−1). Direct rate measurement.
J
DOI: 10.1021/jacs.8b01657
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Table 4. Comparison of the Relative Reactivities of Benzaldehyde (1m) and Imine 14 toward 2a Derived from Rate
Measurements and Competition Studies
a
kCC (M−1 s−1
)
k2
(M−1 s−1
)
krel(rates)
krel(competition)
exptl
electrophile
E
b
1m
14
−12.9
−13.05
2.85 × 106
2.47 × 106
3.20 × 105
2.47 × 106
1
1
c
d
7.7
6.6
a
b
c
d
From eq 1 with E from this table and N = 28.27 and sN = 0.42 for 2a. Calculated with eq 7 using k−CC/krc = 7.9 from Table 3. Reference 2. From
Scheme 14; for a competition experiment with [2a-H]/[1m]/[14] = 1/5/2, a κ = 6.65 was obtained (see the Supporting Information).
Table 5. Quantum Chemically Calculated Frontier Orbital Energies (hartrees), Global (ω) and Local (ωC) Parr Electrophilicity
Indices (eV), and Methyl Anion Affinities (MAAs; kJ mol−1) of Ketones and Aldehydes
a
b
b
b
b
c
d
electrophile
E
εHOMO
εLUMO
global ω
local ωC
ΔGgas(−MAA)
ΔGsol‑sp(−MAA)
1a
1b
1c
1d
1e
1f
−17.5
−21.0
−19.9
−22.1
−18.4
−17.9
−16.9
−18.2
−22.3
−15.6
(−12.9)
(−15.4)
−0.24245
−0.23597
−0.23443
−0.23483
−0.22444
−0.24378
−0.23250
−0.23481
−0.24272
−0.22563
−0.25521
−0.23442
−0.02117
−0.01449
−0.01201
−0.01104
−0.01456
−0.02071
−0.02310
−0.01233
−0.00948
−0.02762
−0.06342
−0.05149
1.07
0.96
0.93
0.92
0.93
1.07
1.06
0.93
0.93
1.10
1.80
1.52
0.16
0.15
0.15
0.14
0.15
0.18
0.19
0.15
0.15
0.14
0.22
0.18
−131.6
−114.2
−126.7
−116.3
−136.2
−147.3
−158.3
−138.5
−114.3
−144.4
−155.2
−143.7
−8.0
14.6
11.7
26.8
2.9
−1.4
−4.1
5.3
1g
1h
1i
16.4
−5.9
−27.8
−13.2
1j
e
e
1m
1o
a
b
Empirical electrophilicity parameters from Tables 2 and 3, as defined in eq 1. Calculated at the B3LYP/6-31G(d,p) level in the gas phase.
Calculated at the B3LYP/6-311+G(3df,2pd)32//B3LYP/6-31G(d,p) level in the gas phase. Based on methyl anion affinities (ΔGgas), which were
c
d
corrected for solvent effects by adding single-point solvation energies calculated at B3LYP/6-31G(d,p) using the solvation model based on density
(SMD)33 (solvent = DMSO) on gas-phase optimized geometries at the same level. Approximate electrophilicities E of aldehydes (see text and
e
Table 3).
eq 10 as summarized in Table 4. The fair agreement between
μ = 1/2(εLUMO + εHOMO
)
(15)
(16)
the relative reactivities of 1m and 14 and the rate constants
calculated by eq 1 confirms the electrophilicity parameter of
benzaldehyde (1m) derived above.
+
ω = ωf
c
k
The local electrophilicity indices ωc at the carbonyl carbon
were calculated as the product of Parr’s electrophilicity index ω
and the nucleophilic Fukui function (fk ) according to eq 16.
Correlation Analysis. To elucidate the origin of the
corresponding electrophilic reactivities, we have determined
various properties of the investigated carbonyl compounds by
quantum chemical calculations (Table 5). As specified in Table
5 and Table S25 (Supporting Information), the computational
methods used in these calculations differ from those employed
in the reaction path calculations (see above) to make them
strictly comparable to our earlier work on nucleophilic
additions to Michael acceptors and carbenium ions.29
+
The Fukui function for nucleophilic attack is defined as the
change of the partial charge q at a certain atom k by adding an
+
electron to the corresponding compound; that is, fk = q(k,N
+1) − q(k,N) with N = total number of electrons.31
Previously, good correlations between the electrophilicities of
benzhydrylium ions and their LUMO energies were reported
by Liu34a and Yu.34b Figure 8 shows a moderate correlation
between the electrophilicity parameters E of ketones and their
LUMO energies in the gas phase. This correlation improves
slightly when the correlation with LUMO energies in DMSO
solution is considered (as depicted in Figure S29B, Supporting
Information). Though the correlation between the electro-
philicity parameters E and LUMO energies of Michael
acceptors has been reported to be very poor,29a Figure 8
shows that ketones are generally more electrophilic than
Michael acceptors of equal LUMO energies.
Methyl anion affinities (MAAs) have been calculated as the
negative of the reaction Gibbs energies for the addition of
methyl anion to ketones and aldehydes (eq 12). In addition, we
calculated Parr electrophilicity indices ω (eq 13) for 12
carbonyl compounds30 from the chemical hardness η (eq 14)
and the electronic chemical potential μ (eq 15). The values of η
and μ have been calculated from the energies of the lowest
unoccupied molecular orbital (εLUMO) and the highest occupied
molecular orbital (εHOMO), which were derived at the gas-phase
B3LYP/6-31G(d,p) level. As in previous studies, the Parr
electrophilicity indices are expressed in electronvolts (eV).
Whereas Figure 9 shows a moderate correlation between
electrophilicities E and Parr’s global electrophilicity index ω, a
plot of E vs the local electrophilicity index ωC (Figure S33A,
Supporting Information) has a correlation coefficient of R2 =
0.30 (!); i.e., ωC is inadequate to predict electrophilic
reactivities of ketones.35
As in our previous investigation of the electrophilic
reactivities of Michael acceptors,29a the electrophilicities E of
the ketones 1a−1j correlate fairly with the calculated gas-phase
methyl anion affinities (R2 = 0.77; Figure S17B, Supporting
ω = μ2/2η
(13)
η = εLUMO − εHOMO
(14)
K
DOI: 10.1021/jacs.8b01657
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 10. Correlation between electrophilicities (E) of ketones and
their MAAsol‑sp values (−ΔGsol‑sp, kJ mol−1) calculated at the
SMD(DMSO)/B3LYP/6-311++G(3df,2pd)//B3LYP/6-31G(d,p)
level of theory. A superscript a indicates the data point was not used
for the construction of the correlation line.
Figure 8. Correlation of the electrophilicities (E) of ketones with their
gas-phase lowest unoccupied molecular orbital energies (εLUMO
)
calculated at the B3LYP/6-31G(d,p) level of theory compared with
the corresponding correlation for Michael acceptors.
Figure 9. Correlation between electrophilicities (E) of ketones 1a−1j
and Parr’s global electrophilicity index (ω) calculated at the B3LYP/6-
31G(d,p) level of theory.
Figure 11. Correlation between the empirical electrophilicities (E, eq
1) and MAA (−ΔGsol‑sp, kJ mol−1) values calculated at the
SMD(DMSO)/B3LYP/6-311++G(3df,2pd)//B3LYP/6-31G(d,p)
level of theory for ketones and Michael acceptors.
Information),21,23,32,36 and this correlation improves further
when solvation is included in the calculated methyl anion
affinities (Figure 10). Obviously, the solvation energy of the
methyl anion is overestimated by this model, since negative
MAAs were calculated for several additions. Though the data
for the benzaldehydes 1m and 1o were not used for the
correlation because of the uncertainty of the experimental E
parameters, they also are on this best fit line, thus justifying the
approximations made for the derivation of their E parameters.
When the plots of E vs MAA for carbonyl compounds and
Michael acceptors are drawn in the same graph (Figure 11),
one can clearly see two correlation lines, which differ in two
aspects. First, the slope for the ketones is significantly larger
than that for Michael acceptors. As pointed out previously, the
slope of the Michael acceptor correlation implies that in
reactions with a nucleophile of sN = 0.7 (see eq 1) about 43% of
the differences of the Gibbs reaction energies are reflected in
the Gibbs activation energies.29a A significantly higher
percentage (75%) of the Gibbs reaction energies is mirrored
by the activation energies of the additions of nucleophiles with
the typical value sN = 0.7 to carbonyl compounds.37 Second, the
different positions of the two correlation lines imply that
additions to carbonyl compounds are significantly faster than
Michael additions of equal thermodynamic driving force
(ΔrG°). In Marcus terminology,38a−e this means that Michael
additions proceed with significantly higher reorganization
energies than nucleophilic additions to carbonyl groups,
which can be explained by the much greater movements of
electrons and structural changes occurring in Michael additions
(Scheme 15).38f
Structure−Reactivity Relationships. As shown in Table
6, the reactivity order cyclobutanone (1a) > cyclohexanone
(1c) > cyclopentanone (1b) > cycloheptanone (1d), which we
found for reactions with carbanions 2a,b, had previously been
L
DOI: 10.1021/jacs.8b01657
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
mol−1 in DMSO solution. As discussed in detail in Figures S24
and S25 (Supporting Information), this change is due to the
fact that the cyclohexanolate conformer with oxygen in the axial
position, which is most stable in the gas phase, is less efficiently
solvated and becomes even less stable in DMSO solution than
the conformer with equatorial oxygen. Thus, the overall poorer
solvation of the cyclohexanolate ion accounts for the fact that
the MAA of cyclohexanone, which is much higher than that of
cyclopentanone in the gas phase, is only slightly higher in
solution. The poor solvation of the cyclohexanolate anion with
oxygen in the axial position analogously accounts for the
finding (Table 5) that the MAAs of cyclobutanone and
cyclohexanone differ only slightly in the gas phase (5 kJ mol−1)
but strongly in solution (20 kJ mol−1).
Scheme 15. Less Movement of Electrons and Nuclei
Required in Nucleophilic Additions to Carbonyl Groups
Than to Michael Acceptors
Table 6. Comparison of the Reactivities of the
Cycloalkanones 1a−d toward Different Nucleophiles
a
b
kCC(2b)
k2(NaBH4)
a
b
c
(M−1 s−1
)
(M−1 s−1
)
krel(2b) krel(NaBH4) krel(Nu)
1a
1b
1c
1d
1.33 × 104
2.14 × 102
4.85 × 102
3.88 × 101
2.66 × 10−2
7.01 × 10−4
1.61 × 10−2
1.02 × 10−4
27
0.44
1.6
1.2
As shown in Table 7, the introduction of electronegative
elements in the 4-position of cyclohexanone leads to an
0.044
1
0.066
1
1
d
Table 7. Influence of Heteroatoms in the γ-Position on the
Reactivities of Cyclic Ketones
0.080
0.0063
0.11
a
b
In DMSO, 20 °C, from Table 2. In iPrOH, 0 °C, from ref 4c.
c
Averaged value derived from reactions with NH2OH, SO32−, CN−,
−
BH4 , and HOC2H4S− in aqueous solution; krel(Nu) = 10B calculated
from Geneste’s B values defined by the relation log k = A log k0 + B in
ref 5f. From the rate constant with 2a (Table 2) divided by 9.8, the
a
b
−
d
ketone 1
krel(2b)
krel(BH4 )
reactivity ratio 2a/2b toward cyclohexanone (1c).
1c (X = CH2)
1e (X = NMe)
1f (X = O)
1.0
5.0
18
1.0
9.9
observed in reactions with other nucleophiles, though the
relative reactivities of the different cycloalkanones depend on
the reaction partner and conditions.39 The uniformly higher
electrophilic reactivity of cyclobutanone (1a) can partially be
explained by the higher release of ring strain during conversion
of the sp2 carbon in the four-membered ring into an sp3 carbon.
Table 5 shows, however, that the gas-phase methyl anion
affinity of 1a is only 5 kJ mol−1 higher than that of
cyclohexanone (1c), indicating that release of strain can only
account for part of the reactivity difference of 1a and 1c. Since
the MAA of 1a is almost 20 kJ mol−1 higher than that of 1c in
DMSO solution, we must conclude that differences of solvation
are the major reason for the higher reactivity of cyclobutanone
(1a) toward 2b in solution. Let us analyze the origin of the
solvation effect in the following comparison of cyclohexanone
with cyclopentanone.
1g (X = S)
59
11.2
a
b
kCC in DMSO from Table 2. In water/dioxane (1/1) at 25 °C (from
ref 5a).
increase of the electrophilic reactivity toward carbanion 2b as
well as toward BH4 . Possibly different solvation accounts for
−
the fact that the relative reactivities in the two reaction series
correlate only moderately. From the fact that the data for the
four six-membered ring ketones 1c, 1e, 1f, and 1g are perfectly
on the correlation line of Figure 10, one can conclude that the
relative reactivities of these ketones are predominantly
controlled by the thermodynamics of the CC-bond-forming
step. Though oxygen is more electronegative than sulfur,
tetrahydrothiopyranone (1g) is more electrophilic than
tetrahydropyranone (1f), which may be due to through-bond
interaction.42
When the β-carbon of ketones is replaced by sulfur (1i → 1j)
or oxygen (1i → 1k), the heteroatom effect is larger (by a
factor of 400 for S and not measurable for O) than the γ-
heteroatom effect shown in Table 7 and follows the
electronegativity order O ≫ S.
H. C. Brown rationalized the 23 times faster reaction of
NaBH4 with cyclohexanone compared to cyclopentanone by a
change of torsional strain: As the hybridization of the carbonyl
carbon changes from sp2 to sp3, the torsional strain increases in
the five-membered ring (eclipsed bonds), but decreases in the
six-membered ring because the equatorial hydrogens are nearly
eclipsed with carbonyl oxygen in cyclohexanone and attain
staggered arrangements in the chair conformation of cyclo-
hexanol.40 Since the opposite rehybridization takes place in the
rate-determining step of SN1 reactions of cycloalkyl halides,
differences in torsional strain were analogously used to explain
the much larger solvolysis rates of cyclopentyl halides
compared to cyclohexyl halides.40 We had already doubted
CONCLUSIONS
■
The arylsulfonyl-substituted chloromethyl anions 2a,b are
suitable reference nucleophiles for the determination of the
electrophilic reactivities of ordinary aliphatic ketones. The rate
constants kCC for the initial nucleophilic attack are accessible by
combination of the directly measured gross rate constants
(k2exptl) for the formation of the epoxides 3 from the reactants 1
3
2
that the change from Csp to Csp is the major contribution to
this difference of the SN1 reactivities because methylenecyclo-
pentane was found to react 50 times faster with benzhydrylium
ions than methylenecyclohexane though the rate-determining
step does not involve rehybridization of a ring carbon.41
The 12.5 kJ mol−1 higher gas-phase methyl anion affinity of
cyclohexanone (1c) compared to cyclopentanone (1b) in
Table 5 supports the torsional strain argument. However, the
difference between the MAAs of 1c and 1b shrinks to 2.9 kJ
and 2 with the degree of reversibility of the initial step (k−CC
/
krc). This ratio was derived from crossover experiments with the
independently synthesized intermediates 4. Two reaction
pathways have been identified for the reactions of the
carbanions 2 with the ketones 1: one which yields the
intermediate halohydrin anions 4 with the C−Cl and C−O−
bonds in anti-arrangements that can undergo direct cyclization
to the epoxides 3 and a second one which gives the halohydrin
M
DOI: 10.1021/jacs.8b01657
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
anions 4 with C−Cl and C−O− in gauche orientation. The
latter undergo retroaddition with regeneration of the reactants
1 and 2, because the barrier for reversal is lower than the barrier
for rotation to give the anti conformer suitable for cyclization.
The cyclopropanation of diethyl maleate with 2a proceeds via
an analogous mechanism, with the difference that the initial
CC-bond-forming step, which also gives different conformers, is
irreversible.
The electrophilicity parameters E of the ketones 1 were
calculated by eq 1 from the rate constants kCC and the
previously reported reactivity parameters N and sN for the
carbanions 2. The E parameters, which refer to the nucleophilic
attack of 2 at the carbonyl groups, correlate moderately with
the gas-phase LUMO energies of the ketones (R2 = 0.76, Figure
8), poorly with Parr’s global electrophilicity index ω (R2 = 0.61,
Figure 9), very poorly with Parr’s local electrophilicity index ωC
(R2 = 0.30, Figure S33A in the Supporting Information), and
best with the methyl anion affinities calculated for DMSO
solution (R2 = 0.87, Figure 10). We thus do not consider Parr’s
electrophilicity indices suitable measures for electrophilic
reactivities, though electrophilic reactivities within a series of
structurally closely related Michael acceptors correlate well with
Parr’s indices.30
Comparison of the electrophilicities E of ketones with those
of Michael acceptors shows that ketones are significantly more
electrophilic than Michael acceptors of equal methyl anion
affinity (≙ Lewis acidity), indicating that nucleophilic additions
to ketones proceed over much lower Marcus intrinsic barriers
due to less electronic and geometrical reorganization than in
Michael additions.
Crossover experiments showed that the initial attack of the p-
cyanophenyl sulfonium ylide at aldehydes is reversible (in
contrast to previous extrapolations), with the consequence that
the previously reported E parameters for aldehydes correspond
to the gross rate constants for these epoxidations and do not
reflect the rates of initial attack of nucleophiles at the carbonyl
group. By using the carbanions 2 as reference nucleophiles,
estimates for the rate of nucleophilic attack at aldehydes have
been obtained, showing that the electrophilicity parameter E of
benzaldehyde (1m) is approximately 7 units greater than that of
cyclohexanone (1c).
Figure 12. Comparison of the empirical electrophilicities E of carbonyl
groups and Michael acceptors in DMSO.
ASSOCIATED CONTENT
■
S
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/jacs.8b01657.
Details of the product studies, kinetic experiments, NMR
spectra of all characterized compounds, single-crystal X-
ray structure of 4ca-H, competition experiments, cross-
over experiments, and computational analysis (PDF)
Coordinates of optimized structures (ZIP)
Crystallographic data for 4ca-H (CIF)
AUTHOR INFORMATION
■
Corresponding Authors
*ofial@lmu.de
As illustrated in Figure 12, the electrophilicities of saturated
aliphatic ketones are comparable to the CC bond reactivities
of β-phenyl-substituted α,β-unsaturated ketones and much
lower than the CC bond reactivities of terminally
unsaturated vinyl ketones. Benzaldehyde, on the other hand,
is more electrophilic than the α,β-unsaturated carbonyl
compounds depicted in Figure 12, in line with the observation
that α,β-unsaturated aldehydes usually undergo 1,2-additions
under kinetically controlled conditions. In earlier applications
of eq 1, we have shown that in reactions of nucleophiles with
carbenium ions and a variety of Michael acceptors the
electrophilicity parameters E can be treated as solvent-
independent quantities, with the consequence that all solvent
effects are shifted into the nucleophile-specific parameters N
and sN. Because of the high basicity of alkoxide ions in aprotic
media, this approximation probably does not hold for the
electrophilicities of carbonyl compounds, and systematic
investigations of solvent effects are now needed to arrive at
reliable predictions of carbonyl reactivities in different solvents.
*zipse@cup.uni-muenchen.de
*herbert.mayr@cup.uni-muenchen.de
ORCID
Armin R. Ofial: 0000-0002-9600-2793
Hendrik Zipse: 0000-0002-0534-3585
Herbert Mayr: 0000-0003-0768-5199
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This paper is dedicated to Professor Ernst-Ulrich Wurthwein
̈
on the occasion of his 70th birthday. We thank the Deutsche
Forschungsgemeinschaft (SFB 749, Projects B1 and C6) and
the China Scholarship Council (fellowship to Z.L.) for financial
support. We are grateful for computing time on the Linux-
Cluster at the Leibniz Supercomputing Centre (www.lrz.de)
and thank Dr. Peter A. Byrne (University College Cork,
Ireland) for helpful discussions.
N
DOI: 10.1021/jacs.8b01657
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(14) (a) Aggarwal, V. K.; Winn, C. L. Acc. Chem. Res. 2004, 37, 611−
620. (b) McGarrigle, E. M.; Myers, E. L.; Illa, O.; Shaw, M. A.; Riches,
S. L.; Aggarwal, V. K. Chem. Rev. 2007, 107, 5841−5883.
(15) (a) Appel, R.; Mayr, H. Chem. - Eur. J. 2010, 16, 8610−8614.
(b) Appel, R.; Hartmann, N.; Mayr, H. J. Am. Chem. Soc. 2010, 132,
17894−17900.
REFERENCES
■
(1) (a) Mayr, H.; Bug, T.; Gotta, M. F.; Hering, N.; Irrgang, B.;
Janker, B.; Kempf, B.; Loos, R.; Ofial, A. R.; Remennikov, G.;
Schimmel, H. J. Am. Chem. Soc. 2001, 123, 9500−9512. (b) Lucius, R.;
Loos, R.; Mayr, H. Angew. Chem., Int. Ed. 2002, 41, 91−95. (c) Mayr,
H.; Kempf, B.; Ofial, A. R. Acc. Chem. Res. 2003, 36, 66−77.
(2) Appel, R.; Mayr, H. J. Am. Chem. Soc. 2011, 133, 8240−8251.
(3) (a) Zimmerman, H. E.; Ahramjian, L. J. Am. Chem. Soc. 1960, 82,
5459−5466. (b) Vogt, P. F.; Tavares, D. F. Can. J. Chem. 1969, 47,
2875−2881. (c) Aggarwal, V. K.; Calamai, S.; Ford, J. G. J. Chem. Soc.,
Perkin Trans. 1 1997, 593−599.
(4) (a) Brown, H. C.; Mead, E. J.; Rao, B. C. S. J. Am. Chem. Soc.
1955, 77, 6209−6213. (b) Brown, H. C.; Wheeler, O. H.; Ichikawa, K.
Tetrahedron 1957, 1, 214−220. (c) Brown, H. C.; Ichikawa, K.
Tetrahedron 1957, 1, 221−230. (d) Brown, H. C.; Ichikawa, K. J. Am.
Chem. Soc. 1962, 84, 373−376. (e) Brown, H. C.; Muzzio, J. J. Am.
Chem. Soc. 1966, 88, 2811−2822. (f) Brown, H. C.; Wang, K. K.;
Chandrasekharan, J. J. Am. Chem. Soc. 1983, 105, 2340−2343.
(5) (a) Geneste, P.; Durand, R.; Hugon, I.; Reminiac, C. J. Org.
Chem. 1979, 44, 1971−1973. (b) Geneste, P.; Lamaty, G.; Moreau, C.;
Roque, J. P. Tetrahedron Lett. 1970, 11, 5011−5014. (c) Geneste, P.;
Lamaty, G.; Roque, J. P. Tetrahedron 1971, 27, 5539−5559.
(d) Geneste, P.; Lamaty, G.; Roque, J. P. Tetrahedron 1971, 27,
5561−5578. (e) Fournier, L.; Lamaty, G.; Natat, A.; Roque, J. P.
Tetrahedron 1975, 31, 1031−1034. (f) Finiels, A.; Geneste, P. J. Org.
Chem. 1979, 44, 1577−1578.
(6) (a) Zenz, I.; Mayr, H. J. Org. Chem. 2011, 76, 9370−9378.
(b) Kaumanns, O.; Lucius, R.; Mayr, H. Chem. - Eur. J. 2008, 14,
9675−9682. (c) Lemek, T.; Mayr, H. J. Org. Chem. 2003, 68, 6880−
6886. (d) For a comprehensive database of nucleophilicity parameters
N and sN as well as electrophilicity parameters E, see http://www.cup.
lmu.de/oc/mayr/DBintro.html.
(7) (a) Olmstead, W. N.; Margolin, Z.; Bordwell, F. G. J. Org. Chem.
1980, 45, 3295−3299. (b) Bordwell, F. G. Acc. Chem. Res. 1988, 21,
456−463.
(8) (a) Arnett, E. M.; Maroldo, S. G.; Schilling, S. L.; Harrelson, J. A.
J. Am. Chem. Soc. 1984, 106, 6759−6767. (b) For a comprehensive list
of pKa values, see http://ibond.nankai.edu.cn/.
(16) Li, Z.; Chen, Q.; Mayer, P.; Mayr, H. J. Org. Chem. 2017, 82,
2011−2017.
(17) Mayr, H.; Ofial, A. R. Angew. Chem., Int. Ed. 2006, 45, 1844−
1854.
(18) Schwesinger, R.; Schlemper, H.; Hasenfratz, C.; Willaredt, J.;
Dambacher, T.; Breuer, T.; Ottaway, C.; Fletschinger, M.; Boele, J.;
Fritz, H.; Putzas, D.; Rotter, H. W.; Bordwell, F. G.; Satish, A. V.; Ji, G.
Z.; Peters, E. M.; Peters, K.; von Schnering, H. G.; Walz, L. Liebigs
Ann. 1996, 1055−1081.
(19) (a) Durst, T.; Tin, K.-C.; De Reinach-Hirtzbach, F.; Decesare, J.
M.; Ryan, M. D. Can. J. Chem. 1979, 57, 258−266. (b) Reutrakul, V.;
Jarussophon, S.; Pohmakotr, M.; Chaiyasut, Y.; U-Thet, S.; Tuchinda,
P. Tetrahedron Lett. 2002, 43, 2285−2288.
(20) Arai, S.; Shioiri, T. Tetrahedron 2002, 58, 1407−1413.
(21) Becke, A. D. J. Chem. Phys. 1993, 98, 5648−5652.
(22) (a) Grimme, S. J. Comput. Chem. 2006, 27, 1787−1799.
(b) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. J. Chem. Phys. 2010,
132, 154104. (c) Wagner, J. P.; Schreiner, P. R. Angew. Chem., Int. Ed.
2015, 54, 12274−12296.
(23) (a) Ditchfield, R.; Hehre, W. J.; Pople, J. A. J. Chem. Phys. 1971,
54, 724−728. (b) Krishnan, R.; Binkley, J. S.; Seeger, R.; Pople, J. A. J.
Chem. Phys. 1980, 72, 650−654.
̀
(24) (a) Cances, E.; Mennucci, B.; Tomasi, J. J. Chem. Phys. 1997,
107, 3032−3041. (b) Tomasi, J.; Mennucci, B.; Cammi, R. Chem. Rev.
2005, 105, 2999−3094. (c) Mayer, R. J.; Tokuyasu, T.; Mayer, P.;
Gomar, J.; Sabelle, S.; Mennucci, B.; Mayr, H.; Ofial, A. R. Angew.
Chem., Int. Ed. 2017, 56, 13279−13282.
(25) Grimme, S. J. Chem. Phys. 2006, 124, 034108.
(26) (a) Weigend, F.; Ahlrichs, R. Phys. Chem. Chem. Phys. 2005, 7,
3297−3305. (b) Weigend, F. Phys. Chem. Chem. Phys. 2006, 8, 1057−
1065.
(27) The corresponding competition experiment with 2b could not
be performed because of the low stability of the resulting oxiranes.
(28) Huisgen, R. Angew. Chem., Int. Ed. Engl. 1970, 9, 751−762.
(9) (a) Darzens, G. Compt. Rend. 1904, 139, 1214−1217.
(b) Darzens, G. Compt. Rend. 1905, 141, 766−768. (c) Darzens, G.
(29) (a) Allgauer, D. S.; Jangra, H.; Asahara, H.; Li, Z.; Chen, Q.;
̈
́
Compt. Rend. 1906, 142, 214−215. (d) Darzens, G.; Lefebure, P.
Zipse, H.; Ofial, A. R.; Mayr, H. J. Am. Chem. Soc. 2017, 139, 13318−
13329. (b) Mayr, H.; Ammer, J.; Baidya, M.; Maji, B.; Nigst, T. A.;
Ofial, A. R.; Singer, T. J. Am. Chem. Soc. 2015, 137, 2580−2599.
Compt. Rend. 1906, 142, 714−715.
(10) (a) Newman, M. S.; Magerlein, B. J. Org. React. 1949, 5, 413−
440. (b) Rosen, T. Darzens Glycidic Ester Condensation. In
Comprehensive Organic Synthesis, 1st ed.; Trost, B. M., Fleming, I., Eds.;
Pergamon Press: Oxford, U.K., 1991; Vol. 2, pp 409−439. (c)
Aggarwal, V. K.; Crimmin, M.; Riches, S. Synthesis of Epoxide by
Carbonyl Epoxidation. In Science of Synthesis; Forysth, C. J., Jacobsen,
E. N., Eds.; Thieme: Stuttgart, Germany, 2008; Vol. 37, pp 321−406.
(d) Reutrakul, V.; Pohmakotr, M. Chloromethyl Phenyl Sulfone. In
Encyclopedia of Reagents for Organic Synthesis, 2nd ed.; Paquette, L. A.,
Crich, D., Fuchs, P. L., Molander, G. A., Eds.; Wiley: Chichester, U.K.,
2009; Vol. 4, pp 2375−2378. (e) Bako, P.; Rapi, Z.; Keglevich, G. Curr.
Org. Synth. 2014, 11, 361−376. For diastereoselective or asymmetric
Darzens reactions, see: (f) Liu, Y.; Provencher, B. A.; Bartelson, K. J.;
Deng, L. Chem. Sci. 2011, 2, 1301−1304. (g) Kuang, Y.; Lu, Y.; Tang,
Y.; Liu, X.; Lin, L.; Feng, X. Org. Lett. 2014, 16, 4244−4247. (h) Li, B.;
Li, C. J. Org. Chem. 2014, 79, 8271−8277. (i) Sakowicz, A.; Loska, R.;
(30) Domingo, L. R.; Per
6585−6591.
(31) (a) Contreras, R. R.; Fuentealba, P.; Galvan
́
ez, P.; Contreras, R. Tetrahedron 2004, 60,
́
́
, M.; Perez, P. Chem.
Phys. Lett. 1999, 304, 405−413. (b) Yang, W.; Mortier, W. J. J. Am.
Chem. Soc. 1986, 108, 5708−5711.
(32) Clark, T.; Chandrasekhar, J.; Spitznagel, G. W.; Schleyer, P. v. R.
J. Comput. Chem. 1983, 4, 294−301.
(33) Marenich, A. V.; Cramer, C. J.; Truhlar, D. G. J. Phys. Chem. B
2009, 113, 6378−6396.
(34) (a) Wang, C.; Fu, Y.; Guo, Q.-X.; Liu, L. Chem. - Eur. J. 2010,
16, 2586−2598. (b) Zhuo, L.-G.; Liao, W.; Yu, Z.-X. Asian J. Org.
Chem. 2012, 1, 336−345.
(35) (a) Aizman, A.; Contreras, R.; Per
889−895. (b) Per
Theoretical Aspects of Chemical Reactivity; Toro-Labbe,
́
ez, P. Tetrahedron 2005, 61,
ez, P.; Domingo, L. R.; Aizman, A.; Contreras, R. In
A., Ed.;
́
́
Maķ osza, M. Synlett 2016, 27, 2443−2446. (j) Chai, G.-L.; Han, J.-W.;
Theoretical and Computational Chemistry, Vol. 19; Elsevier:
Amsterdam, 2007; Chapter 9, pp 139−201. (c) Chamorro, E.;
Wong, H. N. C. Synthesis 2017, 49, 181−187. (k) Chai, G.-L.; Han, J.-
W.; Wong, H. N. C. J. Org. Chem. 2017, 82, 12647−12654.
(11) Ballester, M. Chem. Rev. 1955, 55, 283−300.
Duque-Norena, M.; Per
́
ez, P. J. Mol. Struct.: THEOCHEM 2009, 896,
ez, P. J. Mol. Struct.:
̃
73−79. (d) Chamorro, E.; Duque-Norena, M.; Per
́
̃
(12) (a) Bachelor, F. W.; Bansal, R. K. J. Org. Chem. 1969, 34, 3600−
THEOCHEM 2009, 901, 145−152.
́
3604. (b) Sipos, G.; Schobel, G.; Sirokman, F. J. Chem. Soc., Perkin
̈
(36) Byrne, P. A.; Kobayashi, S.; Wurthwein, E.-U.; Ammer, J.; Mayr,
̈
Trans. 2 1975, 805−808.
H. J. Am. Chem. Soc. 2017, 139, 1499−1511.
(37) ΔΔG⧧ = 2.303RT[Δ(log k)] = 2.303RTsNΔE.
(13) (a) Corey, E. J.; Chaykovsky, M. J. Am. Chem. Soc. 1965, 87,
1353−1363. (b) Johnson, A. W.; LaCount, R. B. J. Am. Chem. Soc.
1961, 83, 417−423.
(38) (a) Marcus, R. A. J. Chem. Phys. 1956, 24, 966−978. (b) Marcus,
R. A. J. Phys. Chem. 1968, 72, 891−899. (c) Albery, W. J. Annu. Rev.
O
DOI: 10.1021/jacs.8b01657
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Phys. Chem. 1980, 31, 227−263. (d) Murdoch, J. R.; Magnoli, D. E. J.
Am. Chem. Soc. 1982, 104, 3792−3800. (e) Chen, M. Y.; Murdoch, J.
R. J. Am. Chem. Soc. 1984, 106, 4735−4743. (f) Hine, J. Adv. Phys. Org.
Chem. 1977, 15, 1−61.
(39) Carey, F. A., Sundberg, R. J. Advanced Organic Chemistry Part A:
Structure and Mechanisms, 5th ed; Springer: New York, 2007; pp 632−
637.
(40) Brown, H. C.; Borkowski, M. J. Am. Chem. Soc. 1952, 74, 1894−
1902.
(41) Roth, M.; Schade, C.; Mayr, H. J. Org. Chem. 1994, 59, 169−
172.
(42) Hoffmann, R. Acc. Chem. Res. 1971, 4, 1−9.
P
DOI: 10.1021/jacs.8b01657
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX