Toward a Unified Mechanism for Oxoammonium Salt-Mediated Oxidation Reactions: A Theoretical and Experimental Study Using a Hydride Transfer Model

Article

pubs.acs.org/joc

Toward a Uniﬁed Mechanism for Oxoammonium Salt-Mediated

Oxidation Reactions: A Theoretical and Experimental Study Using a

Hydride Transfer Model

Trevor A. Hamlin,^†,∥Christopher B. Kelly,^†,∥John M. Ovian,^†Rebecca J. Wiles,^†Leon J. Tilley,^‡

and Nicholas E. Leadbeater*^,†,§

^†Department of Chemistry, University of Connecticut, 55 North Eagleville Road, Storrs, Connecticut 06269, United States

^‡Department of Chemistry, Stonehill College, 320 Washington Street, Easton, Massachusetts 02357, United States

^§Department of Community Medicine & Health Care, University of Connecticut Health Center, The Exchange, 263 Farmington

Avenue, Farmington, Connecticut 06030, United States

S

* Supporting Information

ABSTRACT: A range of oxoammonium salt-based oxidation

reactions have been explored computationally using density

functional theory (DFT), and the results have been correlated

with experimentally derived trends in reactivity. Mechanistically,

most reactions involve a formal hydride transfer from an activated

C−H bond to the oxygen atom of the oxoammonium cation.

Several new potential modes of reactivity have been uncovered and

validated experimentally.

idine-1-oxoammonium-based salts (TEMPO⁺X⁻) center on

1. INTRODUCTION

the oxidation of alcohols to their corresponding carbonyl

species.²This application is not surprising given the ubiquity

of this type of functional group interconversion in organic

synthesis. Early mechanistic studies by Semmelhack et al.

concluded that the mechanism of this transformation involved

nucleophilic attack on the positive nitrogen of the oxoammo-

nium cation by an alkoxide, followed by an intramolecular E2-like

elimination (see I in Figure 2).⁷This is by far the prevailing

mechanism reported in the literature but seems counterintuitive

given the severe steric constraints of the nitrogen atom in this

scaﬀold.^2,5aIndeed, even in this seminal report, Semmelhack et

al. note that carbon nucleophiles attack solely at the oxygen. The

Oxidation is one of the most important and fundamental

transformations in synthetic organic chemistry.¹Consequently,

many reagents have been developed to accomplish oxidations in

an eﬀective and selective manner. Among known oxidizing

reagents, those based on oxoammonium cations, especially

oxidants bearing the 2,2,6,6-tetramethylpiperidyl scaﬀold (Figure

1), have found wide applicability because of their versatility,

Figure 1. Common 2,2,6,6-tetramethylpiperidine-based nitroxides and

oxoammonium salts.

selectivity, and mildness.^2,3Oxoammonium cations can be

generated in situ by a single electron transfer (SET) to a terminal

oxidant and subsequently employed for catalytic oxidation.

Alternatively, oxidations can be conducted stoichiometrically

using oxoammonium salts.^2,3Both approaches have been

employed in the oxidation of an array of functional groups

including alcohols, amines, and aldehydes.²⁻⁵The catalytic

approach is frequently exploited for industrial scale oxidation.⁶

A majority of reports involving 2,2,6,6-tetramethylpiperidine

1-oxyl based (TEMPO) catalysts or 2,2,6,6-tetramethylpiper-

Figure 2. Possible avenues for oxidation using 2,2,6,6-tetramethylpiper-

idine-based oxoammonium cations. (I) Key species proposed by

Semmelhack et al.⁷(II) Our previously proposed hydride transfer based

model for the base-assisted oxidation of methanol using the TEMPO-

based oxoammonium cation.

Received: June 2, 2015

A

DOI: 10.1021/acs.joc.5b01240

J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Figure 3. Oxoammonium cations and salts as well as related species discussed in this paper.

question, therefore, arises as to why other nucleophiles would

behave diﬀerently.

theory^12,13and (ii) single-point solvation free energy calculation

in CH₂Cl₂or MeCN using CPCM with DFT(B3LYP)//6-311+

+G(d,p). The method used for the calculations of iron

containing systems involves (i) geometry optimization and

vibrational frequency calculations in implicit CH₂Cl₂or MeCN

using CPCM at the DFT(B3LYP/LANL2DZ)¹⁴level of theory

for iron and DFT(B3LYP)/6-31+G(d) level of theory for the

rest of the atoms and (ii) single-point solvation free energy

calculation in CH₂Cl₂or MeCN using CPCM with DFT-

(B3LYP/SDD)¹⁵for iron and DFT(B3LYP)/6-311++G(d,p)

for all other atoms. Gibbs free energies in solution are obtained

by adding the thermal correction to the Gibbs free energy from

(i) to the solvation electronic energy calculated in (ii). Radical

species were treated in an analogous fashion as described above

for closed-shell systems; however, unrestricted calculations were

employed. Stationary points were characterized by frequency

calculations at 298 K, with structures at energy minima showing

no imaginary frequencies and transition-state structures showing

one imaginary frequency. All calculations were carried out at 298

K and 1 atm, and the concentration was 1 M. Intrinsic reaction

coordinate (IRC) calculations followed by optimization and

frequency calculations were performed to unambiguously

connect transition-state structures with associated reactants

and products along the reaction coordinate.¹⁶All energy values

shown are in kcal mol⁻¹, bond lengths are reported in angstroms

(Å), and bond angles in degrees (°). Molecular ﬁgures were

generated using CYLView.¹⁷

Two families of 2,2,6,6-tetramethylpiperidine oxoammonium

ions have been studied computationally in this report. Addition-

ally, two diﬀerent oxoammonium salts that have been used in

experimental studies performed by us and others will be

discussed. For clarity, these various species are presented in

Figure 3, along with their respective compound numbers.

2.1. Oxidations α to Oxygen. 2.1.1. Alcohol Oxidation.

Given that oxoammonium salts are primarily used for the routine

oxidation of alcohols to aldehydes, we initially focused our eﬀorts

on exploring the details of this transformation.^4,5In addition,

these reactions have been the subject of preliminary studies by us

and members of our team.⁸We hypothesized that a fundamental

understanding of this type of oxidation would be instrumental in

elucidating the key features governing other oxidation reactions

α to oxygen.

Recently, members of our team suggested that the primary

mode of operation of these cations is as though the positive

charge is on the oxygen atom of the oxoammonium salt,

eﬀectively giving rise to an electrophilic, hydride-accepting

oxygen.⁸From this standpoint, a mechanism originally ruled out

by Semmelhack becomes far more viable, namely, one involving a

hydride transfer from an activated C−H bond to the electrophilic

oxygen of the oxoammonium cation.

Bailey et al. have recently probed the oxidation of alcohols

under neutral and basic conditions using computational

modeling.⁹In their report, they invoked a hydride transfer

under neutral conditions, whereas, under basic conditions, they

argued for a complex similar to the one proposed by Semmelhack

et al.^7,9Building on this work, members of our group, in

conjunction with Bobbitt et al., further investigated the

mechanism of oxoammonium salt oxidations of methanol

under non-base-assisted and pyridyl-base-assisted conditions.⁸

This study suggested that a hydride transfer model was actually

plausible in both cases. We then wondered whether this

mechanism was at play not only for the oxidation of alcohols

but also for a number of other oxidation reactions mediated by

2,2,6,6-tetramethylpiperidine-based oxoammonium salts. We,

therefore, decided to use computational and experimental

methods to probe a variety of oxidative transformations reported

previously by us and others involving these salts. Additionally, we

sought to further explore the intricacies of the oxidation of

alcohols under unassisted and base-assisted conditions, as well as

use our results as a predictive tool for the feasibility of yet

unreported oxidations. The results of this combined exper-

imental and computational endeavor are reported herein.

2. RESULTS AND DISCUSSION

Our analysis is broken down into three parts based on the

substituents around the C−H bond being oxidized: (1) at C−H

bonds α to oxygen, (2) at C−H bonds α to nitrogen, and (3) at

C−H bonds without any heteroatom assistance (C−H bond

activation). Experimental evidence and computationally derived

models are reported in these sections. Quantum chemical

calculations were performed using Gaussian 09.¹⁰The method

used for the calculations of nonmetal containing systems involves

(i) geometry optimization and vibrational frequency calculations

in implicit dichloromethane (CH₂Cl₂) or acetonitrile (MeCN)

using CPCM¹¹at the DFT(B3LYP)/6-31+G(d) level of

We began by using benzyl alcohol as a representative substrate

to probe the diﬀerences between the unassisted and base-assisted

oxidation. While we had evaluated this in our past report in the

B

DOI: 10.1021/acs.joc.5b01240

J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Figure 4. Disparate mechanistic pathways for the oxidation of benzyl alcohol. The reaction proﬁle on the left represents the unassisted mechanism, while

that on the right is the base-assisted process. Values are in kcal mol⁻¹.

Table 1. Energetics of Alcohol Oxidation in the Unassisted Mechanism for Various Substrates

a

b

Values in kcal mol⁻¹. Experimentally obtained values from competitive studies

theoretical oxidation of methanol, there are two key diﬀerences

in this contribution: (1) we invoke the formation of a dipole-

stabilized preoxidation complex between the oxidant and the

alcohol, and (2) we use 1a′ itself rather than 2a′. The formation

of a dipole-stabilized preoxidation complex is predicated on

recent computational studies by Wiberg and co-workers focused

around the oxidation of amines to nitriles,¹⁸while use of 1a′ itself

improves the overall accuracy of our modeling. Using the results

of this study, we were able to construct the potential energy

surface for both reaction types (Figure 4).

In the case of the non-base-assisted mechanism (Figure 4, left),

we observed a concerted, but asynchronous, oxidation event

where the primary molecular motion in the transition-state

structure is that of the hydride transfer from the alcohol to the

electrophilic oxygen of 1a′, followed by a proton transfer to the

now basic nitrogen of the resulting hydroxylamine 1c, generating

1d and the aldehyde. These two distinct events can be observed

C

DOI: 10.1021/acs.joc.5b01240

J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Table 2. Energetics of Alcohol Oxidation in the Base-Assisted Mechanism for Various Substrates

a

b

c

Values in kcal mol⁻¹. Experimentally obtained values from competitive studies. Not assessed due to propensity of product to undergo oxidative

esteriﬁcation; see ref 8.

when performing intrinsic reaction coordinate (IRC) calcu-

lations (see the SI for IRC plots). The ΔG^⧧value from the

preoxidation complex is reasonable (18.3 kcal mol⁻¹), and the

overall reaction is thermodynamically favorable. The required

energy to reach the preoxidation complex (ΔG_complex) is a

consequence of a more organized state, and the pathway has a

ΔS_complex= −0.030 kcal mol⁻¹K⁻¹.

In the base-assisted mechanism, we presupposed the

formation of a hydrogen-bonding complex (energetically deﬁned

as ΔG_H‑bond) between benzyl alcohol and pyridine based on

experimental evidence garnered in our previous report.⁸We then

assumed the likely reversible formation of a dipole-stabilized

preoxidation complex. Not surprisingly, the overall change in

energy to reach this even more entropically demanding state

(ΔS_complex= −0.037 kcal mol⁻¹K⁻¹) was higher as compared to

the unassisted mechanism (7.8 kcal mol⁻¹for ΔG_complexin the

unassisted mechanism vs 8.9 kcal mol⁻¹for ΔG_complexfrom the

hydrogen-bonded complex). While this mechanism is similarly

highly favorable thermodynamically to the unassisted mecha-

nism, the ΔG^⧧value for the preoxidation complex is substantially

lower (10.1 kcal mol⁻¹). This explains the signiﬁcant rate

acceleration observed for the base-assisted mechanism.⁸

oxidation process, and (iii) the overall nature of the oxidation

(concerted or otherwise). To do this, we modeled the oxidation

of numerous alcohols with varying steric and electronic

environments. The results of these studies are shown in Tables

1 and 2. Also included in these studies are our experimental

relative reactivities of alcohols in oxoammonium salt oxidation.

Using these data, we were able not only to elucidate some

interesting trends but also to provide some commentary on the

limitations of relative-reactivities studies in this type of reaction.

Steric and electronic interactions appear to have a noticeable

eﬀect on the ability to form the requisite dipole-stabilized

complex. However, the eﬀect is somewhat complex. Distant

electron-withdrawing groups (EWGs) or sterically demanding

environments deter complexation (Table 1, entries 4−6, 9, and

10), whereas nearby EWGs appear to favor complexation (Table

1, entries 11−13). We quickly ascertained that, while the

experimentally derived relative reactivities are mostly a function

of the ΔG^⧧of oxidation, the ΔG_complexalso plays a role. Alcohols

with near-identical activation barriers but with disparate energies

of complexation have diﬀering reactivity values (e.g., entries 2

and 7 or entries 1 and 8). In this case, the concentration of the

pre-equilibrium complex could be the controlling factor. The

lower energy complex would be higher in concentration in

solution and thus limit the availability of 1a/1a′ to oxidize any

other competing alcohol. In turn, this would result in a disparity

With these potential energy surfaces in hand, we sought to

evaluate factors governing oxidation in both mechanistic

pathways, namely, (i) the disparity between unassisted and

base-assisted oxidation, (ii) the role of electronics in this

D

DOI: 10.1021/acs.joc.5b01240

J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Figure 5. Comparison of the theoretical and experimental oxidation of α-CH₂F and α-CF₂H alcohols in the base and non-base-assisted mechanisms.

in the experimental relative reactivities (which are obtained from

competition studies) of two alcohols.

namely, α-ﬂuoromethyl (3n) and α-diﬂuoromethyl (3o)

substrates (Figure 5). Speciﬁcally, we examined an aliphatic

ﬂuoromethyl alcohol and a benzyl diﬂuoromethyl alcohol. On

the basis of their respective computed values for ΔG^⧧and

ΔG_complexin the unassisted mechanism, one would expect both

alcohols to fail to oxidize. Experimental results conﬁrmed this

prediction. Also, a tentative cutoﬀ value for ΔG^⧧between

successful and unsuccessful unassisted oxidation at room

temperature was determined by this analysis. This barrier must

lie between 23.4 and 29.6 kcal mol⁻¹(most likely closer to the

lower end of the range given the lower relative reactivity of 3j in

Table 1). We performed a similar calculation using the base-

assisted mechanism. The computed values of ΔG^⧧suggested that

both would be readily oxidized, and this was indeed the case

experimentally. The cutoﬀ for oxidation in the base-assisted

mechanism at room temperature is likely around 15.0−18.4 kcal

mol⁻¹based on these results and those for entries 12 and 13 of

Table 2.

In the non-base-assisted mechanism, ΔG^⧧is most certainly a

function of the stability of the forming oxonium-like ion. This

trend is apparent when comparing the eﬀect EWGs and electron-

donating groups (EDGs) have on alcohols of near-identical steric

environments (Table 1, entries 1−6). Alcohols with signiﬁcant

substitution and those that are more apt to participate in

resonance-stabilization have lower overall activation barriers.

Inductively destabilizing groups can negate the beneﬁts of

substitution and dramatically raise ΔG^⧧. Severely sterically

encumbered alcohols also have higher activation barriers, likely

due to steric repulsion (entries 9−13, Table 1).

Using these same alcohols, we probed the energetic diﬀerences

and trends in the base-assisted mechanism. The results of this

study are shown in Table 2. Two trends become immediately

apparent: (1) The magnitude of ΔG^⧧(from the dipole stabilized

complex) is about half that of the analogous non-base-assisted

ΔG^⧧value, and (2) the energy required to form the preoxidation

complex is very close to that needed for the hydride transfer

event (though the magnitude of each value is independent of one

another). Taken together, this implies that oxidation should

proceed rapidly and the oxidation should be less sensitive to

electronic perturbations. This is indeed the case experimentally.

Oxidations under base-assisted conditions proceed much faster

at room temperature (≈1−2 h on average) as compared to

analogous unassisted oxidations (≈6−48 h).^4,8Additionally, the

relative reactivities reﬂect a relative insensitivity to alcohol

structure.

Subtle trends are also apparent. The value of ΔG_H‑bondappears

to be a function of alcohol acidity. This explains the less uniform

ΔG_complexvalues for benzyl alcohols with varying electronic, yet

identical, steric environments. As in the unassisted mechanism,

ΔG_complexis mostly a function of steric interactions. As noted in

our previous report, the value of ΔG_H‑bondis also a function of the

basicity of the pyridyl base.⁸Since this mechanism involves a

more sterically (and entropically) demanding complex, the

values for complexation are, on a whole, larger than in the

unassisted mechanism. While the relative reactivities are

2.1.2. Oxidation of Ethers. In 2009, Bailey and co-workers

reported that benzyl ethers could be oxidatively cleaved under

mild conditions in the presence of 1a (Figure 6).¹⁹In the original

report by these authors, the posited mechanism hinged on a

formal hydride transfer.

Figure 6. Example of the oxidative cleavage of benzyl ethers reported by

Bailey and co-workers.¹⁹

To continue our investigation of oxidations at alkyl groups α to

oxygen, we explored the theoretical oxidation of methyl benzyl

ether (5a) computationally, using our previous ﬁndings with

alcohols as a guide (Figure 7). However, unlike the case of the

alcohols, a preoxidation complex was not preferable given the

new steric constraints of the methyl group attached to the oxygen

(conﬁrmed by IRC calculations). Thus, we modeled the

mechanism without the formation of such a complex.

Interestingly, the ΔG^⧧was markedly lower for 5a as compared

to benzyl alcohol (3a). While this may be due to electronic

diﬀerences between the oxygen atom in the two cases, it more

likely results from modeling the reaction in a more polar solvent

(CH₂Cl₂for benzyl alcohol vs MeCN for benzyl ether).

However, given that these oxidative cleavage processes mediated

inﬂuenced by the respective values of ΔG_H‑bondand ΔG_complex

,

the overriding measure of reaction success is ΔG^⧧from the

dipole-stabilized complex.

As a means to use the data collected in a predictive manner, we

computed the energies of oxidation of two classes of alcohols that

have not been subjected to oxoammonium salt oxidations,

E

DOI: 10.1021/acs.joc.5b01240

J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Figure 7. Computed reaction proﬁle for the oxidation of methyl benzyl ether to benzaldehyde mediated by 1a′.

Table 3. Energetics of Benzyl Ether Cleavage Facilitated by 1a′

a

b

Values in kcal mol⁻¹. Deﬁned as the diﬀerence in ΔG_intbetween the current entry and entry 1.

F

DOI: 10.1021/acs.joc.5b01240

J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

by 1a are experimentally conducted in acetonitrile, we felt that

that using MeCN as the implicit solvent for our solvation model

would be more accurate in describing the energetics of this

reaction. We were unable to ﬁnd transition-state structures

leading to the hydrolysis of the intermediate oxonium ion, but

were able to evaluate both the ΔG_intleading to the intermediate

oxonium ion as well as the overall ΔG_rxn. Both ΔG values

indicated the processes were favorable and spontaneous.

In addition to modeling this speciﬁc reaction, we sought to

both reproduce the observed experimental trends and explore

other subtle details of this process. To do this, we calculated both

the ΔG^⧧and the ΔG_intfor a variety of benzyl ether substrates. We

wanted to evaluate the possibility that the hydride transfer could

be reversible, given that, unlike alcohols, there is no readily

accessible way to neutralize the charge of the intermediate

oxonium ion and this ion itself may serve as an oxidant. Our

results are displayed in Table 3.

surprised to ﬁnd that the oxidation not only was favorable from

the standpoint of ΔG_intbut also had a lower ΔG^⧧than the

oxidation of 5a. These diﬀerences may be a consequence of the

disparate steric environments between the two substrates and the

extended conjugation in the cinnamyl compound. To use this

result in a predictive manner, we prepared this compound and

subjected it to oxidative cleavage using literature conditions for

the cleavage of benzyl ethers. We were pleased to ﬁnd that this

compound could be oxidatively cleaved in good yield (Figure 8).

To test the validity of this hydride transfer model further, we

next explored two disparately diﬀerent substrates: an allylic ester,

7h, and a nonconjugated allylic ether, 7c. While both had higher

values for ΔG^⧧, only the acetyl ester had an unfavorable ΔG_intto

the oxonium intermediate, much like the phenyl ether 5e.

Attempts to oxidize this ester experimentally failed, while

oxidation of the nonconjugated allyl ether 7c succeeded, albeit

needing an extended reaction time (Figure 9).

More information can be garnered from relative trends from a

reference point than from the actual numerical values of the

oxidation. This is especially true given the fact that these values

are not from a preoxidation complex, but rather from the two

isolated species and thus may have artiﬁcially high activation

barriers. To determine relative trends, we selected benzyl methyl

ether (5a) as our reference point and studied diﬀerences in both

the activation energies for the reaction (ΔG^⧧) and the ΔG_intto

the oxonium intermediate (5-I). We were quickly able to discern

some trends. While the substitution pattern of the ethereal alkyl

group has a notable eﬀect, its overall chain length is irrelevant

(compare ΔG^⧧for Table 3, entries 1 and 2). Steric congestion on

the oxygen systematically increases the ΔG^⧧but lowers the

overall ΔG_int, thus simply having an eﬀect on reaction rate but

not reaction success (compare entries 1−4).

Figure 9. Computational and experimental study on the oxidation of 7h

and 7c.

The exception is when both the ΔG^⧧and the ΔG_intare

unfavorable (entry 5, Table 3). This explains the observed

limited oxidation of phenyl ethers when performing oxidations

mediated by 1a′.²⁰Similarly, electronic changes have a noticeable

eﬀect on both ΔG^⧧and ΔG_int. As may be expected, EDGs

lowered ΔG^⧧, whereas EWGs had the opposite eﬀect. However,

the trends in ΔG_intare more revealing. Despite having a lower

ΔG^⧧than 5d, oxidation of 5h is highly unfavorable, thus

explaining the disparity in observed experimental reaction

success in these two cases.¹⁹Additionally, the trends between

ethers resulting in ketones (Table 3, entries 9 and 10) follow

observed experimental trends in which 5i succeeded while 5j

failed. This further evidences the idea that the crux of this

reaction is the overall ΔG_intto 5-I.

On the basis of these encouraging results, we next modeled a

series of allyl and nonallyl ethers to extract the factors governing

ether oxidation (Table 4). As a reference point, we chose 7d as it

represents a middle ground between extensive alkene stabiliza-

tion and a complete lack of stabilization. Several trends, which

correlate with the benzyl ether counterparts, quickly became

apparent. Lack of any oleﬁnic stabilization dramatically increases

both ΔG^⧧and ΔG_int, suggesting that oxidation of aliphatic ethers

should fail (Table 4, entry 1). The inability of these compounds

to undergo oxidation has been conﬁrmed by us and others

experimentally.^4,19Not surprisingly, increasing the substitution

pattern and/or extending the conjugation both lowers ΔG^⧧and

stabilizes the intermediate oxonium ion, thus making the process

less reversible (compare entries 2 and 3). When the alkene is

conjugated, changes to the overall electronics of the π system

(even at remote locations within the conjugated system)

signiﬁcantly impacts ΔG^⧧and ΔG_int(entries 5 and 6). Given

the results in Table 4, we concluded that a broad range of allyl

ethers could likely be oxidized. We pursued this reaction and

developed it into a general methodology, which we recently

disclosed.²¹

Intrigued by our ﬁndings with benzyl ethers, we next examined

whether allyl ethers could be oxidized. Initially, we examined the

theoretical oxidation of 7d to its oxonium ion (7d-I, Figure 8),

without the formation of a dipole-stabilized complex. We were

2.1.3. Oxidative Functionalization of Isochromane. In 2010,

Richter and Garcia-Mancheno disclosed that C(sp³)−H bonds

̃

adjacent to a heteroatom could be functionalized under mild

conditions in the presence of stoichiometric amounts of

22

TEMPO⁺BF₄, 2a (Figure 10). This dehydrogenative

functionalization reaction employs a metal catalyst, Fe(OTf)₂,

and an enolizable species to accomplish alkylation of the

intermediate oxonium ion.

−

Figure 8. Computational and experimental study on the oxidation of 7d.

G

DOI: 10.1021/acs.joc.5b01240

J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Table 4. Energetics of Allyl and Aliphatic Ether Cleavage Facilitated by 1a′

a

b

c

Values in kcal mol⁻¹. Deﬁned as the diﬀerence in ΔG^⧧between the current entry and entry 3. Deﬁned as the diﬀerence in ΔG_intbetween the

current entry and entry 3.

Figure 10. Oxidative functionalization of isochromane mediated by 2a reported by Richter and Garcia-Mancheno.²²

̃

The key step in the reaction sequence is the formation of the

oxonium intermediate (9-I). The authors proposed two putative

pathways to this species, one radical and one ionic. Their

mechanistic studies were inconclusive, and therefore, the

operative mechanism remains in question. To continue our

study of oxidations at alkyl groups α to oxygen, we explored the

theoretical oxidative functionalization of isochromane with a

diester. The energetics of both the radical and the ionic pathways

were calculated, and the respective potential energy surfaces are

shown in Figure 11. In the radical pathway, a high-energy single

electron transfer (SET) precedes a low barrier hydrogen atom

transfer (9′-TS1, red line) while the ionic pathway involves a

hydride transfer (9-TS1, green line) in the transition-state

structure. The latter was found to be favored by 8.5 kcal mol⁻¹

compared to the radical pathway. Thus, the likely pathway is

ionic.

simpliﬁed by assuming complete dissociation of the triﬂate

ligands from Fe(OTf)₂. Multiple conformations of the transition-

state structure were computed, with the most energetically

favorable structure being that shown in Figure 11. The overall

reaction scheme is exergonic, and the theoretical results are in

agreement with a reaction that readily proceeds at room

temperature.

2.2. Oxidations α to Nitrogen. 2.2.1. Oxidation of Imines

and Amines. The oxidation of amines to nitriles using 1a has

recently been reported by Bailey and co-workers.¹⁸The reaction

is thought to proceed through an imine intermediate which

forms following a hydride transfer and a deprotonation.

Subsequently, this imine is rapidly oxidized to the nitrile by a

second hydride transfer event with a concerted deprotonation.

Using ethylamine (11) as a model substrate, the energetics of

oxidation were explored computationally in the context of a

hydride transfer model, and the results supported their putative

mechanism. This sequence is shown in Figure 12. The activation

barriers for each hydride transfer event were reasonable (15.8

kcal mol⁻¹). Not surprisingly, the activation barrier for a hydride

transfer in aliphatic α-nitrogen substituted cases is markedly

The formation of the cyclic ether cationic intermediate (9-I) is

favorable by −7.0 kcal mol⁻¹. This oxonium intermediate then

reacts with an iron−enolate complex and has a reasonable

activation barrier of 20.7 kcal mol⁻¹from the ion and an overall

favorable ΔG_rxn. The alkylation transition-state structure was

H

DOI: 10.1021/acs.joc.5b01240

J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Figure 11. Computed energetic proﬁle for the SET pathway and ionic pathway for the oxidative functionalization of 9.

Figure 12. Theoretical oxidation of ethylamine to acetonitrile reported by Bailey and co-workers.¹⁸

lower than that in analogous α-oxygen examples, thus supporting

the concept that the hydride transfer process is a function of the

stability of the forming cationic species. The oxidation to the

imine diﬀered in its ΔG_rxnas compared to similar oxidations of

with the ﬂuoride anion) is 2.4 times more exergonic (−14 kcal

mol⁻¹reported by Bailey and Wiberg vs ≈−34 kcal mol⁻¹

reported here). The observed energetic disparities between the

two reactions account for the experimental observations.

2.2.2. Amine Oxidation and Functionalization. In addition

to exploring oxidative functionalization at the α position to

alcohols to aldehydes (ΔG_imine: −1.8 kcal mol⁻¹vs ΔG_aldehyde

:

≈−20 kcal mol⁻¹), but this is to be expected given the disparity in

bond strength between CN and CO bonds.

oxygen, Richter and Garcia-Mancheno have also reported that

̃

We also recently reported on a method to convert aldehydes

and alcohols to nitriles using hexamethyldisilazane (HMDS) and

1a.²³Mechanistic studies conﬁrmed the intermediacy of an N-

silyl imine (13). Interestingly, this reaction was highly

exothermic but required at least room-temperature conditions

to initiate. The reaction proceeded rapidly once initiated and

resulted in the formation of Me₃Si-F as a byproduct of oxidation.

While conﬁdent that this oxidation proceeded similarly by a

hydride transfer mechanism, we wanted to contrast the

energetics of this oxidation with that report by Bailey and co-

workers.¹⁸

C(sp³)−H bonds α to a nitrogen atom can also be functionalized

under mild conditions in the presence of stoichiometric amounts

of 2a and an iron catalyst (Figure 14).²²

We explored the theoretical oxidation and functionalization of

1,2,3,4-tetrahydroisoquinoline (15a) using 2a′ (Figure 15). As in

the case of isochromane (9), both the radical and the ionic

pathways were analyzed (Figure 15). The SET preceding the

hydrogen atom transfer was slightly more preferable (0.7 kcal

mol⁻¹) than an ionic hydride transfer. However, taken together

with the hydrogen atom transfer step (15a′-TS1), the hydride

transfer pathway (15a-TS1) is kinetically more favorable.

Ultimately, both pathways are exergonic and lead to the cyclic

iminium ion intermediate (15a-I). The activation for the hydride

transfer step for this amine substrate is lower than its ether

counterpart by 9.9 kcal mol⁻¹. This is expected given that

nitrogen species are much more easily oxidized by 2a′ than the

The results can be seen in Figure 13. In comparison to the

analogous oxidation from an imine to a nitrilium ion, the

activation barrier for our more sterically congested N-silyl imine

is higher (>5 kcal mol⁻¹). However, the overall ΔG_rxnof the

process (including desilylation of the intermediate nitrilium ion

I

DOI: 10.1021/acs.joc.5b01240

J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Figure 13. Computed reaction proﬁle for the oxidation of N-silyl imines to nitriles.

Figure 14. Oxidative functionalization of 1,2,3,4-tetrahydroisoquinolines using 2a reported by Richter and Garcia-Mancheno.²²

̃

oxygen containing analogues, likely due to enhanced ability of the

nitrogen lone pair to participate in the hydride transfer event.

This also matches the observed trend when comparing alcohol

and amine oxidations. The reaction leading to the amine

intermediate is very energetically favorable and seems to form a

cation sink. The reaction of the iminium ion (15a-I) with the

iron−enolate complex has a relatively high activation barrier of

31.4 kcal mol⁻¹(15a-TS2) from this energy well. The alkylation

of dimethyl malonate is endergonic by 8.7 kcal mol⁻¹relative to

the iminium ion (which is likely an artifact of the coordination to

the iron), but overall exergonic from the starting material.

Unlike isochromane, we can directly inﬂuence the electronic

nature of the heteroatom in tetrahydroisoquinoline by

functionalizing it. We elected to explore one of the substrates

energetics of oxidation. When conjugated into the amide

functionality, we see a larger disparity between radical (15b′-

TS1) and ionic pathways (15b-TS1). In fact, in this case, the

ionic pathway is not only very clearly favored by 7.7 kcal mol⁻¹

when combining the SET and hydrogen atom steps but also

preferred over just the SET step alone by 3.1 kcal mol⁻¹. An

additional rationale for the overall endergonic nature of this

particular reactant is needed. We suggest that, unlike its

nonacetylated counterpart, (and the isochromane discussed

earlier), it has enhanced steric constraints in its coordinated state.

This strain would be relieved upon decomplexation. This lower

energy state would prevent the intermediate cation from

becoming a thermodynamic sink. Additionally, the irreversibility

of forming this C−C bond likely drives this process.

experimentally employed by Richter and Garcia-Mancheno,

2.3. Oxidations of Activated C−H Bonds. 2.3.1. Oxida-

tion of Cycloheptatriene. Cycloheptatriene, 17, contains a

methylene group that is particularly reactive toward oxidation

̃

namely, an N-acetyl-protected 1,2,3,4-tetrahydroisoquinoline

(15b).²²Using this scaﬀold, we probed the changes in the

J

DOI: 10.1021/acs.joc.5b01240

J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Figure 15. Computed reaction proﬁles for the theoretical oxidative functionalization of 1,2,3,4-tetrahydroisoquinoline (15a, left) and its acylated

congener (15b, right) mediated by 2a′.

Figure 16. Energetics of the theoretical oxidative functionalization of cycloheptatriene mediated by 1a′.

due to the aromatic tropylium ion that results upon hydride

abstraction.²⁴This ion can itself be used as an oxidant of certain

species such as amines.²⁵Given the dual role of the tropylium ion

(17-I) as both a hydride acceptor (an oxidizing agent remarkably

similar to the oxoammonium cation) and a hydride donor, some

ambiguity exists as to whether an oxidation of 17 by an

oxoammonium salt would be possible. Electrochemical studies

using 2b suggested that it may be plausible using 1a.²⁶We

modeled the oxidation involving a putative hydride transfer event

(Figure 16, 17-TS1) using the parameters for acetonitrile in our

solvation model. The ΔG^⧧for the hydride transfer was quite

modest, and the ΔG_cationwas highly favorable. We next modeled

an alkylation reaction (17-TS2) with the free enolate of methyl

acetoacetate. The activation barrier for this event was rather high

from the tropylium ion (likely due to dearomatization), but the

overall reaction was exergonic. On the basis of this potential

energy surface, we speculated that oxidative functionalization of

17 should proceed readily. We, therefore, attempted this net

process experimentally. Oxidation did indeed proceed rapidly

and was complete in less than 1 h, as conﬁrmed by the presence

1

of the characteristic H NMR signal of the tropylium cation at

9.36 ppm. Unfortunately, attempts to isolate tropylium

tetraﬂuoroborate proved problematic due to the near identical

solubility properties of 1c and this tropylium salt. Rather than

isolate the ion, we proceeded to functionalize it in a manner

similar to our theoretical oxidation. Conrow reported that

K

DOI: 10.1021/acs.joc.5b01240

J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

tropylium tetraﬂuoroborate can be eﬀectively functionalized in

the presence of acetylacetone and pyridine.²⁷Using this report

and our computations as a guide, we conducted the sequence

outlined in Figure 18. After a short optimization of workup

conditions based on the relative solubilities of 1c and 18, we were

able to isolate the functionalized cycloheptatriene in good yield

(Figure 17). This experimental result is the strongest support to

date for the hydride transfer model in oxoammonium salt

oxidations to date.

proceeds through a concerted ene-like mechanism (Figure 19).²⁷

The nucleophilic addition was said to occur in a simultaneous

manner with removal of the allylic proton and C−O bond

formation. To determine the operative pathway, we explored the

energetics of both the literature proposed route as well as our

putative hydride transfer mechanism. When modeling the former

(Figure 20), we found the process to be concerted as previously

posited. The activation barrier for the ene-like process (19-TS)

was found to be 28.4 kcal mol⁻¹for 2-methyl-2-butene as a

substrate. Comparatively, the rate-determining hydride transfer

step in the theoretical stepwise pathway (19-TS′) has an

activation barrier of 29.7 kcal mol⁻¹. The ΔΔG^⧧between the

hydride transfer and the ene-like pathway is modest (1.3 kcal

mol⁻¹). Each respective transition-state structure is shown in

Figure 21. Unlike the oxidation of the tropylium ion, the ΔG_cation

is highly unfavorable when proceeding via the stepwise pathway.

Thus, the resulting cation is a stronger oxidant (i.e., a better

hydride acceptor) than the oxoammonium cation. Therefore,

just as in the oxidation of benzyl ethers, a thermodynamically

unfavorable hydride transfer results in the inability to proceed

forward in the reaction, and although kinetically equal, the

unfavorable ΔG_intprecludes a hydride transfer mechanism in this

reaction.

Figure 17. Experimental oxidative functionalization of 17 mediated by

1a.

2.3.2. Allylic Oxidation. On the basis of the results with 17, we

wondered whether the reported facile reaction of an

oxoammonium species with a trisubstituted alkene reported²⁸

by Bailey and co-workers also proceeded by a hydride transfer

pathway. The reaction leads to the formation of allylic

alkoxyamines (Figure 18). Bailey suggested that this reaction

2.3.3. Oxidation of Other Activated C−H Bonds. Given the

results on the theoretical oxidation of reactive C−H bonds in the

preceding sections, we became interested in the factors

inﬂuencing oxidation of C−H bonds using the putative hydride

transfer model. We, therefore, modeled a range of compounds

possessing such C−H bonds with various steric and electronic

environments. The computed results are in Table 5.

In general, the determining factor for oxidation of such bonds

by 1a′ is not kinetic in origin but thermodynamic, supporting our

earlier assertions in the case of the oxidation of allyl and benzyl

ethers. Oxidations leading to cyclopropenium ions (Table 5,

entries 1−5) are highly favorable kinetically and thermodynami-

cally, though not to the same degree as 17. Electronic changes do

have an eﬀect on both energetic parameters with a magnitude

similar to those of allyl and benzyl ethers. For example, the

electron-donating p-OMe group (entry 2) both stabilizes the

corresponding cyclopropenium ion thermodynamically and

lowers ΔG^⧧, making the hydride transfer more preferable

kinetically. Steric eﬀects in this case are only noticeable when

adding groups at the incipient cationic center (compare ΔG^⧧of

entries 1, 4, and 5). Oxidations resulting in stabilized, but

nonaromatic, ions face steeper challenges. Unless highly

stabilized, such as in the case of the trityl group (entry 6),

oxidation is an endergonic process. Thus, oxidation in these

systems is likely to be diﬃcult, if not impossible. Indeed, to date,

we have not observed oxidation experimentally for toluene,

diphenylmethane, or norbornadiene (entries 7, 8, and 9).

Similarly, oxidation of triphenylmethane has not met with

Figure 18. Ene-like addition of the 1a′ to alkenes.

Figure 19. Ene-like mechanism proposed by Bailey and co-workers²⁸

leading to allylic alkoxyamines.

Figure 20. Ene-like mechanism proposed by Bailey and co-workers²⁸leading to allylic alkoxyamines.

L

DOI: 10.1021/acs.joc.5b01240

J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Figure 21. Transition-state structures for the ene-like reaction, stepwise hydride transfer TS (left, 19a-TS), concerted cyclic TS (right, 19a′-TS).

a

Table 5. Energetics of the Theoretical Oxidation of Activated C−H Bonds

a

All values are in kcal mol⁻¹.

experimental success by us to date. This latter observation likely

relates to the high value of ΔG^⧧for oxidation rather than the

thermodynamics of the reaction. A high activation barrier in this

case is likely a consequence of the sterically hindered

environment surrounding the C−H bond and the need to

form a planar geometry upon carbocation formation. In fact, the

failure of this oxidation is the exact evidence used by Semmelhack

et al. to discount a hydride transfer mechanism for oxidations

involving oxoammonium cations. However, the hydride transfer

model we suggest explains both the failure of triphenylmethane

and the success of cycloheptatriene to oxidize.²⁹

tributions to the energetics (and ultimately the reactivity) in

oxidations. Several reactions discussed here, speciﬁcally the

oxidation of cycloheptatriene, lend credence to a hydride transfer

mechanism being operative. Such a ﬁnding supports the notion

that the hydride transfer mechanism originally discounted by

Semmelhack et al. is a valid alternative to the currently accepted

mechanism for oxoammonium cation-mediated oxidations.

4. EXPERIMENTAL SECTION

General Considerations. All chemical transformations requiring

inert atmospheric conditions or vacuum distillation utilized Schlenk line

techniques with a 3- or 4-port dual-bank manifold. Nitrogen was used to

provide such an atmosphere. NMR spectra (¹H, ¹³C, ¹⁹F) were

performed at 298 K on either a 300, 400, or 500 MHz NMR. ¹H NMR

spectra obtained in CDCl₃were referenced to residual nondeuterated

chloroform (7.26 ppm) in the deuterated solvent. ¹³C NMR spectra

obtained in CDCl₃were referenced to CDCl₃(77.3 ppm). ¹⁹F NMR

spectra were referenced to hexaﬂuorobenzene (−164.9 ppm).³⁰IR

spectra were obtained on an FT-IR spectrometer. Reactions were

monitored by a gas chromatograph attached to a mass spectrometer, ¹H

NMR, and/or by TLC on silica gel plates (60 Å porosity, 250 μm

thickness). TLC analysis was performed using hexanes/ethyl acetate as

the eluent and visualized using permanganate stain, p-anisaldehyde stain,

Seebach’s Stain, and/or UV light. Flash chromatography and silica plugs

3. CONCLUSION

In closing, we have thoroughly explored the validity of a hydride

transfer mechanism for oxoammonium cations in an array of

diﬀerent oxidative reactions. A hydride transfer-based model

appears to be a viable mechanistic path for oxidations at positions

α to oxygen and nitrogen as well as in activated (prearomatic)

compounds. This hydride transfer model explains experimental

trends and can be used to predict reactivity during oxidations.

Electronic inﬂuences by groups neighboring the hydridic

hydrogen as well as steric constraints given the bulky nature of

the 2,2,6,6-tetramethylpiperidyl scaﬀold have signiﬁcant con-

M

DOI: 10.1021/acs.joc.5b01240

J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

utilized Dynamic Adsorbents Inc. Flash Silica Gel (60 Å porosity, 32−63

μm).

([M]⁺,8%) 150 (20%) 117 (76%) 115 (16%) 105 (16%) 91 (100%) 77

(23%) 65 (25%) 63 (14%) 51 (13%) 43 (11%) 369 (12%).

Chemicals. CDCl₃was stored over 4 Å molecular sieves. Alcohols

used for relative rate studies were purchased from commercial suppliers.

The oxoammonium salt 4-acetamido-2,2,6,6-tetramethyl-1-oxopiper-

idin-1-ium tetraﬂuoroborate, 1a, was prepared according to our recently

published protocol.^4b

2,2-Diﬂuoro-1-(p-tolyl)ethanone³³(4o). To a 100 mL round-

bottom ﬂask were added crushed magnesium turnings (2.92 g, 0.120

mol, 1.2 equiv) and a stir bar. The ﬂask was sealed with a rubber septum,

the atmosphere was evacuated from the ﬂask via an inlet needle, and the

ﬂask was ﬂame-dried under vacuum. The ﬂask was ﬂushed with nitrogen

and placed in a room-temperature water bath. p-Bromotoluene (17.16 g,

0.100 mol, 1 equiv) was dissolved in a minimal amount of anhydrous

Et₂O (45 mL) and added dropwise to the ﬂask Caution! Exothermic.

After complete addition, the solution was allowed to stir at room

temperature for 1 h and the mixture gradually became dark gray.

To a separate ﬂame-dried 250 mL round-bottom ﬂask equipped with

a stir bar, rubber septum, and N₂inlet needle were added ethyl

diﬂuoroacetate (15.63 g, 0.110 mol, 1.1 equiv) and anhydrous Et₂O (61

mL). The ﬂask was cooled to −78 °C via a dry ice/acetone bath and

stirred at this temperature for 5 min. After this time, the Grignard

solution was transferred to a syringe and added to this chilled ﬂask over

15 min. The solution was allowed to stir at −78 °C for 3 h. After this

time, the solution was warmed to room temperature and was quenched

with saturated NH₄Cl. The quenched mixture was warmed to 0 °C and a

0.5 M aqueous H₂SO₄solution (75 mL) was added. The biphasic

mixture was transferred to a separatory funnel and diluted with pentane

(100 mL) and deionized water (100 mL). The phases were separated,

and the aqueous layer was extracted with pentane (3 × 50 mL). The

combined organic layers were washed with brine and dried with Na₂SO₄.

The solvent was removed in vacuo by rotary evaporation. The crude

ketone was puriﬁed by recrystallization (hexanes) to give the pure CF₂H

ketone (8.92 g, 52%) as an oﬀ-white, crystalline solid (mp 48−50 °C) .

¹H NMR (CDCl₃, 400 MHz) δ ppm 2.45 (s, 3 H) 6.27 (t, J = 53.62 Hz, 1

H) 7.33 (dd, J = 8.56, 0.58 Hz, 2 H) 7.98 (d, J = 8.17 Hz, 2 H); ¹³C NMR

(CDCl₃, 100 MHz) δ ppm 22.2 (CH₃) 111.5 (t, J_C‑F= 253.5 Hz, CF₂H)

129.4 (t, J_{C‑C‑C‑F}= 1.5 Hz, C) 130.0 (CH) 130.0 (t, J_{C‑C‑C‑C‑F}= 2.3 Hz,

CH) 146.5 (C) 187.5 (t, J_C‑C‑F= 25.4 Hz, C); ¹⁹F NMR (CDCl₃, 377

MHz) δ ppm −173.13 (d, J = 53.13 Hz); GC−MS (EI) 170 ([M]⁺, 4%),

119 (100%), 91 (82%), 89 (14%), 65 (32%), 63 (15%), 51 (17%), 39

(11%).

Synthesis of α-Fluoro and α-Diﬂuoro Carbinols via Their

Respective Ketones. 1-Fluoro-4-phenylbutan-2-one³¹(4n). The

following is a modiﬁcation of the procedure outlined by Alexakis and co-

workers.³¹To a 50 mL round-bottom ﬂask were added crushed

magnesium turnings (1.23 g, 50.4 mmol, 1.44 equiv) and a stir bar. The

ﬂask was sealed with a rubber septum, the atmosphere was evacuated

from the ﬂask via an inlet needle, and the ﬂask was ﬂame-dried under

vacuum. The ﬂask was cooled by ﬂushing with nitrogen and then placed

in a room-temperature water bath. (2-Bromoethyl)benzene (7.77 g, 42

mmol, 1.2 equiv) was dissolved in a minimal amount of anhydrous Et₂O

(16 mL) and added dropwise to the ﬂask. Caution! Exothermic. After

complete addition, the solution was allowed to stir at room temperature

for 1 h and the mixture gradually became dark gray.

In a separate ﬂame-dried 500 mL round-bottom ﬂask equipped with a

stir bar, rubber septum, and N₂inlet needle were added ﬂuoroacetoni-

trile (2.07 g, 35 mmol, 1 equiv) and anhydrous Et₂O (210 mL, 0.166 M

in the nitrile). The ﬂask was cooled to 0 °C via an ice−water bath and

stirred at this temperature for 5 min. After this time, the Grignard

solution was transferred to a syringe and added to this chilled ﬂask over

the course of 15 min. The solution was allowed to stir at 0 °C for 2 h,

whereupon a white precipitate formed. After this time, the solution was

quenched with saturated 10% aqueous HCl. The biphasic mixture was

transferred to a separatory funnel and diluted with Et₂O (100 mL) and

deionized water (100 mL). The phases were separated, and the aqueous

layer was extracted with Et₂O (3 × 100 mL). The combined organic

layers were washed with deionized water (≈200 mL) and brine (≈200

mL) and dried with Na₂SO₄. The solvent was removed in vacuo by rotary

evaporation. The crude ketone was puriﬁed by vacuum distillation (53−

55 °C @ 0.1 mmHg) to give the pure CFH₂ketone (4.18 g, 72%) as a

clear, colorless oil. ¹H NMR (CDCl₃, 400 MHz) δ ppm 2.87−2.94 (m, 2

H) 2.99 (apparent triplet, J = 7.70 Hz, 2 H) 4.79 (d, J = 47.59 Hz, 1 H)

7.25 (apparent triplet, J = 7.50 Hz, 3 H) 7.33 (apparent triplet J = 7.30 Hz,

2 H); ¹³C NMR (CDCl₃, 100 MHz) δ ppm 28.9 (d, J_{C‑C‑C‑C‑F}= 1.8 Hz,

CH₂) 40.1 (CH₂) 85.3 (d, J_C‑F= 184.9 Hz, CH₂) 126.6 (CH) 128.55

(CH) 128.8 (CH) 140.7 (C) 206.3 (d, J_C‑C‑F= 19.4 Hz, C); ¹⁹F NMR

(CDCl₃, 377 MHz) δ ppm −230.79 (t, J = 47.70 Hz); GC−MS (EI) 166

([M]⁺, 43%) 133 (44%) 115 (8%) 105 (82%) 91 (100%) 77 (28%) 65

(16%) 61 (6%) 51 (14%).

2,2-Diﬂuoro-1-(p-tolyl)ethanol³⁴(3o). To a 100 mL round-bottom

ﬂask equipped with a stir bar was added the α-diﬂuoro ketone 4o (4.00 g,

0.0235 mol) dissolved in methanol (47 mL). The ﬂask was cooled to 0

°C via an ice−water bath. After stirring at 0 °C for 10 min, NaBH₄(1.79

g, 0.0470 mol, 2 equiv) was added slowly portionwise to the ﬂask.

Caution! Exothermic, evolves H₂gas. The reaction mixture was allowed to

stir at 0 °C for 5 min after complete addition of NaBH₄. The ﬂask was

then warmed to room temperature and allowed to stir for 2 h. After this

time, the reaction was carefully quenched with deionized H₂O (30 mL).

The quenched reaction mixture was transferred to a separatory funnel

and diluted with deionized water (100 mL) and Et₂O (150 mL). The

layers were separated, and the aqueous layer was extracted with Et₂O (3

× 75 mL). The combined organic layers were washed with deionized

H₂O (2 × 100 mL) and brine (150 mL). The organic layer was dried

with Na₂SO₄, and the solvent was removed in vacuo by rotary

evaporation to give the crude alcohol. Further puriﬁcation was

accomplished by vacuum distillation (56−58 °C @ 0.1 mmHg) to

1-Fluoro-4-phenylbutan-2-ol³²(3n). To a 100 mL round-bottom

ﬂask equipped with a stir bar was added the α-ﬂuoro ketone 4n (3.18 g,

0.0191 mol) dissolved in methanol (40 mL). The ﬂask was cooled to 0

°C via an ice−water bath. After stirring at 0 °C for 10 min, NaBH₄(1.46

g, 0.0386 mol, 2 equiv) was added slowly portionwise to the ﬂask.