Mechanism of Electrochemical Generation and Decomposition of Phthalimide-N-oxyl

Article abstract of DOI:10.1021/jacs.1c04181

PhthalimideN-oxyl (PINO) is a potent hydrogen atom transfer (HAT) catalyst that can be generated electrochemically fromN-hydroxyphthalimide (NHPI). However, catalyst decomposition has limited its application. This paper details mechanistic studies of the

Full text of DOI:10.1021/jacs.1c04181

pubs.acs.org/JACS
Article
Mechanism of Electrochemical Generation and Decomposition of
Phthalimide-N‑oxyl
Cheng Yang, Luke A. Farmer, Derek A. Pratt,* Stephen Maldonado,* and Corey R. J. Stephenson*
Cite This: J. Am. Chem. Soc. 2021, 143, 10324−10332
Read Online
sı
*
Metrics & More
Article Recommendations
Supporting Information
ACCESS
ABSTRACT: Phthalimide N-oxyl (PINO) is a potent hydrogen
atom transfer (HAT) catalyst that can be generated electrochemically
from N-hydroxyphthalimide (NHPI). However, catalyst decom-
position has limited its application. This paper details mechanistic
studies of the generation and decomposition of PINO under
electrochemical conditions. Voltammetric data, observations from
bulk electrolysis, and computational studies suggest two primary
aspects. First, base-promoted formation of PINO from NHPI occurs
via multiple-site concerted proton−electron transfer (MS-CPET). Second, PINO decomposition occurs by at least two second-order
paths, one of which is greatly enhanced by base. Optimal catalytic efficiency in PINO-catalyzed oxidations occurs in the presence of
bases whose corresponding conjugate acids have pKa’s in the range of ∼11−15, which strikes a balance between promoting PINO
formation and minimizing its decay.
INTRODUCTION
■
Electrocatalysts that promote anodic half reactions have been
specifically developed for oxidative organic transformations
due to longstanding industrial interest.1 The use of oxidative
electrocatalysts, shuttling electrons between molecular sub-
strates and anode interfaces, have realized chemical trans-
formations in an environmentally friendly manner, as electrons
are collected by electrode interfaces instead of stoichiometric
chemical oxidants.2 For example, the electrochemically
generated N-oxoammonium ion derived from 2,2,6,6-tetrame-
thylpiperidine-N-oxyl (TEMPO) serves as an active catalytic
species and readily reacts with alcohols to produce carbonyls.3
Both TEMPO and TEMPO+ salts are isolable, which facilitates
mechanistically guided development of this and related redox
mediators for selective and efficient oxidations.3
PINO is another promising electrocatalyst generated by the
facile oxidation of its precursor NHPI (Figure 1A).3 PINO can
abstract allylic4 and benzylic5−7 hydrogen atoms, as well as the
other relatively weak C−H bonds,8 making it a potential HAT
electrocatalyst. However, PINO is short-lived, which has
caused a limited operational utility for methodology develop-
ment due to PINO decomposition.8 The practical conse-
quence of this decomposition is the need to use NHPI at high
Figure 1. Previous studies on PINO decomposition under electro-
catalyst loading (10−20 mol %) to achieve efficient reactivity,
as exemplified in the application to the oxidation of β-O-4
linkages in lignin.6 Furthermore, the short lifetime of PINO
has made mechanistic studies challenging.3 Despite the interest
in the NHPI/PINO couple as a valuable redox mediator,3 a full
assessment of off-cycle (i.e., PINO decomposition) pathways
has not been performed.
chemical conditions.
Received: April 21, 2021
Published: July 2, 2021
PINO decomposition under electrochemical conditions was
first reported by Masui and co-workers in 1987.9 A trimeric
https://doi.org/10.1021/jacs.1c04181
J. Am. Chem. Soc. 2021, 143, 10324−10332
© 2021 American Chemical Society
10324

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
Figure 2. (A) Cyclic voltammograms for oxidation of 10 mM NHPI (yellow); in the presence of 10 mM water (purple); in the presence of 10 mM
2,6-lutidine (blue) at the scan rate of 0.1 V s−1. (B,C) Cyclic voltammograms for oxidation of 10 mM NHPI in the presence of 10 mM 2,6-lutidine
at the scan rate varying from 0.01 to 0.5 V s−1.
species (hereafter trimer) was isolated and characterized as the
major PINO decomposition product after bulk electrolysis of
NHPI in the presence of pyridine (Figure 1B). The formation
of the trimer by nucleophilic attack of NHPI upon its
oxoammonium ion was suggested (Figure 1C, path a). Later,
Pedulli and co-workers proposed a separate radical pathway
involving coupling between PINO and the acyl radical arising
from C−N bond fragmentation of another PINO (Figure 1C,
path b).10 Both decomposition pathways have been presumed
as relevant in the literature without detailed scrutiny.
Advancing NHPI/PINO as the redox mediator in an efficient
catalytic system requires disambiguation of the PINO
decomposition mechanisms.
To enable development of new strategies for selective
electrochemical oxidation using NHPI or similar species, we
have examined elementary aspects of PINO generation, and
ascertained how it participates in oxidation of benzylic alcohols
using electroanalytical and computational methods. The data
indicate a multiple-site concerted proton−electron transfer
(MS-CPET) mechanism of PINO generation from NHPI,
where the proton and electron move to different locations.11
Moreover, it suggests that in lieu of the two aforementioned
mechanistic proposals for the decomposition of PINO,
electrochemically generated PINO undergoes two distinct
second-order decay processes, one of which is promoted by
base. The influence of base strength on the catalytic efficiency
of benzylic alcohol oxidation is quantitively analyzed. Taken
together, this study provides a foundation for the development
of more effective N-oxyl electrocatalysts.
common assumption that NHPI can be reversibly oxidized
both with and without a Brønsted base.4−7,13,14
Variation in the electrolyte did not affect the voltammetric
response of NHPI unless care was not taken to exclude water
(Supporting Information, SI). Adventitious and/or intentional
addition of water to the electrolyte caused the voltammetric
response to appear quasi-reversible, implicating water as an
active agent during the oxidation of NHPI and suggesting
similar wet conditions were likely used in prior reports of
electrochemical oxidation of NHPI without added
base.4−7,13,14 In fact, we discovered that leakage of water
(H2O) from an aqueous reference electrode was sufficient to
cause quasi-reversible behavior. The formal standard potential
for NHPI oxidation in the presence of added H2O, as
determined from the midpoint of the redox waves (E0’ ≈ E1/2),
was a function of the concentration of added water (SI). When
one equivalent of water was used, E1/2 = +1.04 V versus
E0’(Fc+/Fc) (Figure 2A).15
Cyclic voltammetric responses for the oxidation of NHPI in
the presence of 1 equiv of 2,6-lutidine were collected from 0.01
to 0.5 V s−1 (Figure 2B). Normalizing the current densities in
Figure 2C by v highlighted the fact that the voltammetric
response shape was sensitive to the experimental time scale,
indicating the oxidation of NHPI was coupled with at least one
additional chemical reaction. At scan rates ≤0.01 V s−1, the
voltammetric response became sigmoidal, attaining a steady-
state current density (Figure 2C) that suggested redox active
species were being generated at the electrode by follow-up
reactions in solution. A steady-state current density with the
same magnitude was observed consistently for bases with pKa
values <15 and was insensitive to base concentration; see
Supporting Information for further details.
RESULTS
■
Redox Response of NHPI with/without Bases. The
redox response of NHPI at a glassy carbon electrode was
measured in anhydrous MeCN with recrystallized tetrabuty-
lammonium hexafluorophosphate (NBu4PF6).12 Figure 2A
shows the voltammetric response of NHPI without base in
the electrolyte. In contrast to numerous previous reports,
including our own,4−7,13,14 a reversible voltammetric response
was not observed. Instead, NHPI could not be readily oxidized
at 0.1 V s−1 in this potential range under anhydrous conditions.
Moreover, the lack of cathodic current on the return sweep
suggested oxidized NHPI (NHPI•+) was unstable on this time
scale. The redox response of NHPI oxidation speaks against a
To examine the possible contribution of inner-sphere effects
in the redox responses, voltammetric measurements were
performed with anodically activated glassy carbon.16 However,
this deliberate change in electrode surface chemistry did not
impact the primary voltammetric features, as the observed peak
splitting was invariant with the nature of the glassy carbon
surface; see Supporting Information for further details.
NHPI-Base Complex. Identification of any possible
reactions between NHPI and base was performed first, as
most PINO-catalyzed oxidations are carried out in the
presence of bases. When one equivalent of 2,6-lutidine was
10325
https://doi.org/10.1021/jacs.1c04181
J. Am. Chem. Soc. 2021, 143, 10324−10332

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
Figure 3. (A) IR spectroscopy of 0.05 M NHPI in MeCN, followed by addition of 2,6-lutidine. (B) Selected cyclic voltammograms for titration of
10 mM NHPI with 2,6-lutidine from 0 to 5 equiv at 0.1 V s−1 and the anodic current density as a function of ratio of [2,6-lutidine]/[NHPI]. (C)
The free energy of the formation of the conjugate acid of NHPI, NHPI/base complex formation (B3LYP/6-311+G(d) with IEFPCM (MeCN)).
used as base, a quasi-reversible voltammogram was obtained
with E1/2 = +0.39 V versus E0’(Fc+/Fc) (Figure 2A). The
presence of base decreased the standard potential and
increased the reversibility of NHPI oxidation. The reported
pKa values of NHPI (23.5)11 and the 2,6-lutidinium ion
(14.1)17 in MeCN suggest that proton exchange is highly
unfavorable. Nevertheless, the plausibility of a hydrogen bond
formation between NHPI and pyridine derivative was assessed
through IR spectroscopy. NHPI in MeCN shows a character-
istic broad O−H stretch at 3210 cm−1 (Figure 3A). Addition
of 2,6-lutidine caused a decrease in this absorption and the
appearance of a very broad absorption at ∼2450 cm−1,
corresponding to the O−H stretch of the hydrogen-bonded
complex.18 The cumulative spectra were consistent with an
equilibrium constant Keq = 11.8 M−1 for a hydrogen-bonded
complex (Figure 3A).
To determine the stoichiometry of the hydrogen-bonded
complex, cyclic voltammetric titration experiments were
conducted at 0.1 V s−1 (Figure 3B). A solution of NHPI in
MeCN was titrated with 2,6-lutidine. The quasi-reversible
voltammetric response at ∼+0.4 V versus E0’(Fc+/Fc)
corresponding to oxidation of the NHPI−lutidine complex
grew in magnitude as the concentration of 2,6-lutidine
increased with a concomitant disappearance of the voltam-
metric response observed in the absence of a base. The anodic
peak current density for the oxidation of the complex was
saturated until one equivalent of 2,6-lutidine was added,
suggesting NHPI and 2,6-lutidine form a 1:1 complex in
solution (Figure 3B). The E1/2 was dependent upon the
concentration of 2,6-lutidine (SI), consistent with pre-
equilibrium complexation.
A set of calculations on the energies for both an ion pair of
deprotonated NHPI and pyridinium and for a hydrogen-
bonded complex was performed (Figure 3C). The free energy
change corresponding to the formation of the 1:1 hydrogen-
bonded complex between NHPI and pyridine was calculated to
be −2.3 kcal/mol, rendering a formation constant (Kf) of 47.9.
In reasonable agreement with this computation, Abraham’s
hydrogen-bonding parameters determined for pyridine (
β2H = 0.62)19 and NHPI (α2H = 0.37)20 yield an estimate a
Kf of 3.9.21 In contrast, the Brønsted acid−base reaction (i.e.,
deprotonation) was predicted to be highly endergonic (ΔG =
+10.6 kcal/mol). With 4-dimethylamino-pyridine (DMAP),
deprotonation was expectedly more favorable compared to
pyridine but still calculated to be endergonic (ΔG = +1.9 kcal/
mol). Similarly, the formation of a NHPI−DMAP H-bonded
complex is more thermodynamically favored (ΔG = −4.1 kcal/
mol) than its NHPI−pyridine counterpart. The smaller ΔG
between full deprotonation and NHPI−DMAP formation
(+6.0 kcal/mol) is consistent with the higher pKa of DMAP’s
conjugate acid (17.7).17
10326
https://doi.org/10.1021/jacs.1c04181
J. Am. Chem. Soc. 2021, 143, 10324−10332

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
Figure 4. (A) (a) Selected cyclic voltammograms and calculations for oxidation of NHPI with/without base. (b,c) E1/2 for oxidation of NHPI in the
presence of 1 equiv of base 0.1 V s−1. IP for oxidation of NHPI complexes, and free energy of dissociation of PINO and the associated conjugate
acid (B3LYP/6-311+G(d) with IEFPCM (MeCN)). (B) E1/2 versus pKa (blue) and ipa/ipc versus pKa (red).
Base Effect. To further probe the role of base in PINO
generation and decomposition, a series of experiments were
carried out with organic bases whose conjugate acids have well-
defined pKa values in MeCN.17 Cyclic voltammograms for the
oxidation of NHPI in the presence of one equivalent of
different organic bases were collected at 0.1 V s−1 (selected
examples presented in Figure 4A, the others in SI). In the
presence of stronger bases, NHPI was oxidized at less positive
potentials. The dependence of the E1/2 values on pKa are
summarized in Figure 4B (blue filled circles). Across a pKa
range of 2 to 19, the linear least-squares fit yielded a slope of
limide-N-oxide, was prepared22 and its cyclic voltammogram
was collected at 0.1 V s−1 in MeCN. Taking E1/2 to be
diagnostic of the formal potential, these data implied E0′
(PINO/PINO−) = +0.07 V versus E0′(Fc+/Fc) (Figure 4A).
The small cathodic peak further suggested PINO itself is
unstable on the time scale of the voltammetry.23 On the basis
of the cumulative E1/2 versus pKa data, the voltammetry
indicates a pKa of ∼19 for NHPI in MeCN.
For additional insight on the effect of base on electro-
chemical oxidation, we expanded the computations in Figure
3C to compare oxidation of NHPI, the H-bonded complexes
thereof, and PINO− (Figure 4A). Expectedly, NHPI was the
most difficult species to oxidize, followed by the hydrogen-
bonded complexes, and finally PINO−. Importantly, the H-
bonded complexes of NHPI and either pyridine or DMAP
were predicted to decrease the ionization potential by 24.4 and
30.4 kcal/mol, respectively, relative to the NHPI−water
complex. The decrease in ionization potential (IP) parallels
the trend of decreasing standard potential for oxidation of
NHPI observed experimentally, though a lack of quantitative
agreement is not surprising given the absence of implicit
solvation, which is presumably particularly relevant in the case
of water. Notably, the complexes resulting from oxidation (i.e.,
PINO−H−base+) were found to favor dissociation (ΔG =
−4.5 and −4.7 kcal/mol, respectively), rendering pyridinium
or dimethylaminopyridinium and free PINO. Hence, the
increased reversibility of NHPI oxidation in the presence of
pyridine does not appear to be due to product complexes
which increase the persistence of PINO.
56
1 mV per pKa unit, consistent with the Nernst factor
(2.303 × RT/nF = 59 mV/pKa) for proton-coupled electron
transfer (PCET).11
A separate analysis of the apparent electron transfer rate
constant (k0) for each voltammogram in Figure S7 did not
yield an obvious trend with pKa. A likely factor was the
differences in steric crowding around the pyridine nitrogen
atom. For example, voltammetry with bipyridine yielded a
much smaller k0 value in comparison to pyridine, despite both
voltammograms showing similar E1/2
.
Besides the potential shift, we also observed that the
magnitude of the cathodic peak current varied with base
strength (Figure 4A). Across these organic bases, the ratio of
the anodic and cathodic peak current (ipa/ipc) was a function of
the pKa of the base, approaching 1 as the pKa increased to 15
and then increasing at larger pKa values (Figure 4B, red filled
circles). In the presence of stronger bases, for example, DMAP
and quinuclidine, the oxidation was irreversible. While the role
of base as a proton acceptor is generally recognized,4−8,11 the
reason for the erosion of the reversibility of the electrochemical
oxidation is not immediately apparent. On the basis of the lack
of reversibility above a certain range of pKa, it would seem that
at least under electrochemical conditions, the decomposition
of PINO is accelerated under increasingly basic conditions.
In an effort to delineate the effects of individual bases, the
redox response of the conjugate base of NHPI (i.e., PINO−)
was determined. An NHPI salt, tetrabutylammonium phtha-
A test substrate, 1-phenylethanol, was used to assess whether
the base affects the catalytic efficiency. A catalytic current
enhancement was observed when the substrate was added to a
solution of NHPI and organic bases (pKa < 15) (SI). The
catalytic current was independent of the pKa of the base (SI).
Importantly, no current enhancement was observed when
stronger bases (i.e., DMAP, NEt3 or quinuclidine) were used.
Comparing the catalytic currents measured for 1-phenyl-
ethanol and 1-phenylethanol-d1 yielded a kinetic isotope effect
10327
https://doi.org/10.1021/jacs.1c04181
J. Am. Chem. Soc. 2021, 143, 10324−10332

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
Figure 5. (A) Electrolysis conditions: 1-phenylethanol (0.16 mmol), NHPI (10 mol %, 0.016 mmol), base (2 equiv., 0.32 mmol), KPF6 (0.1 M in 8
mL MeCN), open-air, divided cell, glassy carbon plate anode (2 cm2), graphite cathode (4 cm2), potential controlled electrolysis (0.2 V more
positive than anodic peak potential for each redox couple). (a) RSM = remaining starting material. Yields determined by GC-MS. (B)
Chronoamperometry recorded by potentiostat.
of 1.8, suggesting that the abstraction of the benzylic hydrogen
atom is rate-determining in the electrochemical oxidation (SI).
Electrolyses of 1-phenylethanol catalyzed by NHPI were
performed in the presence of different bases at a constant
potential for 6 h (Figure 5). The potential was set to +0.2 V
relative to the anodic peak potential for each redox couple. The
resulting solutions were analyzed by gas chromatography
coupled to a mass spectrometer (GC-MS). The best catalytic
efficiency was observed for bases with a pKa within the range of
∼11 to 15 with full conversion within 6 h. Studying the
reaction by integrating chronocoulometry (i.e., charge as a
function of time) indicated that ∼1 equiv of electrons passed
through the cell, suggesting the absence of chain reactions.24
The correlation of reaction time and base strength was in line
with the following premise: Weaker bases (pKa < 10) displayed
a current drop during the reaction and low conversions were
obtained. On the other hand, stronger bases (pKa > 17) were
ineffective for catalysis, resulting in no desired oxidation.
Consistent with the foregoing, electrolysis of 1-phenylethanol
in the presence of 1 equiv of the NHPI salt resulted in >90%
recovery of the starting material. A substituent effect was
noted, as the current magnitude and time to reach full
conversion was dependent upon the steric bulk of the base.
The less sterically hindered bases gave higher currents and
shorter reaction times. However, when the pKa was above 15,
even sterically hindered bases (e.g., 2,2,6,6-tetramethylpiper-
idine) were not able to promote the oxidation.
oxoammonium ion derived from TEMPO, PINO+ has not yet
been reported in the literature. Indeed, the computed free
energy difference between two molecules of PINO and PINO−
+ PINO+ (ΔG = +51.9 kcal/mol, Figure 6, path a) suggests
Figure 6. Reaction pathways initially considered for decay of PINO
(B3LYP/6-311+G(d) with IEFPCM (MeCN)).
Mechanistic Insights on PINO Decomposition. To
understand the causes of catalyst decomposition in order to
develop more efficient catalytic systems, PINO decomposition
was investigated using computational methods to interrogate
both the existing mechanistic proposals as well as others which
we considered plausible. Initially, we examined the second-
order decay originally proposed by Masui (Figure 1C, path a).9
This decomposition path proceeds first by disproportionation
of two molecules of PINO as the (presumed) rate-determining
step. The resulting oxoammonium PINO+ (2) could then
undergo acyl substitution by either NHPI or its conjugate base
PINO− to yield the putative dimeric intermediate, which could
subsequently lead to the isolated trimer (vide infra). Unlike the
that disproportionation is highly endergonic. Furthermore, the
IP of PINO was computed to be highly similar to that of NHPI
(ΔG = +158.6 vs +159.9 kcal/mol), suggesting that the
potential required for anodic oxidation would be well beyond
the potential at which oxidation of the NHPI-base complex
occurs. On this basis, the formation of a discrete intermediate
2, as originally depicted by Masui, is unlikely to occur.
The previously suggested C−N fragmentation was also
examined (Figure 1C, path b).10 The only productive mode of
C−N cleavage identified by computational means was through
the rotation of one of the C(aryl)−C(acyl) bonds (Figure 6,
path a), which led to a high energy charge-separated activated
10328
https://doi.org/10.1021/jacs.1c04181
J. Am. Chem. Soc. 2021, 143, 10324−10332

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
Figure 7. Free energy diagram of proposed bimolecular decomposition pathways of PINO (B3LYP/6-311+G(d) with IEFPCM (MeCN) and GD3
empirical dispersion correction).
complex 3 with acylium cation and amide anion N-oxyl
moieties (ΔG = +53.9 kcal/mol), not an acyl radical. Intrinsic
reaction coordinate (IRC) calculations revealed that this
transition state collapses to phthalic anhydride monoimine-N-
oxyl 4, obviously a consequence of the conformational
restriction imposed on the reactive moieties by the aryl ring
to which they are attached. To provide some insight on the
intrinsic strength of the C−N bond in question, we determined
the energy difference between maleimide-N-oxyl 5 and the
(E)-isomer of its C−N bond fragmentation product 6 (Figure
6, path b). This transformation has essentially the same free
energy cost as formation of the activated complex of PINO
heterolysis (ΔG = +54.1 kcal/mol). The results imply that the
suggested C−N bond cleavage is also unlikely to occur under
relevant conditions. In addition, the concerted fragmentation
of both C−N bonds analogous to the reverse reaction of
quinone dimethide-based nitric oxide cheletropic traps
(NOCTs) was considered (Figure 6, path c).25 Although a
transition state was not found, considering the endergonicity of
this reaction (ΔG = +59.3 kcal/mol), it is also unlikely to
contribute to the decay of PINO. Admittedly, this point is
perhaps not surprising given NOCT reactions have not been
reported as being reversible.
Considering the modes by which two molecules of PINO
may dimerize, a lower energy pathway similar to Masui’s
original disproportionation proposal was found (Figure 7). An
activated complex corresponding to the combination of
PINO− and PINO+ was identified with ΔG = +36.5 kcal/
mol relative to two isolated PINO molecules and is much
lower in energy than the sum computed from the energies of
the discrete PINO− and PINO+ ions (ΔG = +51.9 kcal/mol,
Figure 6A). IRC calculations suggest that this transition state
connects Masui’s aforementioned dimeric intermediate to a
PINO contact ion pair at ΔG = +16.8 kcal/mol relative to two
isolated PINO molecules. It seems reasonable to suggest that
this contact ion pair results from charge transfer from a
corresponding PINO radical pair to avoid formation of the
higher energy separated PINO− and PINO+ ions. Indeed, the
corresponding PINO (singlet) radical pair is 7.0 kcal/mol
lower in free energy than the contact ion pair. Suspecting that
the free energy difference between the PINO radical pair and
two separated PINO radicals (9.8 kcal/mol) was overestimated
on account of DFT’s failure to account for dispersion
interactions, we incorporated Grimme’s empirical dispersion
correction (GD3)26 in our calculations. Doing so led to a
decrease in the free energy change for association of the PINO
radical pair (to +5.7 kcal/mol), the contact ion pair (to +10.3
kcal/mol) and overall barrier for substitution (to +30.7 kcal/
mol); see Supporting Information for further details.
Given the lack of reversibility observed when the
tetrabutylammonium salt of NHPI is oxidized or when
stronger bases are present, we wondered if PINO− may attack
PINO directly. Indeed, a comparatively low energy transition
state structure was readily identified for this substitution
(Figure 7). The transition state was characterized by significant
O−C bond formation and C−N bond cleavage. IRC
calculations revealed that this structure connected a PINO/
PINO− complex and the substitution product independent of a
tetrahedral intermediate. Again, inclusion of the GD3 empirical
dispersion correction systematically decreased the overall
computed free-energy barrier from +17.9 to +12.5 kcal/mol.
Although, as in the case of the reaction of two PINO
molecules, the reaction is endergonic (ΔG = +10.7 kcal/mol),
the resultant amide anion N-oxyl should oxidize readily since
its ionization potential is essentially equivalent to that
calculated for PINO− (ΔG = 106.3 vs 107.1 kcal/mol,
respectively).27 This oxidation would generate Masui’s
proposed dimeric intermediate.
The optimized structure of Masui’s proposed dimer was
characterized by a conspicuously long C−N bond between the
carbonyl carbon and nitroso nitrogen atoms (1.59 Å), implying
it to be quite weak. Indeed, the calculated bond dissociation
10329
https://doi.org/10.1021/jacs.1c04181
J. Am. Chem. Soc. 2021, 143, 10324−10332

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
Figure 8. Mechanisms of the electrochemical generation and decomposition of PINO.
free energy (BDFE) is +10.2 kcal/mol, implying the homolysis
would occur readily under the reaction conditions, releasing
nitric oxide. The incipient acyl radical could be trapped by
PINO in a reaction that is predicted to be 42.6 kcal/mol
downhill.28 It should be noted that, in absolute terms, the
B3LYP/6-311+G(d) method used for the foregoing calcu-
lations is likely to underestimate these bond strengths.
Therefore, to provide further insight, we carried out high
accuracy CBS-QB3 calculations on a model wherein the
“spectator” PINO moiety is replaced with a hydroxyl group,
which yielded corresponding free energies of +15.1 and −58.7
kcal/mol, respectively).29 Nevertheless, the results support a
transitory “dimeric” intermediate and a stable “trimeric”
product in line with the observations made to date.
Considering the PINO decomposition mechanism is base
promoted, the decomposition rate must be base dependent. A
series of low scan rate cyclic voltammetric measurements was
performed at 0.01 V s−1 as in Figure 2c. Among these
investigated bases, the steady-state currents were observed
consistently (SI). With weak bases (pKa < 15), the steady-state
current was independent of base strength and base
concentration. When stronger bases (pKa > 15) were used,
the steady-state current was proportional to the base
concentration. Taking this information together, we posit
that the observed steady state current indicated that PINO
decomposition produces a new redox active species that is
easier to oxidize. These voltametric observations are therefore
in agreement with aforementioned computational studies.
analogy to their study on base mediated oxidation of N-
hydroxy-2,2,6,6-tetramethylpiperidine (TEMPO-H).18 The
E1/2 versus pKa correlation addresses two further aspects.
First, these data indicate that E0’(NHPI•+/NHPI) is ≥+1.05 V
versus E0′(Fc+/Fc), in agreement with prior predictions from
theory.11 Second, these data indicate that E0′(PINO/PINO−)
is closer to +0.07 V versus E0′(Fc+/Fc) − which is slightly
more positive than prior theoretical predictions of E0′(PINO/
PINO−) = −0.1 V versus E0′(Fc+/Fc).11 Similarly, the
cumulative voltammetric measurements shown here indicate
that the pKa of NHPI in MeCN is 19 (rather than 23.5).11
The subject of PINO decomposition has been a matter of
interest since the seminal kinetic studies conducted by Masui,9
although the details have remained hitherto nebulous. Both
first30 and second order decay9,31 with respect to PINO have
been reported leading to a bifurcation of proposed decay
processes that have been presumed to lead to the same trimer
decomposition product.8 Our data suggests that ascribing the
observed first-order decay30 of PINO to a unimolecular
phenomenon perhaps takes too much for granted. The
underlying assumption is that the principal product of the
process is the same trimer previously observed by Masui.
Additionally, as pointed out by Pedulli, the kinetics of decay
are quite dependent on solvent composition,30a an observation
which seems inconsistent with unimolecular decay being rate-
determining. However, without more information on the actual
product(s) formed as a result of the first-order decay
process(es), or the prerequisites for its occurrence, searching
for the rate defining reaction in silico may be futile. Suffice to
say, the oft-invoked fragmentation of the C−N bond in PINO
to form a transient acyl radical which can be captured appears
unlikely.
The presented data suggest that the strength of the base
alters the catalytic efficiency by influencing PINO decom-
position. The voltammetric measurements and electrolysis
experiments suggest an optimal base strength in the pKa range
of ∼11−15. Specifically, we posit that PINO has two operative
decomposition pathways in MeCN (Figures 7 and 8). One
decomposition involves the dimerization of two molecules of
PINO via a charge-transfer complex. In the presence of weak
bases (pKa < 15), this pathway is dominant. In the presence of
comparatively strong bases (pKa > 15), a second significantly
faster decomposition pathway consisting of the reaction
between PINO− and PINO is operative. The resulting
intermediate is rapidly oxidized at the anode to produce the
DISCUSSION
■
The presented data speak to the following three points. First,
PINO-catalyzed electrochemical oxidation occurs by a MS-
CPET mechanism. Second, the reversibility of electrochemical
oxidation of NHPI is dependent on the added base strength,
and consequently the base strength affects the conversion
efficiency of the electrocatalytic oxidation of 1-phenylethanol.
Third, PINO decomposition is promoted by base via
nucleophilic substitution by PINO− on a carbonyl of a second
PINO followed by single electron oxidation of the resultant
complex.
The linear dependence of E1/2 on pKa with a slope of 56
1
mV/pKa, observed for the oxidation of NHPI in the presence
of base, is a hallmark of PCET reactions.11 Indeed, Mayer and
co-workers have suggested MS-CPET as a plausible mecha-
nism for the base-mediated formation of PINO from NHPI, in
10330
https://doi.org/10.1021/jacs.1c04181
J. Am. Chem. Soc. 2021, 143, 10324−10332

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
same dimer. This oxidation event is observable at slow scan
rates. The dimer undergoes C−N cleavage to release nitric
oxide and form an acyl radical which reacts with another
molecule of PINO to form the observed trimer.
Corey R. J. Stephenson − Willard Henry Dow Laboratory,
Department of Chemistry, University of Michigan, Ann
Arbor, Michigan 48109, United States; orcid.org/0000-
0002-2443-5514; Email: crjsteph@umich.edu
Our expansion on early observations regarding base-
promoted oxidations and elucidation of a reasonable PINO
decomposition mechanism provides guidance to advance
PINO-catalyzed electrochemical oxidations. Previous efforts
to design improved PINO-like HAT catalysts have largely
focused on adjusting BDFEsup to yield a more reactive
catalyst or down to yield a more selective catalyst. In contrast,
an approach to improving catalysis through extension of the
effective catalytic lifetime of PINO by obviating degradation
pathways has not been a focus of research efforts. On the basis
of the results presented herein, we suggest that both of these
aspects should be considered in the design of more effective N-
oxyl catalysts for C−H oxidation reactions.
Authors
Cheng Yang − Willard Henry Dow Laboratory, Department
of Chemistry, University of Michigan, Ann Arbor, Michigan
48109, United States
Luke A. Farmer − Department of Chemistry and Biomolecular
Sciences, University of Ottawa, Ottawa, Ontairo K1N 6N5,
Canada
Complete contact information is available at:
https://pubs.acs.org/10.1021/jacs.1c04181
Notes
The authors declare no competing financial interest.
Furthermore, we posit the following directions for future N-
oxyl catalyst development. First, it is of great importance to
develop electrocatalysts compatible with strong bases.
Avoiding rapid base-promoted decay will allow one to operate
reactions at lower applied potentials using stronger bases,
which will result in a broader substrate scope and an extension
to the reactions that require stronger bases for which NHPI is
poorly suited. Considering the electrophilic carbonyl is key to
decomposition, replacing one or both carbonyls with other
electron-withdrawing functional groups could solve this
problem. Second, strategic design of electrocatalysts with a
much slower bimolecular decay should be pursued. The
resulting longer lifetime of the catalyst will improve such
oxidation reactions by decreasing catalyst loading. Rendering
the carbonyls less electrophilic could be a way to decrease
bimolecular decay. Studies in these directions are ongoing in
our laboratories.
ACKNOWLEDGMENTS
■
C.R.J.S. acknowledges the financial support from the National
Science Foundation (CHE-1565782 and CBET-2033714), and
the University of Michigan. S.M. acknowledges the generous
support from the Department of Energy (DE-SC0006628).
D.A.P. acknowledges the financial support from the Natural
Sciences and Engineering Research Council of Canada. This
work was enabled by support provided by Compute Ontario
and Compute Canada (www.computecanada.ca).
REFERENCES
■
(1) Yan, M.; Kawamata, Y.; Baran, P. S. Synthetic Organic
Electrochemical Methods Since 2000: On the Verge of a Renaissance.
Chem. Rev. 2017, 117, 13230−13319.
(2) Moeller, K. D. Synthetic Applications of Anodic Electro-
chemistry. Tetrahedron 2000, 56, 9527−9554.
(3) Nutting, J. E.; Rafiee, M.; Stahl, S. S. Tetramethylpiperidine N-
Oxyl (TEMPO), Phthalimide N-Oxyl (PINO), and Related N-Oxyl
Species: Electrochemical Properties and Their Use in Electrocatalytic
Reactions. Chem. Rev. 2018, 118, 4834−4885.
CONCLUSION
■
In summary, we have described a comprehensive electro-
analytical study of the NHPI/PINO redox mediator and
conducted computational studies to uncover the mechanisms
by which PINO degrades as a function of the identity of the
base. The data provides a roadmap for developing more
effective N-oxyl catalysts for C−H oxidation reactions.
(4) Horn, E. J.; Rosen, B. R.; Chen, Y.; Tang, J.; Chen, K.; Eastgate,
M. D.; Baran, P. S. Scalable and Sustainable Electrochemical Allylic C-
H Oxidation. Nature 2016, 533, 77−81.
(5) Masui, M.; Ueshima, T.; Ozaki, S. N-Hydroxyphthalimide as an
Effective Mediator for the Oxidation of Alcohols by Electrolysis. J.
Chem. Soc., Chem. Commun. 1983, 479−480.
(6) Bosque, I.; Magallanes, G.; Rigoulet, M.; Kärkäs, M. D.;
Stephenson, C. R. J. Redox Catalysis Facilitates Lignin Depolymeriza-
tion. ACS Cent. Sci. 2017, 3, 621−628.
(7) Rafiee, M.; Wang, F.; Hruszkewycz, D. P.; Stahl, S. S. N-
Hydroxyphthalimide-Mediated Electrochemical Iodination of Meth-
ylarenes and Comparison to Electron-Transfer-Initiated C-H
Functionalization. J. Am. Chem. Soc. 2018, 140, 22−25.
(8) Recupero, F.; Punta, F. Free Radical Functionalization of
Organic Compounds Catalyzed by N-Hydroxyphthalimide. Chem.
Rev. 2007, 107, 3800−3842.
(9) Ueda, C.; Noyama, M.; Ohmori, H.; Masui, M. Reactivity of
Phthalimide-N-oxyl: A Kinetic Study. Chem. Pharm. Bull. 1987, 35,
1372−1377 Note: In Masui’s report, the potential controlled
electrolysis was performed at 0.9 V vs SCE, which is 0.52 V vs E0’
(Fc+/Fc) (see ref 15.)..
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge at
https://pubs.acs.org/doi/10.1021/jacs.1c04181.
■
sı
Cyclic voltammetry, electrolysis, compound character-
ization, and calculations (PDF)
AUTHOR INFORMATION
Corresponding Authors
■
Derek A. Pratt − Department of Chemistry and Biomolecular
Sciences, University of Ottawa, Ottawa, Ontairo K1N 6N5,
Canada; orcid.org/0000-0002-7305-745X;
Email: dpratt@uottawa.ca
Stephen Maldonado − Willard Henry Dow Laboratory,
Department of Chemistry, University of Michigan, Ann
Arbor, Michigan 48109, United States; Program in Applied
Physics, University of Michigan, Ann Arbor, Michigan 48109,
United States; orcid.org/0000-0002-2917-4851;
Email: smald@umich.edu
(10) Amorati, R.; Lucarini, M.; Mugnaini, V.; Pedulli, G. F.; Minisci,
F.; Recupero, F.; Fontana, F.; Astolfi, P.; Greci, L. Hydroxylamines as
Oxidation Catalysts: Thermochemical and Kinetic Studies. J. Org.
Chem. 2003, 68, 1747−1754.
(11) Warren, J. J.; Tronic, T. A.; Mayer, J. M. Thermochemistry of
Proton-Coupled Electron Transfer Reagents and its Implications.
Chem. Rev. 2010, 110, 6961−7001. Note: The reported pKa of NHPI
10331
https://doi.org/10.1021/jacs.1c04181
J. Am. Chem. Soc. 2021, 143, 10324−10332

Journal of the American Chemical Society
pubs.acs.org/JACS
Article
in MeCNis an extrapolated value based on the pKa measured in
̈
DMSO as opposed to a direct measurement in MeCN. See: Kutt, A.;
Transfer between Nitroxides and HOO· Radicals and Implication for
Catalytic Coantioxidant Systems. J. Am. Chem. Soc. 2018, 140, 10354.
In fact, at the potential of the experiment HOO• is probably oxidized.
See: Koppenol, W. H.; Stanbury, D. M.; Bounds, P. L. Electrode
Potentials of Partially Reduced Oxygen Species, from Dioxygen to
Water. Free Radical Biol. Med. 2010, 49, 317−322.
Leito, I.; Kaljurand, I.; Sooväli, L.; Vlasov, V. M.; Yagupolskii, L. M.;
Koppel, I. A. A Comprehensive Self-Consistent Spectrophotometric
Acidity Scale of Neutral Brønsted Acids in Acetonitrile. J. Org. Chem.
2006, 71, 2829−2838.
(25) Korth, H.-G.; Sustmann, R.; Lommes, P.; Paul, T.; Ernst, A.; de
Groot, H.; Hughes, L.; Ingold, K. U. J. Am. Chem. Soc. 1994, 116,
2767−2777.
(12) Sawyer, D. T.; Hage, J. P.; Sobkowiak, A. J. Am. Chem. Soc.
1995, 117, 106.
(13) Kishioka, S.; Yamada, A. Kinetic Study of the Catalytic
Oxidation of Benzyl Alcohols by Phthalimide-N-Oxyl Radical
Electrogenerated in Acetonitrile using Rotating Disk Electrode
Voltammetry. J. Electroanal. Chem. 2005, 578, 71−77.
(26) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A Consistent and
Accurate ab initio Parameterization of Density Functional Dispersion
Correction (DFT-D) for the 94 Elements H-Pu. J. Chem. Phys. 2010,
132, 154104.
(27) Shown is the IP for amide anion N-oxyl without GD3
correction applied, and it was found to have no impact on the
computed value regardless (106.3 vs 106.1 kcal/mol).
(28) The DFT BDFE values were calculated both with and without
GD3 empirical correction for the C−N of the dimer complex (+10.2/
+7.2 kcal/mol respectively) and the C−O of the trimer complex
(+42.1/+38.4 kcal/mol respectively). The former is highlighted in the
main text for direct comparison with the bimolecular pathways
described vide supra.
(14) Rafiee, M.; Karimi, B.; Alizadeh, S. Mechanistic Study of the
Electrocatalytic Oxidation of Alcohols by TEMPO and NHPI.
ChemElectroChem 2014, 1, 455−462.
(15) 0 V vs E0’ (Fc+/Fc) = + 0.38 V vs SCE = + 0.63 V vs NHE. See:
Pavlishchuk, V. V.; Addison, A. W. Conversion Constants for Redox
Potentials Measured Versus Different Reference Electrodes in
Acetonitrile Solutions at 25°C. Inorg. Chim. Acta 2000, 298, 97−102.
(16) Sawant, T. V.; McKone, J. R. Flow Battery Electroanalysis. 2.
Influence of Surface Pretreatment on Fe(III/II) Redox Chemistry at
Carbon Electrodes. J. Phys. Chem. C 2019, 123, 144−152.
(29) The C−O and C−N BDE/BDFE values were calculated using
a high accuracy CBS-QB3 method (gas-phase) with the phthalimide
on the other carbonyl truncated to a hydroxyl group. Replacing the
phthalimide with hydroxyl was found to have a negligible impact on
the BDE/BDFE when using the B3LYP/6-311+G(d) method sans
GD3 (≤ 1.5 kcal/mol). The calculated C-O BDE/BDFE of the model
trimer was 71.3/58.7 kcal/mol. The calculated C-N BDE/BDFE of
the model intermediate dimer was 26.7/15.1 kcal/mol. For
comparison, 26.7 kcal/mol is ∼10 kcal/mol lower than the O-O
BDE of a typical alkyl peroxide. See: Bach, R. D.; Ayala, P. Y.;
Schlegel, H. B. J. Am. Chem. Soc. 1996, 118, 12758−12765.
(30) (a) Amorati, R.; Lucarini, M.; Mugnaini, V.; Pedulli, G. M.;
Minisci, F.; Recupero, F.; Fontana, F.; Astolfi, P.; Greci, L. J. Org.
Chem. 2003, 68, 1747−1754. (b) Kushch, O.; Hordieieva, I.;
Novikova, K.; Litvinov, Y.; Kompanets, M.; Shendrik, A.; Opeida, I.
J. Org. Chem. 2020, 85, 7112−7124.
(17) (a) Lokov, M.; Tshepelevitsh, S.; Heering, A.; Plieger, P. G.;
̃
Vianello, R.; Leito, I. On the Basicity of Conjugated Nitrogen
Heterocycles in Different Media. Eur. J. Org. Chem. 2017, 2017,
4475−4489. (b) Kaljurand, I.; Rodima, T.; Leito, I.; Koppel, I. A.;
Schwesinger, R. Self-Consistent Spectrophotometric Basicity Scale in
Acetonitrile Covering the Range between Pyridine and DBU. J. Org.
Chem. 2000, 65, 6202−6208. (c) Kolthoff, I. M.; Ikeda, S.
Polarographic and Acid Properties of Thorium Perchlorate in
Acetonitrile. J. Phys. Chem. 1961, 65, 1020−1026. (d) iBonD 2.0
databank: http://ibond.nankai.edu.cn. Accessed on June 1st, 2020.
(18) Morris, W. D.; Mayer, J. M. Separating Proton and Electron
Transfer Effects in Three-Component Concerted Proton-Coupled
Electron Transfer Reactions. J. Am. Chem. Soc. 2017, 139, 10312−
10319.
(19) Abraham, M. H.; Grellier, P. L.; Prior, D. V.; Morris, J. J.;
Taylor, P. J. J. Chem. Soc., Perkin Trans. 2 1990, 2, 521−529.
(31) (a) Koshino, N.; Saha, B.; Espenson, J. H. J. Org. Chem. 2003,
68, 9364−9370. (b) Baciocchi, E.; Gerini, M. F.; Lanzalunga, O. J.
Org. Chem. 2004, 69, 8963−8966.
(20) The Kf of NHPI with MeCN (βH = 0.44)18 is reported (1.3
2
M−1
)
from which the α2H of NHPI (0.37) is calculated
27a
(log Kf = 7.354α2HβH − 1.094).
2
(21) While at first glance the 12-fold difference (1.5 kcal/mol)
between the two predicted formation constants for the H-bonded
complex between NHPI and pyridine seems quite large, we consider it
remarkably good agreement given the two approaches. One is derived
from empirical solvent H-bond acidity/basicity parameters derived
from IR measurements with a variety of donor/acceptor combinations
in CCl4, and the other is derived from gas phase thermochemistry to
which a solvent reaction field corresponding to the bulk electrostatic
properties of acetonitrile has been applied.
(22) Dekamin, M. G.; Mokhtari, J.; Naimi-Jamal, M. R. Organo-
catalytic cyanosilylation of carbonyl compounds by tetrabutylammo-
nium phthalimide-N-oxyl. Catal. Commun. 2009, 10, 582−585.
(23) The current magnitude for oxidation of the NHPI salt was
smaller than the other NHPI/base complex. This was possibly due to
a self-complexation between the ionic species, which was presented in
the SI.
(24) It would appear that it is not a chain reaction, or the chain
reaction is extremely short. We suspect this may be since the rate
constant for reaction of a hydroperoxyl radical (HOO•) and benzylic
alcohol substrate is so much slower than HOO• and PINO, which is
presumably diffusion controlled given that the reaction between
HOO• and TEMPO occurs with k ∼ 109 M−1 s−1. See: (a) Harrison,
K. A.; Haidasz, E. A.; Griesser, M.; Pratt, D. A. Inhibition of
hydrocarbon autoxidation by nitroxide-catalyzed cross-dismutation of
hydroperoxyl and alkylperoxyl radicals. Chem. Sci. 2018, 9, 6068−
6079. (b) Baschieri, A.; Valgimigli, L.; Gabbanini, S.; DiLabio, G. A.;
Romero-Montalvo, E.; Amorati, R. Extremely Fast Hydrogen Atom
10332
https://doi.org/10.1021/jacs.1c04181
J. Am. Chem. Soc. 2021, 143, 10324−10332