Redesign of a Pyrylium Photoredox Catalyst and Its Application to the Generation of Carbonyl Ylides

Article abstract of DOI:10.1021/acs.orglett.7b01222

We report the exploration into photoredox generation of carbonyl ylides from benzylic epoxides using newly designed 4-mesityl-2,6-diphenylpyrylium tetrafluoroborate (MDPT) and 4-mesityl-2,6-di-p-tolylpyrylium tetrafluoroborate (MD(p-tolyl)PT) catalysts. These catalysts are excited at visible wavelengths, are highly robust, and exhibit some of the highest oxidation potentials reported. Their utility was demonstrated in the mild and efficient generation of carbonyl ylides from benzylic epoxides that otherwise could not be carried out by current common photoredox catalysts.

Full text of DOI:10.1021/acs.orglett.7b01222

Letter
pubs.acs.org/OrgLett
Redesign of a Pyrylium Photoredox Catalyst and Its Application to
the Generation of Carbonyl Ylides
Edwin Alfonzo,† Felix Steven Alfonso,‡ and Aaron B. Beeler*,†
†Department of Chemistry, Boston University, Boston, Massachusetts 02215, United States
‡Department of Chemistry, Stanford University, Stanford, California 94305, United States
S
* Supporting Information
ABSTRACT: We report the exploration into photoredox
generation of carbonyl ylides from benzylic epoxides using
newly designed 4-mesityl-2,6-diphenylpyrylium tetrafluoroborate
(MDPT) and 4-mesityl-2,6-di-p-tolylpyrylium tetrafluoroborate
(MD(p-tolyl)PT) catalysts. These catalysts are excited at visible
wavelengths, are highly robust, and exhibit some of the highest
oxidation potentials reported. Their utility was demonstrated in
the mild and efficient generation of carbonyl ylides from benzylic
epoxides that otherwise could not be carried out by current
common photoredox catalysts.
hotoredox catalysis has become a valuable tool for the
Psynthesis of building blocks and biologically relevant
scaffolds. The ability to produce open shell intermediates under
low concentration and mild conditions has enabled numerous
elusive transformations, changing the way synthetic chemists
consider bond construction.1 Most notable are the catalysts that
use visible light to promote this reactivity, as UV light can cause
unproductive pathways to occur. In particular, organic dyes
have become a popular choice for photoredox reactions,
offering a metal-free alternative, lower cost, and high oxidation
potentials that are not available to their metal counterparts.2
Therefore, there has been a great deal of interest in the
discovery of organic photoredox catalysts that can both absorb
longer wavelengths of light and promote thermodynamically
challenging redox processes.3
We became interested in methods for the photoredox
generation of carbonyl ylides from benzylic epoxides. Such
reactions could be used to rapidly access cyclic ethers, which
are present in many biologically active natural products (Figure
1a).4 Carbonyl ylides are highly versatile intermediates which
Figure 1. (a) Representative natural products. (b) Previous work by
Whiting on the photoredox formation of carbonyl ylides. (c)
Mechanism proposed by Arnold and Albini.
can be generated from silicon-based 1, 3 eliminations, samarium
reduction, and photochemical reactions.5 The most common
approach relies on carbonyl interception of transient
carbenoids.5d These methods often suffer from issues such as
+1.99 V vs SCE), a visible light absorbing catalyst, was
employed (Figure 1b). This finding was critical because it
avoided direct excitation of the epoxide which leads to poor
yields. Although successful, this reaction was only demonstrated
on a single example, required super stoichiometric amounts of
the dipolarophile, and used degassed solvent due to the
formation of superoxide, generated by DCA, which led to
use of precious metals, superstoichiometric amounts of
reagents, toxicity, designed substrates, and poor yields. Thus,
we felt that a general and mild method to access carbonyl ylides
from benzylic epoxides would have broader application.
Early accounts of the photochemical formation of carbonyl
ylides from stilbene oxide and their cycloaddition to a wide
variey of dipolarophiles were reported by Arnold and Albini,6
Griffin,7 and others.8 Later, Whiting9 demonstrated that a
carbonyl ylide from an electron-rich epoxide could only be
Received: April 22, 2017
obtained when 9,10-anthracenedicarbonitrile (DCA) (Es1
=
red
© XXXX American Chemical Society
A
DOI: 10.1021/acs.orglett.7b01222
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
undesired side products.10 The reaction mechanism is proposed
to proceed through single electron transfer (SET) from the
epoxide to the excited DCA, forming both a reduced DCA and
a radical cation which rearranges to an oxonium radical (Figure
1c).6 A subsequent back electron transfer furnishes the desired
carbonyl ylide. Independent of the epoxide relative config-
uration, the cis product is favored, likely to avoid allylic-1,3
(A1,3) interaction in the oxonium radical intermediate.
Therefore, use of cis epoxide usually leads to higher
diastereoselectivity.
oxidizing 2,4,6-triphenylpyrylium tetrafluoroborate (TPT)
(Es1 = +2.55 V vs SCE) was used conversion to the desired
red
product could be observed, but major decomposition of the
catalyst limited this method (Figure 2a).
Although TPT has a higher oxidation potential than Mes-
Acr-Me, the latter and its derivatives, made popular by Nicewicz
and co-workers, have recently received more attention.11 This
is, in part, due to the protective nature of the orthogonal
mesityl group, which minimizes undesirable reactivity with
exogenous nucleophiles and transient radicals at the acridinium
chromophore.12 This unique feature has allowed a method in
which even highly reactive fluoride anions were employed.13
Unfortunately, TPT is deactivated by homocoupling of its
pyranyl form or by nucleophile and radical addition to its
pyrylium core, limiting its synthetic use.14 We considered
whether chemical modification of TPT could lead to a more
robust catalyst with an oxidation potential suitable for more
challenging reactions requiring higher oxidation potentials, such
as the carbonyl ylide formation of electron-deficient substrates.
TPT is known to possess two independent chromophores, the
4-arylpyrylium moiety which absorbs UV light and the 2,6-
diarylpyrylium that absorbs visible light. The latter is
responsible for the activity most often associated with TPT
mediated photoredox processes. We hypothesized that
replacing the 4-aryl moiety with a mesityl group will give
TPT a protective environment, while maintaining the photo-
physical properties attributed to the 2,6-diarylpyrylium
chromophore (Figure 3a). If successful, this will allow us to
access a catalyst with greater stability and high oxidation
potential.
Since the dipolarophile scope has been well explored and the
carbonyl ylide formation from epoxides had only been
demonstrated on a very limited number of diaryl substrates,
we began investigating this [3 + 2] dipolar cycloaddition by
trapping acetylenedicarboxylate (DMAD). Early experiments
revealed that DCA and Fukuzumi’s catalyst (Mes-Acr-Me)
(Es1
= +2.18 V vs SCE) could catalyze the desired
red
cycloaddition of stilbene oxide and DMAD (Figure 2a). After
Figure 2. Evaluation of photoredox catalyst for the [3 + 2] dipolar
cycloaddition and the scope when using DCA. a Standard conditions:
0.05 mmol of catalyst, 0.1 mmol of DMAD, 0.12 mmol of epoxide,
acetonitrile (MeCN) 0.5 M. b 0.01 mmol of DCA, 1 mmol of DMAD,
c
1.2 mmol of epoxide, 0.5 M degassed Chloroform (CHCl3). CHCl3
instead of MeCN.
optimization using DCA as the catalyst, we began probing
electron-rich diaryl epoxides in a preparative scale (Figure 2b).
These include examples having chlorine and alkyl ethers
incorporated in them (1−2). Additionally, an example bearing
naphthalene and an ortho electron-donating group was well
tolerated and the desired product (3) was isolated in excellent
yield. Heterocycles such as furan (4) and thiophene (5)
proceeded in moderate to excellent yields. Overall, diastereo-
meric ratios >4:1 were consistently obtained when using the cis
epoxide. The reactions with moderate yields were due to a
competing reaction, wherein C−O cleavage and subsequent
aryl migration afforded diaryl acetaldehyde (6).9 Unfortunately,
DCA and Mes-Acr-Me failed with electron-deficient epoxides
under several conditions (Figure 2a). However, when higher
Figure 3. (a) Rationalization for the synthesis of MDPT. (b) Emission
and absorbance spectra of MDPT and absorbance spectrum of TPT.
After considerable experimentation, we were able to
synthesize the desired 4-mesityl-2,6-diphenylpyrylium tetra-
fluoroborate (7, MDPT). Photophysical analysis of MDPT
revealed that it exhibits two well-defined maximums in its
absorbance spectra, an overlap peak of the mesityl and benzene
groups, and the 2,6-diarylpyrylium chromophore, located at 274
and 399 nm, respectively (Figure 3b). Interestingly, MDPT
lacks the 4-arylpyrylium chromophore commonly present in
TPT due to restricted rotation about the mesityl pyrylium bond
B
DOI: 10.1021/acs.orglett.7b01222
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
(Figure 3b). More striking is the emission spectrum which
features two maximums, the first ascribed as the singlet
transition at 438 nm and a broader, yet well-defined, peak at
590 nm (Figure 3b).15 From these spectroscopic data the
singlet excited state energy, at one-half the Stokes shift, is 2.96
V. Furthermore, cyclic voltammetry (CV) experiments were
conducted and MDPT was found to have an irreversible
reduction peak at Ep/2 = −0.34 V vs SCE in acetonitrile. Taking
together this data, with the Gibbs free energy of photoinduced
electron transfer equation, we have estimated the singlet excited
state redox potential of MDPT to be Es1red = +2.62 V vs SCE.16
These physical attributes make MDPT one of the strongest
visible light photoredox oxidants reported to date.17
We proceeded to explore the scope of this reaction (Figure
4a) and found that a variety of electron-deficient and -neutral
We also synthesized 4-mesityl-2,6-di-p-tolylpyrylium tetra-
fluoroborate (8, MD(p-tolyl)PT) with the goal of creating a
more red-shifted absorbing catalyst.16 This catalyst exhibits
absorbance and emission maximums at 420 and 479 nm,
respectively, and its singlet excited state redox potential is Es1
red
= +2.27 V vs SCE. This provides evidence that, like TPT, the
electronic properties of the catalyst can be tuned via structural
modifications.
The stage was now set for the investigation of the [3 + 2]
dipolar cycloaddition with a more electron-deficient system
such as 2,3-bis(4-chlorophenyl)oxirane. Simply by switching
the solvent to acetonitrile (MeCN) and maintaining the same
stoichiometry, MDPT promoted the cycloaddition to give 9 in
excellent yields with no apparent photobleaching observed over
24 h (Table 1, entry 1). Under the same conditions, TPT also
Figure 4. Scope of the [3 + 2] dipolar cycloaddition. a 0.005 mmol of
MD(p-tolyl)PT, 0.1 mmol of DMAD, 0.12 mmol of epoxide, MeCN.
b 0.05 mmol of MDP(p-tolylPT), 1 mmol of DMAD, 1.2 mmol of
epoxide, 0.5 M MeCN. c 1 mmol scale. d 1 mmol of epoxide and 5
mmol of DMAD.
Table 1. Investigation of the [3 + 2] Dipolar Cycloaddition
a
with an Electron-Deficient Epoxide
substrates can now undergo the reaction when using MD(p-
tolyl)PT. Halogenated and alkyl-containing compounds were
well tolerated (9−16). Examples with substitution at the para
and ortho positions reacted much faster than that of the meta.
We observed no dehalogenation in the crude reaction mixture
for any of these substrates. Higher diastereomeric ratios were
obtained for ortho substituted substrates (14−15), likely due to
enhanced A1,3-interaction. It was found that a tetrasubstituted
epoxide afforded a single regioisomer (17). A nonsymmetric
example bearing a trifluoromethyl group was obtained in good
yields and selectivity (18). Furthermore, a bis-halogenated
substrate, bearing orthogonal groups appropriate for iterative
cross-coupling reactions, afforded dihydrofuran (19) in good
yield.
entry
catalyst
yield (%)
b
1
2
3
4
5
6
MDPT, 24 h
TPT, 24 h
93
b
11
Mes-Acr-Me (5 mol %) or DCA (5 mol %)
MD(p-tolyl)PT, 1 h
0
b
>95
b
MD(p-tolyl)PT, 1 h (not degassed)
no light or no catalyst
>95
0
a
Standard conditions: 0.1 mmol of DMAD, 0.12 mmol of epoxide,
MeCN 0.5 M. Average isolated yield of two reactions.
b
After identifying a complementary method to access carbonyl
ylides from both electron-rich and -deficient diaryl epoxides, we
began exploring substrates that lacked one aryl substituent,
which have not been reported to participate in the reaction. We
were excited to observe that epoxides bearing dialkyl and aryl
substitution (Figure 4b) proceeded in moderate yields (20−
21). We then proceeded to investigate an alkynyl aryl epoxide.
Indeed, it underwent the desired cycloaddition with modest
yield and selectivity (22). These examples reacted much slower,
needing reaction times of 48 h.19 Even a vinyl aryl epoxide
afforded the cycloaddition (23) with modest yield and
selectivity and reacted much faster than all the previously
shown examples. Thus, in an effort to trap the carbonyl ylide
faster we made the epoxide the limiting reagent and we were
able to obtain the cycloaddition in good yields. Remarkably, no
isomerization of the olefin was observed. It is important to note
that DCA could not catalyze any of these aforementioned
examples. Overall, we found that careful choice of catalyst must
promoted the cycloaddition, but in low yields (Table 1, entry
2). This result validates our original hypothesis and shows the
significant improvement of the catalyst‘s robustness over TPT.
It is important to note that at this stage we cannot be sure
which excited state of the catalyst is responsible for the initial
SET; however, we are currently conducting detailed photo-
physical characterization of MDPT to better understand the
mechanism.18
Under these optimized conditions DCA and Mes-Acr-Me
failed to produce the desired product (Table 1, entry 3). We
also used MD(p-tolyl)PT and found that it promoted the
cycloaddition at a significantly faster rate, likely due to its
greater molar absorptivity in the optical window of use (Table
1, entry 4). Moreover, no degassing or drying of the MeCN is
needed with these catalysts (Table 1, entry 5). Control
experiments demonstrated that both light and catalyst are
needed to promote the cycloaddition (Table 1, entry 6).
C
DOI: 10.1021/acs.orglett.7b01222
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
(6) Albini, A.; Arnold, D. Can. J. Chem. 1978, 56, 2985.
(7) Wong, J. P. K.; Fahmi, A. A.; Griffin, G. W.; Bhacca, N. S.
Tetrahedron 1981, 37, 3345−3355.
(8) (a) Lee, G. A. J. Org. Chem. 1978, 43, 4256−4258. (b) Pan, J.;
Zhang, W.; Zhang, J.; Lu, S. Tetrahedron Lett. 2007, 48, 2781.
(9) Clawson, P.; Lunn, P. M.; Whiting, D. A. J. Chem. Soc., Perkin
Trans. 1 1990, 153−157.
(10) Futamura, S.; Kusunose, S.; Ohta, H.; Kamiya, Y. J. Chem. Soc.,
Chem. Commun. 1982, 1223−1224.
(11) (a) Kottisch, V.; Michaudel, Q.; Fors, B. P. J. Am. Chem. Soc.
2016, 138, 15535−15538. (b) Perkowski, A. J.; You, W.; Nicewicz, D.
A. J. Am. Chem. Soc. 2015, 137, 7580−7583. (c) Xuan, J.; Xia, X. D.;
Zeng, T. T.; Feng, Z. J.; Chen, J. R.; Lu, L. Q.; Xiao, W. J. Angew.
Chem., Int. Ed. 2014, 53, 5653−5656. (d) Hu, X.; Zhang, G.; Bu, F.;
Lei, A. ACS Catal. 2017, 7, 1432−1437. (e) Huo, H.; Harms, K.;
Meggers, E. J. Am. Chem. Soc. 2016, 138, 6936−6939. (f) Perkowski,
A. J.; Nicewicz, D. A. J. Am. Chem. Soc. 2013, 135, 10334−10337.
Perkowski, A. J.; Cruz, C. L.; Nicewicz, D. A. J. Am. Chem. Soc. 2015,
137, 15684−15687.
(12) (a) Fukuzumi, S.; Kotani, H.; Ohkubo, K.; Ogo, S.; Tkachenko,
N. V.; Lemmetyinen, H. J. Am. Chem. Soc. 2004, 126, 1600−1601.
(b) Lin, Y. C.; Chen, C. T. Org. Lett. 2009, 11, 4858−4861.
(13) Wilger, D. J.; Grandjean, J. M.; Lammert, T. R.; Nicewicz, D. A.
Nat. Chem. 2014, 6, 720−726.
be considered when targeting a particular product. When
dihydrofuran 1 was irradiated with 5 mol % of MD(p-tolyl)PT,
major decomposition was observed, providing evidence that the
products can be redox active.20
In summary, we have significantly expanded the scope of the
formation of carbonyl ylides from epoxides. Overcoming
current limitations led to the development of powerful organo
photoredox catalysts that are robust and possess some of the
highest oxidation potentials reported. These pyrylium catalysts
could find immediate application in already established
methods where more common catalysts such as Mes-Acr-Me
failed to produce a single electron transfer.
ASSOCIATED CONTENT
* Supporting Information
■
S
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.or-
glett.7b01222.
Experimental procedures for synthesis of substrates and
catalysts and for photoredox reactions; full character-
ization data and NMR spectra for all new compounds
(PDF)
(14) (a) Miranda, M. A.; Garcia, H. Chem. Rev. 1994, 94, 1063−
1089. (b) Martiny, M.; Steckhan, E.; Esch, T. Chem. Ber. 1993, 126,
1671−1682.
(15) Currently, we do not have sufficient data to assign this
transition. We speculate it may be a charge transfer state due to the
similarities to Fukuzumi’s catalyst.
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: beelera@bu.edu.
ORCID
(16) See Supporting Information for details.
(17) Other organic photoredox catalysts that possess higher
oxidation potentials absorb UV light. See ref 2 for a thorough inquiry.
(18) It is possible that the triplet excited state redox potential of
MDPT and MD(p-tolyl)PT, which would have a longer lifetime, could
be oxidizing enough to generate the desired carbonyl ylides.
(19) The epoxide and DMAD could be detected in the crude
material after the 48 h irradiation period.
Aaron B. Beeler: 0000-0002-2447-0651
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
(20) 1H NMR analysis of the crude material showed peaks
corresponding to DMAD and byproduct 10.
Financial support from Boston University and National Science
Foundation (ABB CHE-1555300) is gratefully acknowledged.
We thank Professor Scott Schaus for insightful discussions and
Dr. Norman Lee (Boston University) for high-resolution mass
spectrometry data. NMR (CHE-0619339) and MS (CHE-
0443618) facilities at Boston University are supported by the
NSF.
REFERENCES
■
(1) (a) Schultz, D. M.; Yoon, T. P. Science 2014, 343, 1239176.
(b) Bach, T. Angew. Chem., Int. Ed. 2015, 54, 11294−11295.
(2) Romero, N. A.; Nicewicz, D. A. Chem. Rev. 2016, 116, 10075−
10166.
(3) (a) Zhang, Y.; Petersen, J. L.; Milsmann, C. J. Am. Chem. Soc.
2016, 138, 13115−13118. (b) Rybicka-Jasinska, K.; Shan, W.; Zawada,
́
K.; Kadish, K. M.; Gryko, D. J. Am. Chem. Soc. 2016, 138, 15451−
15458. (c) Yin, H.; Jin, Y.; Hertzog, J. E.; Mullane, K. C.; Carroll, P. J.;
Manor, B. C.; Anna, J. M.; Schelter, E. J. J. Am. Chem. Soc. 2016, 138,
16266−16273. (d) Li, H.; Zhang, M. T. Angew. Chem., Int. Ed. 2016,
55, 13132−13136. (e) Luo, J.; Zhang, J. ACS Catal. 2016, 6, 873−877.
(4) (a) Fraga, B. M.; Diaz, C. E.; Guadano, A.; Gonzalez-Coloma, A.
J. J. Agric. Food Chem. 2005, 53, 5200−5206. (b) Esumi, T.; Hojyo, D.;
Zhai, H.; Fukuyama, Y. Tetrahedron Lett. 2006, 47, 3979−3983.
(c) Liao, H. B.; Huang, G. H.; Yu, M. H.; Lei, C.; Hou, A. J. J. Org.
Chem. 2017, 82, 1632.
(5) (a) Hojo, M.; Ohkuma, M.; Ishibashi, N.; Hosomi, A.
Tetrahedron Lett. 1993, 34, 5943−5946. (b) Hojo, M.; Aihara, H.;
Ito, H.; Hosomi, A. Tetrahedron Lett. 1996, 37, 9241−9244.
(c) Karnischky, L. A.; Arnold, D. R. J. Am. Chem. Soc. 1970, 92,
1404−1406. (d) Padwa, A. Acc. Chem. Res. 1991, 24, 22−28. (e) .
D
DOI: 10.1021/acs.orglett.7b01222
Org. Lett. XXXX, XXX, XXX−XXX