A phosphonium ylide as a visible light organophotoredox catalyst

Article abstract of DOI:10.1039/d1cc00996f

A phosphonium ylide-based visible light organophotoredox catalyst has been designed and successfully applied to halohydrin synthesis using trichloroacetonitrile and epoxides. An oxidative quenching cycle by the ylide catalyst was established, which was confirmed by experimental mechanistic studies.

Full text of DOI:10.1039/d1cc00996f

ChemComm
View Article Online
View Journal | View Issue
COMMUNICATION
A phosphonium ylide as a visible light
organophotoredox catalyst†
Cite this: Chem. Commun., 2021,
57, 3591
Yasunori Toda, *a Katsumi Tanaka,b Riki Matsuda,b Tomoyuki Sakamoto,b
b
a
a
Fuyuki Ito *c and Hiroyuki Suga
*
Received 22nd February 2021,
Accepted 8th March 2021
Shiho Katsumi,
Masahiro Shimizu,
DOI: 10.1039/d1cc00996f
rsc.li/chemcomm
A phosphonium ylide-based visible light organophotoredox catalyst there are no examples of phosphonium ylides that serve as visible
has been designed and successfully applied to halohydrin synthesis light organophotoredox catalysts.4a For this inquiry, we reasoned
using trichloroacetonitrile and epoxides. An oxidative quenching that a phosphonium ylide 1 with a p-conjugated system5 would be
cycle by the ylide catalyst was established, which was confirmed by suitable due to the following points: (1) 1 bears no protons a to the
experimental mechanistic studies.
phosphorus center, preventing undesired carbyne formation3d by
+
deprotonation from radical cation 1ꢀ ; (2) the spin density deloca-
+
The field of phosphonium ylide chemistry, having a history of lization and the steric hindrance of 1ꢀ may prevent undesired
over 100 years, has contributed greatly to the development of radical reactions; (3) the p-conjugated system is expected to
organic synthesis.1 The CQC bond forming reaction using increase the HOMO energy level of 1, enhancing reactivity for
phosphonium ylides, known as the Wittig reaction, is a well- electron acceptors; (4) the system can lower the HOMO–LUMO
studied transformation of carbonyl compounds to alkenes.2 energy gap of 1, which aids in visible light excitation. Herein, we
Not only the nucleophilic aspects of the carbanion but also the describe the exploration of organophotoredox catalysis by 1.
nature of the corresponding radical species have gained atten- Recently, we reported a metal salts-catalyzed photodecomposition
tion from researchers.3 In 1997, Janssen et al. reported of CCl3CN for halohydrin synthesis under visible light irradiation.6a
methylenephosphorane-derived cation radicals that were pre- Since CCl3CN is considered a good electron acceptor, this reaction
pared by one electron oxidation with the use of iodine, AgBF4, was selected as the initial model reaction to validate the photoredox
or electrochemical methods.3a Recently, Liu and co-workers catalysis via oxidative quenching of 1*. Taking into account the
ꢀ
were the first to use carbonyl-stabilized phosphonium ylides mechanism of the CCl2CN radical generation, the ylide catalysis
as reagents in photoredox catalysis, enabling (Z)-selective alkene was satisfactorily applied to trifluoromethylation reactions involving
synthesis or carbyne equivalent generation (Scheme 1a).3d,e ꢀCF3 radical intermediate.7 In addition, the first example of
A carbon radical attached to the phosphonium ion is invoked as
the key intermediate of the transformation. Therefore, it could be
feasible to establish photoredox catalysis4 by phosphonium ylides if
the photoexcited ylide is allowed to undergo photoinduced electron
transfer (PET) to electron acceptors (A) utilizing light energy
(Scheme 1b). PET is an initiative process in photoredox
catalysis, and the excited state oxidation potential of the catalysts
+
[E*ox(catꢀ /cat*)] is negative when the excited states act as reduc-
tants. Thermodynamically favorable PET is expected to occur
+
favorably when E*ox(catꢀ /cat*) is lower than the reduction potential
of the acceptor [Ered(A/AꢀÀ)]. To the best of our knowledge, however,
a Department of Materials Chemistry, Faculty of Engineering, Shinshu University,
4-17-1 Wakasato, Nagano 380-8553, Japan. E-mail: ytoda@shinshu-u.ac.jp
b Graduate School of Science and Technology, Shinshu University, Japan
c Department of Chemistry, Institute of Education, Shinshu University,
6-ro, Nishinagano, Nagano, 380-8544, Japan
† Electronic supplementary information (ESI) available. See DOI: 10.1039/
Scheme 1 Development of photoredox catalysis by phosphonium ylides.
d1cc00996f
This journal is © The Royal Society of Chemistry 2021
Chem. Commun., 2021, 57, 3591–3594
| 3591

View Article Online
Communication
ChemComm
Table 1 Optimization of reaction conditionsa
organocatalyzed direct C–H imidations of arenes under visible light
irradiation was demonstrated since arylamines are ubiquitous in a
range of biologically active compounds.8
We began our study with UV-vis absorption spectroscopy
experiments (Fig. 1). CCl3CN and epoxide 2a have no absorp-
tion bands in the visible light region, whereas phosphonium
ylide 4 (labs = 326 nm) exhibits an absorption band around
330 nm. In contrast, the band of 1 (labs = 370 nm) extends past
400 nm, suggesting that blue LED irradiation can be utilized for
photoreaction by 1. Next, the reducing power of 1 in the excited
state [E*ox(1ꢀ+/1*)] was estimated in accordance with eqn (1),
Entry
Catalyst
Solvent (0.1 M)
3ab (%)
1
2
3
4
5
6
7
8
1
1
1
1
1
1
1
4
MeCN : H2O = 99 : 1
1,4-Dioxane : H2O = 99 : 1
DMF : H2O = 99 : 1
NMP : H2O = 99 : 1
DMA : H2O = 99 : 1
DMF : H2O = 9 : 1
DMF
DMF : H2O = 99 : 1
DMF : H2O = 99 : 1
DMF : H2O = 99 : 1
DMF : H2O = 99 : 1
DMF : H2O = 99 : 1
DMF : H2O = 99 : 1
33
92
495 (91)
495
78
+
E*ox(1ꢀ /1*) = Eox(1ꢀ+/1) À E0,0
(1)
where Eox is the oxidation potential in the ground state
obtained from CV experiments (1: +0.72 V vs. SCE; 4: +1.12 V vs.
SCE3d) and E0,0 is the excitation energy based on emission
77
74c
54
9
None
24
energy (hc/lem
) obtained from the fluorescence spectra
10d
11d,e
12f
13g
1
1
1
1
0
(1 : 436 nm; 4 : 481 nm). We therefore calculated the E*ox values
of À2.36 V vs. SCE for 1* and À1.95 V vs. SCE for 4*.9 As
expected, the reducing power of 1* was higher than that of 4*. It
should be noted that the redox potential of 1* is similar to that
of 10-phenylphenothiazine in the excited state.4a Furthermore,
the reduction potential of CCl3CN was measured to be À0.28 V
vs. SCE, revealing the feasibility of PET from 1 to CCl3CN. We
then investigated the Gibbs free energy change in the single
electron transfer. The values of DGET = +23.1 kcal molÀ1
(ground-state electron transfer) and DGPET = À48.0 kcal molÀ1
(excited-state electron transfer) clearly indicated that the pro-
cess could take place only when in the photoexcited state.10
Having elucidated the photochemical properties of 1, we
tested the HCl addition to epoxide 2a via formation of the
ꢀCCl2CN radical with the use of 1 and CCl3CN as summarized
in Table 1. The initial experiment was performed in a 99 : 1
mixture of MeCN and H2O (0.1 M) with irradiation by blue LEDs
(40 W Kessil lamp, 50 mW cmÀ2) under argon atmosphere at
25 1C for 6 h (Table 1, entry 1). This resulted in the desired
halohydrin 3a being obtained, albeit in low yield. Switching the
solvent to 1,4-dioxane/H2O, DMF/H2O or NMP/H2O drastically
improved the yield of 3a (Table 1, entries 2–4). The amount of
H2O was important to achieve a higher yield (Table 1, entry 6).
Notably, 17% of byproduct 5a was observed in dry DMF, where
48
86
o5
a
Unless otherwise noted, all reactions were carried out with 2a
(0.2 mmol), catalyst (10 mol%) and CCl3CN (0.6 mmol) in 0.1 M
b
solution at 25 1C for 6 h under blue LED irradiation. Determined by
1H NMR analysis (isolated yield is shown in parentheses). 17% of 5a
c
d
e
f
was observed. In dark. At 100 1C. CCl4 was used instead of CCl3CN.
g
CHCl3 was used instead of CCl3CN.
3a seemed to be coupled with DMF (Table 1, entry 7).
A decreased yield was obtained with 4 (Table 1, entry 8). An
attempt without catalysts resulted in low conversion (Table 1,
entry 9), and the reaction under dark conditions led to recovery
of 2a (Table 1, entry 10). The thermal reaction at 100 1C gave 3a
in modest yield (Table 1, entry 11). CCl4 (Ered = À0.64 V vs. SCE)
and CHCl3 (Ered = À0.90 V vs. SCE) were less effective, pre-
sumably due to their lower reduction potentials (Table 1,
entries 12 and 13).11
The scope of substrates is summarized in Table 2. The HCl
addition to monoalkyl-substituted epoxides 2b–h was first
examined (Table 2, left box). Glycidyl ethers afforded the
corresponding chlorohydrins 3b–f in good yields. The regioisomer
of 3g was also observed in the reaction of 1,2-epoxyhexane (2g).
Enantiopure (S)-N-glycidyl phthalimide (2h) was smoothly converted
to (S)-3h with high enantiomeric excess without stereochemical
erosion. Oxirans 2i–k bearing two or three substituents were toler-
ated to furnish 3i–k with moderate to high yields (Table 2, center
box). In the cases of indene oxide (2l) and styrene oxide (2m), the
ring-opening regioselectivity was different from the alkyl ones
(Table 2, right box). Owing to the aryl group, the chloride ion could
attack the benzylic position of epoxides 2l and 2m.12 The same
tendency was observed for the reaction of aziridine 2n.
We obtained intriguing outcomes from control experiments.
The amount of HCl generated in situ can be estimated by Mohr’s
method, validating that 40% of CCl3CN was converted to HCl
under the standard conditions (Scheme 2a).6b On the other
hand, a radical trap experiment was successfully carried out
(Scheme 2b). TEMPO adducts 6 and 7 were detected by ESI-TOF
Fig. 1 UV-vis absorption spectra of ylide 1 (blue), epoxide 2a (gray),
CCl3CN (red), and 4 (purple) in MeCN (2.0 Â 10À4 M), and fluorescence
emission spectra of 1 (green) and 4 (light green) in MeCN (5.0 Â 10À5 M) at
room temperature.
This journal is © The Royal Society of Chemistry 2021
3592
| Chem. Commun., 2021, 57, 3591–3594

View Article Online
ChemComm
Communication
Table 2 Scope of HCl additiona
a
Unless otherwise noted, all reactions were carried out with
2
(0.2 mmol), 1 (10 mol%), and CCl3CN (0.6 mmol) in 0.1 M DMF/H2O
(99 : 1) at 25 1C for 6 h under blue LED irradiation. Isolated yields are
Fig. 2 (a) Stern–Volmer plots for quenching of 1 by fluorescence lifetime
measurements. (b) Proposed mechanism. (c) Spin density distributions
b
c
d
shown. Isolated after esterification. In 0.5 M DMF. 1 (20 mol%),
+
+
(isoval = 0.002) in the radical cations 1ꢀ and 4ꢀ calculated by DFT.
e
CCl3CN (5.0 equiv.) in 0.1 M 1,4-dioxane. 1 (20 mol%), CCl3CN
(5.0 equiv.) in 0.1 M MeCN.
electron transfer progress, kq was corrected in accordance with
eqn (2),
HRMS analysis, which supported the presence of radical cation
+
1ꢀ and an a-amino radical derived from DMF.
To gain more insight into the mechanism, we constructed
Stern–Volmer plots based on quenching experiments (Fig. 2a).
The excited state lifetime of 1 (t0) was measured at 2.0 ns in the
absence of a quencher. The steady decrease in excited state
lifetime was observed by increasing the amount of CCl3CN, while
epoxide 2a did not strongly affect the lifetime (Fig. S8, ESI†). The
Stern–Volmer constant (KSV) of 1 was determined to be 1.81 MÀ1
and thus, the quenching rate constant (kq = KSV/t0) was estimated
1/kq = 1/kdiff + 1/kact
(2)
where kact is the rate constant for activated quenching.14 The
kact value was therefore determined to be 9.63 Â 108 MÀ1 sÀ1
,
revealing that the rate of electron transfer is slower than that of
diffusion. In bimolecular electron transfer reactions, an initial
encounter complex is generated that leads to an active ion pair.
Our kinetic study suggested that the electron transfer step is
activation-controlled, while the formation of encounter
complex [1*Á Á ÁCCl3CN] proceeds as a diffusion-controlled pro-
cess. Based on the experimental evidence, we proposed a
mechanism for the catalytic HCl generation in DMF by 1
to be 9.16 Â 108 MÀ1 sÀ1. The diffusion-limited rate constant (kdiff
)
can be calculated as 1.9 Â 1010 MÀ1 sÀ1 (MeCN, 25 1C) by using
the Stokes–Einstein equation.13 In order for kinetic analysis of the
(Fig. 2b). First, visible light irradiation causes excitation of 1,
+
ꢀ
leading to generation of CCl2CN and 1ꢀ through an electron
ꢀ
transfer process. The generated CCl2CN abstracts a hydrogen
atom from DMF to provide a-amino radical I, which in turn is
oxidized by 1ꢀ+ to result in formation of iminium intermediate
II and regeneration of 1.15 Hydrolysis of II generates HCl,
whereas the reaction of II with halohydrin 3 gives byproduct
+
5. Moreover, DFT calculations with respect to radical cation 1ꢀ
were performed to predict spin density distributions (Fig. 2c).
+
+
Compared with 4ꢀ , spin densities of 1ꢀ can be delocalized
over the p system and this could be deemed one of the key
features for establishing the organophotoredox system.
Lastly, we attempted trifluoromethylation reactions to
empower the ylide-based photoredox catalysis (Scheme 3 and
Scheme S1, ESI†). Considering Àour mechanistic studies, elec-
tron acceptors having Ered(A/Aꢀ ) higher than E*ox(1ꢀ+/1*) are
expected to undergo PET to release active radical species via an
Scheme 2 Control experiments.
This journal is © The Royal Society of Chemistry 2021
Chem. Commun., 2021, 57, 3591–3594
| 3593

View Article Online
Communication
ChemComm
thankful to Mr Yoshifumi Mochizuki (Shinshu University) for
his kind support.
Conflicts of interest
There are no conflicts to declare.
Notes and references
1 (a) O. I. Kolodiazhnyi, Phosphorus Ylides, Wiley-VCH, 2007;
(b) A. W. Johnson, Ylides and imines of phosphorus, Wiley, 1993.
2 (a) B. E. Maryanoff and A. B. Reitz, Chem. Rev., 1989, 89, 863;
(b) P. A. Byrne and D. G. Gilheany, Chem. Soc. Rev., 2013, 42, 6670;
(c) M. Edmonds and A. Abell, in Modern Carbonyl Olefination, ed.
T. Takeda, Wiley-VCH, 2004, pp. 1–17.
3 (a) M. M. Wienk and R. A. J. Janssen, J. Am. Chem. Soc., 1997,
119, 5398; (b) T. W. Hudnall, C. L. Dorsey, J. S. Jones and
¨
F. P. Gabbaı, Chem. – Eur. J., 2016, 22, 2882; (c) T. Miura,
Y. Funakoshi, J. Nakahashi, D. Moriyama and M. Murakami, Angew.
Chem., Int. Ed., 2018, 57, 15455; (d) M. Das, M. D. Vu, Q. Zhang and
X.-W. Liu, Chem. Sci., 2019, 10, 1687; (e) M. D. Vu, W.-L. Leng,
H.-C. Hsu and X.-W. Liu, Asian J. Org. Chem., 2019, 8, 93; ( f ) See also:
X.-D. Xia, L.-Q. Lu, W.-Q. Liu, D.-Z. Chen, Y.-H. Zheng, L.-Z. Wu and
W.-J. Xiao, Chem. – Eur. J., 2016, 22, 8432.
Scheme 3 Trifluoromethylation and imidation reactions by 1.
4 For selected recent reviews, see: (a) N. A. Romero and D. A. Nicewicz,
Chem. Rev., 2016, 116, 10075; (b) C.-H. Wang, P. H. Dixneuf and
´
J.-F. Soule, Chem. Rev., 2018, 118, 7532; (c) L. Marzo, S. K. Pagire,
oxidative quenching cycle. If the radical can be trapped by CQC
bonds, various photoredox transformations might be accom-
plished. Hence, 8 (Ered = À0.37 V vs. SCE)7c was tested for
performance validation of 1. Indeed, trifluoromethylation of 9
provided 10 in 70% NMR yield (44% isolated yield after GPC).
In addition, 11 underwent trifluoromethylative lactonization to
afford 12 in 63% yield.7b Further synthetic application was
demonstrated in the direct arene C–H imidations using 13,
where ꢀNPhth could be generated through reductive cleavage of
the N–O bond by 1 under visible light irradiation (Scheme 3c).8
The imidations of 2,6-lutidine (14a), 2,4,6-collidine (14b), and
mesitylene (14c) successfully proceeded to give 15a–c in good
yields. 1,3,5-Trimethoxybenzene (14d) and benzene (14e) were
also tolerated to furnish 15d and 15e, respectively.
¨
O. Reiser and B. Konig, Angew. Chem., Int. Ed., 2018, 57, 10034.
5 (a) Y. Toda, T. Sakamoto, Y. Komiyama, A. Kikuchi and H. Suga,
ACS Catal., 2017, 7, 6150; (b) Y. Toda, K. Hashimoto, Y. Mori and
H. Suga, J. Org. Chem., 2020, 85, 10980.
6 (a) Y. Toda, K. Tanaka, R. Matsuda and H. Suga, Chem. Lett., 2019,
48, 1469; (b) Y. Toda, R. Matsuda, S. Gomyou and H. Suga,
Org. Biomol. Chem., 2019, 17, 3825.
˜
7 (a) S. Barata-Vallejo, B. Lantano and A. Postigo, Chem. – Eur. J., 2014,
20, 16806; (b) Y. Yasu, T. Koike and M. Akita, Chem. Commun., 2013,
49, 2037; (c) R. Tomita, Y. Yasu, T. Koike and M. Akita, Beilstein
J. Org. Chem., 2014, 10, 1099; (d) Y. Yasu, Y. Arai, R. Tomita, T. Koike
and M. Akita, Org. Lett., 2014, 16, 780; (e) R. Tomita, Y. Yasu,
T. Koike and M. Akita, Angew. Chem., Int. Ed., 2014, 53, 7144;
( f ) T. Koike and M. Akita, Acc. Chem. Res., 2016, 49, 1937.
8 (a) L. J. Allen, P. J. Cabrera, M. Lee and M. S. Sanford, J. Am. Chem.
Soc., 2014, 136, 5607; (b) C. B. Tripathi, T. Ohtani, M. T. Corbett and
T. Ooi, Chem. Sci., 2017, 8, 5622; (c) J. Jiao, K. Murakami and
K. Itami, ACS Catal., 2016, 6, 610.
In summary, we have developed a novel visible light orga-
nophotoredox catalyst. The catalytic system enabled not only
HCl generation from CCl3CN but also C–H functionalization
reactions. 1 behaves as an excited-state reductant, which was
9 E*ox values in DMF were also calculated: À2.33 V vs. SCE for 1*;
À2.02 V vs. SCE for 4*.
10 The Gibbs free energy changes are estimated according to eqn (3) and
À
(4), DGET = ÀF[Ered(A/Aꢀ ) À Eox(1ꢀ+/1)] (3) DGPET = ÀF[Ered(A/AꢀÀ) –
E*ox(1ꢀ+/1*)] (4) where F is the Faraday constant (23.061 kcal VÀ1 molÀ1).
unambiguously verified by mechanistic investigations. We 11 A. A. Isse, C. Y. Lin, M. L. Coote and A. Gennaro, J. Phys. Chem. B,
2011, 115, 678.
believe that the potential utility of phosphonium ylides as
photoredox catalysts will lead to fascinating visible light-
12 The reaction of 2l using 1.2 equiv. of HCl afforded a trans/cis mixture
of 3l (see ESI† for details).
triggered transformations.
13 S. L. Murov, I. Carmichael and G. L. Hug, Handbook of Photo-
chemistry, Marcel Dekker, 2nd edn, 1993.
14 (a) L. Troian-Gautier, M. D. Turlington, S. A. M. Wehlin,
A. B. Maurer, M. D. Brady, W. B. Swords and G. J. Meyer, Chem.
Rev., 2019, 119, 4628; (b) R. Bevernaegie, S. A. M. Wehlin,
E. J. Piechota, M. Abraham, C. Philouze, G. J. Meyer, B. Elias and
L. Troian-Gautier, J. Am. Chem. Soc., 2020, 142, 2732.
This work was partially supported by the Japan Society for
the Promotion of Science (JSPS) through a Grant-in-Aid for
Young Scientists (B) (Grant No. JP17K14483). We thank Prof.
Dr Naoki Kanayama (Shinshu University) for the light inten-
sity measurements. We also thank Prof. Dr Kennosuku Itoh
(Kitasato University) for fruitful discussions. We are extremely
15 Partial involvement of the radical chain process cannot be ruled out
in the reaction using CCl3CN.
This journal is © The Royal Society of Chemistry 2021
3594
| Chem. Commun., 2021, 57, 3591–3594