Selective endoperoxide formation by heterogeneous TiO2 photocatalysis with dioxygen

Article abstract of DOI:10.1016/j.tet.2018.03.040

We report the selective formation of endoperoxides by aerobic TiO₂ photocatalysis through the cyclic addition of dioxygen and a non-conjugated diene, the first heterogeneous catalytic system for endoperoxide synthesis. This green protocol does not require any additive and the photocatalyst is abundant and recyclable, providing a yield up to 64% and >20:1 diastereoselectivity. Mechanistic investigations were carried out by using product analysis, kinetic studies, O-18 labelling experiments, electron-spin resonance and a set of quenching experiments. Superoxide (but not singlet oxygen, triplet oxygen or peroxide) is directly involved in the reaction cascade to form the endoperoxide product. The new findings may be helpful for future for designing eco-friendly and energy sustainable strategies for selective oxygenation reactions using semiconductors, O₂ and sunlight.

Full text of DOI:10.1016/j.tet.2018.03.040

Accepted Manuscript
Selective endoperoxide formation by heterogeneous TiO photocatalysis with
2
dioxygen
Xiaofeng Qiao, Subharanjan Biswas, Wenli Wu, Feng Zhu, Chen-Ho Tung, Yifeng
Wang
PII:
S0040-4020(18)30305-3
10.1016/j.tet.2018.03.040
TET 29380
DOI:
Reference:
To appear in: Tetrahedron
Received Date: 7 February 2018
Revised Date: 18 March 2018
Accepted Date: 20 March 2018
Please cite this article as: Qiao X, Biswas S, Wu W, Zhu F, Tung C-H, Wang Y, Selective endoperoxide
formation by heterogeneous TiO photocatalysis with dioxygen, Tetrahedron (2018), doi: 10.1016/
2
j.tet.2018.03.040.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT
Graphical abstract
The selective formation of endoperoxides by aerobic TiO2 photocatalysis was
successfully realized. The mechanism involves the reaction of superoxide and a
distonic radical cation at TiO2 surfaces.

ACCEPTED MANUSCRIPT
Selective Endoperoxide Formation by Heterogeneous TiO2
Photocatalysis with Dioxygen
Xiaofeng Qiao, Subharanjan Biswas, Wenli Wu, Feng Zhu, Chen-Ho Tung, Yifeng Wang*
Key Lab for Colloid and Interface Science of Ministry of Education, School of Chemistry and
Chemical Engineering, Shandong University, Ji’Nan 250100, P. R. China
ABSTRACT: We report the selective formation of endoperoxides by aerobic TiO2 photocatalysis
through the cyclic addition of dioxygen and a non-conjugated diene, the first heterogeneous catalytic
system for endoperoxide synthesis. This green protocol does not require any additive and the
photocatalyst is abundant and recyclable, providing a yield up to 64% and >20:1 diastereoselectivity.
Mechanistic investigations were carried out by using product analysis, kinetic studies, O-18 labelling
experiments, electron-spin resonance and a set of quenching experiments. Superoxide (but not singlet
oxygen, triplet oxygen or peroxide) is directly involved in the reaction cascade to form the endoperoxide
product. The new findings may be helpful for future for designing eco-friendly and energy sustainable
strategies for selective oxygenation reactions using semiconductors, O2 and sunlight.
KEYWORDS: Heterogeneous photocatalysis, Oxygenation, Selective oxidation, Solar synthesis,
Cascade reaction
1

ACCEPTED MANUSCRIPT
1. Introduction
There has been a considerable interest in synthesis of endoperoxides because they are not only
promising antimalarial, anticancer and antiviral agents, but also precursors of 1,4-diols, –hydroxyketones
γ
and similar substitution patterns that are difficult to access using standard enolate chemistry.1-6 Early work
on endoperoxide synthesis used cycloadditions of dienes with singlet oxygen produced by
photosensitization.6-10 Most recently, intriguing work by Yoon5 and Nicewicz6 reported highly efficient
synthesis of endoperoxides using visible light photocatalysis; however, their photocatalysts are expensive
Ru-complexes and harmful pyrylium salts, respectively (Scheme 1A). Cyclization of hydroperoxides
leading to endoperoxides was also explored,11-15 but these approaches suffer severely from the use of
expensive and toxic transition metals like Pd and Au, ecotoxic ligands like PPh3 and bipyridine, harsh
reaction conditions, and/or multisteps.11-15 Also, all the above protocols are based on homogeneous
catalysts which are not recyclable. By contrast, semiconductor photocatalysis,16-22 which adopts low-cost,
recyclable and environmentally-friendly metal-oxide catalysts, and uses costless sunlight and atmospheric
dioxygen as the reagents, is an ideal alternative to the current homogeneousely catalyzed
endoperoxidation reactions. However, there has been no report on heterogeneous photocatalytic syntheses
of endoperoxide products to date.
Scheme 1. Photocatalytic endoperoxide synthesis using light and dioxygen as the reagents.
As one of the most important heterogeneous photocatalysts,16-22 TiO2 has long been studied in synthetic
2

ACCEPTED MANUSCRIPT
chemical transformations.23,24 However, the highly oxidative holes (hVB+, ca. +2.7 V) generated at the
photo-excited TiO2 valence band and the ꢀOH radicals (ca. +2.8 V) generated from H2O2 and H2O, can
non-selectively oxidize substrates and intermediates; and meanwhile, the more polar intermediates which
are usually better adsorbed at the surfaces of TiO2 are prone to oxidative degradation (Scheme 1B). These
features are beneficial for the destructive oxidation of organic molecules into CO2, water and mineral
acids in organic pollutant degradation, but leads to unwanted over-oxidation and degradative C-C
cleavage when TiO2 photocatalysis is adopted for synthetic purposes.25 Singlet oxygen, which is a useful
reagent in homogeneous selective oxygenation, could be generated by energy transfer from the photo-
3
excited TiO2 nanoparticles to ground-state O2, but its quantum yield is so low that it is a not practical
reagent for applying in oxygenation reactions in TiO2 photocatalysis.26 Superoxide (O2 ) is another active
–
oxygen species in TiO2 photocatalysis, generated by 1e-reduction of O2 which is the primary step for O2
–
to engage into the photocatalytic reaction. Although O2 had been proposed to react at the photo-excited
TiO2 nanoparticle surface with short-lived charged radical species like the benzene radical cation,27
selective formation of the corresponding peroxidation products was not reported, and by contrast, a most
–
recent study by Zhao et al. pointed out that O2 may be inert and it is to be further converted into
ꢀ
OH
–
radicals to react with a substrate.28 Despite these debates, we envisioned that if O2 can be properly
adopted in a selective oxygenation/peroxidation reaction, the quantum yield of incident photons should be
very high since this needs only one step to activate O2 and avoids highly active ꢀOH radicals, and TiO2
photocatalytic cascade reaction29 may be developed into a useful tool for selective endoperoxide
formation.
In this article, we report the use of TiO2 nanoparticles for photocatalytic selective synthesis of
endoperoxides. This protocol uses only TiO2, O2 and the substrate without any additive, and the catalyst
can be readily recycled for reuse. By this protocol, sequential C-C bond formation and peroxidation
reaction are integrated in a cascade which is rarely seen in heterogeneous photocatalysis. Different from
homogeneous systems, herein O2– is the direct peroxidation reagent. The new findings may be helpful for
future for using TiO2 photocatalysis as a tool for selective oxygenation reactions in an eco-friendly and
energy-sustainable way.
2. Results and Discussion
2.1 Selective endoperoxide formation.
Under aerobic conditions, bis(styrene) 1, bearing both an electron-rich and an electron-deficient
counterpart connected via an ether moiety, selectively produces endoperoxide (1e; yield = 64%,Scheme 2)
3

ACCEPTED MANUSCRIPT
in the presence of photo-excited TiO2 nanoparticles (P25, 80% anatase and 20% rutile, average diameter =
1
25 nm). The diastereomeric ratio (d.r.) is > 20:1 as indicated by H NMR. From the conversion of the
substrate, a quantum yield of as high as 5.2% at 395 nm was calculated. No reaction was detected in the
dark (Figure S1) or in the absence of TiO2, indicating both light and TiO2 are essential for the
photocatalytic reaction. TiO2 showed remarkable stability and recyclability during the photocatalytic
reactions. We examined five consecutive catalytic runs, and excellent recyclability and negligible activity
loss were found for the TiO2 photocatalyst (Figure 1). This is understandable since TiO2 is well-known for
its high chemical stability. The turnover number (TON) of this catalytic reaction, defined as amount of
substrate divided by amount of TiO2, reached 2.4, showing TiO2 is a catalyst. Notably, only the surface
atoms of this nanocrystalline catalyst are used in the photocatalytic reaction, and for the 25 nm-diameter
commercial P25 TiO2, less than 1/104 of the atoms are located on the surfaces. Hence, the real TON
should be much larger than 2.4. In addition, the current protocol can also be carried out using sunlight
following the previous protocol.29,30 The yield of 1e reached 52% after 10 h outdoor reaction, showing the
potential of using TiO2 and sunlight in synthesis of the endoperoxides. However, for convenience, all the
following results were obtained using 395 nm LED.
Scheme 2. The model photocatalytic reaction. Conditions: 0.15 mmol 1, 25 mg TiO2, 2.0 mL
acetonitrile, 1.0 atm O2, 0 °C, 30 min.
4

ACCEPTED MANUSCRIPT
70
62
60
59
60
50
40
30
20
10
0
57
57
1
2
3
Catalytic run
4
5
Figure 1. Yields of 1e as functions of catalytic runs.
We then examined the synthetic potential of the TiO2 photocatalysis in endoperoxide formation. For a
series of bis(styrene) bearing both an electron-rich and an electron-deficient counterparts, just analogous
to 1, the current TiO2 photocatalytic system shows good synthetic power to give the corresponding
endoperoxide derivatives (substrates 1–5 of Table 1). This is different from our previous study, in which a
set of electron-rich, symmetrical bis(styrene)s were used as substrates of TiO2 photocatalysis and
tetralones were the major products.29 Meanwhile, the linker of the two counterparts of a substrate can be
extended to other functional moieties like diethyl malonate (6). For these substrates, byproducts like γ-
hydroxyketone (f and g) and benzaldehydes were also detected. Their relationship with the major product
and the mechanism of endoperoxide formation are focused in the following sections.
Table 1. Outcome of the substrates in TiO2 photocatalysis.
Product yield (%)
Substrate
f+g
Total
e
64
13
77
1
2
3
4
5
54
47
50
50
n.a.a
15
54
62
79
50
29
n.a.a
5

ACCEPTED MANUSCRIPT
41
n.a.a
41
6
a n.a. means yield was not quantified.
2.2 Product analysis and kinetic profiles.
We analyzed the products of 1 by TiO2 photocatalysis, as shown in Scheme 3 and Figure 2. Apparently,
among the identified products, 1a, 1b and 1c were formed via breakage of the C=C bonds of 1 due to
degradative over-oxidation, which has been a common phenomenon in TiO2 photocatalysis. The
cyclobutane derivative, 1d, was formed by photocatalytic oxidative intramolecular [2+2]
cycloaddition.29,31 The endoperoxide (1e) and the
γ-hydroxyketone (1f and 1g) moieties were formed by
the tandem reaction of intramolecular C-C coupling followed by oxygenation.
Scheme 3. Products of bis(styrene) 1 by TiO2 photocatalysis (reaction for 30 min).
1e
1a
1b
1g
1f
1d
1c
10
20
30
40
Retention time (min)
Figure 2. HPLC spectrum of a reaction solution of 1 during the TiO2 photocatalytic reaction (detection
wavelength = 254 nm). For characterization of the compounds, see Supporting Information.
The kinetic profiles of the photocatalytic reaction of 1 were then studied to investigate the correlations
and the possibility of interconversion among compounds 1a–g. As shown in Figure 3, the substrate, 1,
shows a nearly exponential decay, typical of pseudo-first order photocatalytic reaction occurring at the
6

ACCEPTED MANUSCRIPT
TiO2–solvent interface. The yield of the major product, 1e, increased with reaction time and reached its
maxima (64%) at ca. 30 min. Further reaction leaded to decrease of the yield of 1e and emergence of
some unidentified species (e.g., those corresponding to the peaks at 24 – 32 min in Figure 2). Both 1a and
1b were rapidly formed in the beginning of the photocatalytic reaction, suggesting both were derived
directly from the substrate. After ca. 50 min, when the substrate 1 was completely consumed, amounts of
both 1a and 1b were still increasing, suggesting there were other pathways for their formation, e.g.,
+
photocatalytic degradation of 1e–g. The benzoic acid, 1c, could be derived via oxidation of 1a by hVB
.
No interconversion of 1e and 1f/1g can be inferred from the kinetic profiles. In fact, they are derived from
the substrate 1 through parallel pathways (shown later). The amount of total carbon was found to decrease
as the photocatalytic reaction proceeded, which is reasonable because the unidentified species are not
accounted.
A
B
100
80
60
40
20
0
12
8
1e
1
1a
1b
1c
1f
4
0
0
50 100 150 200 250
0
50 100 150 200 250
Time (min)
Time (min)
C
100
80
60
40
20
0
total carbon
0
50 100 150 200 250
Time (min)
Figure 3. Kinetic profiles of the photocatalytic reaction. Conditions: 0.15 mmol 1, 25 mg TiO2, 2.0 mL
acetonitrile, 1.0 atm O2, 0 °C.
2.3 The tandem reaction pathways.
In a control experiment, it was found out that the cyclobutane compound 1d was quickly (within ca. 25
min) converted to generate the endoperoxide 1e with 73% yield, together with complementary amounts of
1f and 1g (total yield of 1f and 1g is ca. 26%; Scheme 4). Hence, the cyclobutane 1d is the key
intermediate for the formation of 1e (as well as of 1f and 1g). This is consistent with the HPLC data
7

ACCEPTED MANUSCRIPT
which shows concentration of 1d was very low throughout the reaction (Figure 2).
Scheme 4. Conversion of cyclobutane (1d) to endoperoxide (1e) and γ-hydroxyketone (1f/1g) by TiO2
photocatalysis.
The above evidences infer that 1e, 1f and 1g are derived from the same active radical species, i.e., the
radical cation 1•+ (Scheme 4), which is resulted from oxidative single-electron transfer (SET) of the
substrate (1) followed by rearrangement,29,31 or formed via oxidative SET of the cyclobutane moiety of
1d.32,33 Additional control experiments also support this mechanism for initiating the photocatalytic
reaction of 1 and formation of 1e. First, isopropanol was used to examine the role of hVB+, since
+
isopropanol can effectively quench hVB and thereby inhibit the photocatalytic oxidations and in turn
promote the reduction pathways. It was found out that with added 2 eq of isopropanol the conversion of 1
+
was completely quenched, indicating that the present photocatalytic reaction was initiated by hVB
oxidation. Next, for a series of solvents, it was found that the yield of 1e decreased as the solvent polarity
decreased, e.g., MeCN (64%) > MeNO2 (53%) > Et2O (48%) > toluene (38%), for 100% conversion of 1.
This is consistent with the higher chemical stability of the radical cation 1•+ in a more polar solvent,29,34
which is beneficial not only for its formation but also for the useful further oxygenation reactions.
We then examined if there was any interconversion between 1e and 1f/1g, since it is known that an
endoperoxide could produce the corresponding γ-hydroxyketone nearly in quantitative amount via the
Kornblum–DeLaMare rearrangement in the presence of triethylamine.5 For this, 1e was subjected to the
TiO2 photocatalytic reaction; however, negligible conversion of 1e was noticed (3 h reaction) and
meanwhile 1f/1g was not detected. In addition, when isopropanol was added to quench hVB+, 1e was not
found to react with eCB–. The above evidences are consistent with the kinetic profiles shown in Figure 3,
which do suggest that 1e was not an intermediate during the formation of 1f/1g. Therefore, 1e and 1f/1g
were derived directly from the radical cation 1•+ through independent and competing kinetic pathways.
2.4 O-18 labelling experiments.
8

ACCEPTED MANUSCRIPT
We speculate that the residual H2O in the system (e.g., H2O molecules bound to the TiO2 surfaces)
served as a nucleophile to react with the cationic center of 1•+, which thereby resulted in the formation of
the γ-hydroxyketone (1f/1g). To examine this, 5 eq of H2O was deliberately added to the reaction solution,
and it was found that the yield of 1e was slightly decreased to 54% and meanwhile the total yield of 1f
and 1g was greatly enhanced to 28% (yield = 13% without added H2O). Using 18O-labeled water (5 eq) as
the additive, it was found out that >99% of the hydroxyl moiety of a γ
-hydroxyketone was from H218O
(Scheme 5). These observations further confirms the involvement of the radical cation 1•+ in the
photocatalytic reaction (see Scheme 4).
Scheme 5. Oxygen sources for formation of γ-hydroxyketone. See Figure S3 and discussion for HR-
ESI-MS spectra and data analyses.
2.5 Engagement of superoxide in the formation of 1e.
–
–
To probe the role of O2 in the current TiO2 photocatalytic model reaction, formation of O2 and its
–
reaction with 1•+ was investigated by ESR spectroscopy. Since O2 is a transient, short-lived species, a
–
spin trap DMPO was used to capture it (resulting in the DMPO-O2 adduct) for ESR measurements. As
shown in Figure 4A, while no signal was detected before light irradiation (curve a), a six-peak signal
emerged with light-excitation of the TiO2 suspension without added substrate (curve b). Such a
–
–
characteristic signal of DMPO-O2 adduct in acetonitrile indicates generation of O2 by the reduction of
–
dissolved triplet O2 molecules by eCB–. The concentration of the DMPO-O2 adduct gradually decreased as
more and more substrate (1) was added to the system (Figure 4B), indicating competition of 1•+ (direct
-
reaction of O2 and 1 can be ruled out since isopropanol quenched the conversion of 1 in Section 3.3) and
–
–
DMPO in capturing O2 . Note that the second order rate constant of DMPO and O2 is as large as 104 M-2
–
s-1.35 Hence capture of O2 at the TiO2 surfaces by 1•+ should be comparably fast.
9

ACCEPTED MANUSCRIPT
A
a
dark
b
without 1
c
with 1
simulated
d
3320
3340
3360
3380
3400
Field (Gauss)
B
100
90
80
70
60
50
0.00
0.05
0.10
0.15
0.20
Added [1] (M)
Figure 4. ESR measurements. (A) The spectra of an acetonitrile suspension of TiO2 with DMPO, (a)
before light irradiation; (b) after irradiation for 15 s; (c) with added 0.1 M 1 and irradiated for 15 s; (d)
–
simulated spectrum of DMPO-O2 adduct with hyperfine coupling constants of aN = 13.0 and aH = 9.20
Gauss. Conditions: 2 mL MeCN, 25 mg TiO2, 1.0 mM DMPO. Panel B shows change of the signal
–
intensity of DMPO-O2 adduct (based on total area of the six peaks) as functions of added [1].
–
Next, a set of quenching experiments were conducted to show O2 , but not any other [O]-carriers, is
directly involved in the endoperoxidation reaction.
–
Effects of superoxide dismutase (SOD) which can selectively quench O2 on the photocatalytic reaction
–
was investigated. On addition of SOD to quench O2 , conversion of 1 still reached 98% but no trace of 1e
was found, while the yield of the γ-hydroxyketone (1f/1g) was not affected (entry 2 of Table 2). Control
experiments ruled out the possibility of decomposition of the endoperoxide (1e) by added SOD. On the
other hand, since [Cp2Fe]BF4 can effectively scavenge eCB– 36 and thereby quenches superoxide formation
at the TiO2 surfaces, effect of [Cp2Fe]BF4 additive provides additional information for investigating a
photocatalytic reaction which is driven solely by hVB+. In contrast to SOD, upon addition of [Cp2Fe]BF4, 1
was completely converted, but both 1e and 1f/1g were not formed (Table 2, entries 1-3). In addition, with
either added SOD or [Cp2Fe]BF4, the benzaldehydes 1a/1b were the major products. The above SOD and
10

ACCEPTED MANUSCRIPT
–
[Cp2Fe]BF4 quenching experiments provide complementary evidences for engagement of 1•+ and O2 in
–
formation of 1e. Meanwhile, they indicate formation of 1f/1g requires eCB and formation of 1a/1b is a
+
process that is driven solely by hVB
.
Table 2. Reaction of 1 by the aerobic TiO2 photocatalysis under various conditions.
Yield (%)
Entry
Additives
Conv. of 1 (%)
1a + 1b a
1f + 1g
13
1e
1
2
3
4
5
6
7
a
No additive b
100
98
11
80
95
16
78
23
20
64
0
SOD, 2 mg
11
[Cp2Fe]BF4, 2 eq
4-oxo-TMP, 2 eq
100
95
0
0
43
0
10
α-terpinene, 2 eq
95
11
H2O2, 10 eq
93
31
35
17
tBuOOH, 5 eq
94
17
Amounts of benzaldehyde was routinely included; however, formation of benzaldehyde from a
styrene involves C=C breakage and hence, it is very complicated and discussion of its mechanism is
beyond the scope of the present study. Standard conditions: 0.15 mmol 1, 25 mg TiO2, 2.0 mL MeCN, 1
b
atm O2, 0 °C, 30 min.
Further discussion rules out the reaction of 1•+ and triplet O2 to form 1e in this heterogeneous system. It
was proposed in a homogeneous system (e.g., photocatalysis using Ru complexes as the photocatalysts5)
•+
that 1•+ could react with triplet oxygen to give an transient intermediate (Scheme 6; path A), i.e., 1O2 ,
which could rearrange and then abstract an electron from the catalyst to produce 1e. If this were the case
in the current system, from the observation that SOD could quench the formation of 1e, one could
•
conclude that SOD should be able to disproportionate peroxyl radicals, namely, both 1O2•+ and 1(OH)O2
(shown in Scheme 6) should be quenched simultaneously; hence both the endoperoxide (1e) and the γ-
hydroxyketone (1f/1g) would have been simultaneously quenched with added SOD. Obviously this
contradicts the experimental results. Hence 1e and 1f/1g must be formed through independent pathways
–
(as previously stated), i.e., 1e was formed by the reaction of 1•+ and O2 (neither of the paths in Scheme 6)
while 1f/1g was formed by the reaction of 1•+ and 3O2 (path B).
11

ACCEPTED MANUSCRIPT
Scheme 6. The conceivable pathways for reaction of 1•+ with 3O2 in photocatalytic reactions.
1
1
We also checked any role of O2 in the current TiO2 photocatalysis. For this, a O2 scavenger which
–
does not quench O2 , i.e., 2,2,6,6-tetramethyl-4-piperidone (4-oxo-TMP), was used as the additive.37
However, it was observed that 4-oxo-TMP showed little effect on the product yields (Table 2, entry 4),
1
indicating O2 was not the [O]-source for 1e. Alternatively, when
α
-terpinene was added (Table 2, entry
5),30,37 the GC-MS results indicated only the product of reaction between
α
-terpinene and O2 , i.e., 4-
–
isopropyltoluene. The product of
α
-terpinene and singlet oxygen, namely ascaridole, was not observed
(see Figure S4 and discussion). Meanwhile, formation of 1e was indeed quenched, just as expected since
-terpinene. Therefore, 1O2 was not involved in the present system. Also, from
–
O2 was captured by added
α
1
+
theoretical point of view, involvement of O2 could be ruled out because in the presence of an hVB
scavenger like 1, the energy transfer from excited TiO2 to 3O2 to produce 1O2 should be greatly inhibited.
Furthermore, we examined whether H2O2 and ROOH were responsible for the formation of any of the
identified products. Notably, peroxide species are common in TiO2 photocatalysis, and they are formed by
2e-reduction of O2 and by combination of two ꢀOH radicals. Meanwhile, transition-metal- and acid-
catalyzed cyclization of hydroperoxides could lead to endoperoxide formation in homogeneous
t
systems.11-15 However, in TiO2 photocatalysis, we found out that with added H2O2 or BuOOH, conversion
of 1 was not affected, while the yields of both 1e and 1f/1g were not enhanced (Table 2, entries 6 and 7).
Therefore, any role of H2O2 and/or ROOH was not observed in this study.
2.6 Reaction Mechanism.
The above multiple lines of evidence and discussion lead to a mechanism for the oxygenation routes of
1 at the photo-excited TiO2 nanoparticle surfaces (Scheme 7). The substrate 1 is activated by the oxidative
SET followed by a fast rearrangement to form 1•+. Dioxygen, on activation by eCB– at the surfaces of TiO2,
–
produces O2 which has a direct role in the formation of the endoperoxide product 1e. H2O and ground-
state O2 are involved in the formation of 1f/1g as the byproducts.
12

ACCEPTED MANUSCRIPT
Scheme 7. A proposed mechanism.
In the two previous reports, one of which chose a tris(bipyrazyl) ruthenium(II) complex and the other
chose a triarylpyrylium salt as the homogeneous photocatalyst to perform the syntheses of endoperoxides
3
from similar substrates,5,6 the mechanism was proposed to be the [4+2] cycloaddition of 1•+ and O2,5 or
1
the cycloaddition of 1 with O2.6 In the current heterogeneous photocatalytic system, we believe the
TiO2/solvent interface is essential since it provides a platform for the two short-lived species i.e., 1•+ and
–
O2 , to accumulate and to perform the cycloaddition, just consistent with the Langmuir–Hinshelwood
mechanism in heterogeneous catalysis. Meanwhile, the TiO2 surface may be beneficial for the stabilities
of both O2 and 1•+.
–
3. Conclusions
In conclusion, we have presented a green protocol for selective endoperoxide synthesis by using TiO2,
light and O2. The diastereoselectivity is > 20:1 and the quantum yield is as high as 5.2% using 395 nm
LED. The selective C-C coupling and endoperoxidation cascade at TiO2 surfaces under aerobic conditions
in this study is unprecedented. Ample evidences have been provided to indicate superoxide, but not
singlet oxygen, triplet oxygen or peroxide, was the sole [O]-source of the endoperoxide product. This
study may be helpful for future for exploring new applications of TiO2 photocatalysis in eco-friendly and
energy-sustainable oxygenation reactions.
4. Experimental Section
4.1 Materials.
13

ACCEPTED MANUSCRIPT
TiO
2
(Degussa P25) was provided by Evonik Specialty Chemicals (Shanghai) Co. Ltd, and used as
received. Superoxide dismutase (SOD;
≥1,400 units/mg dry weight) was purchased from Aladdin
company. 5,5-Dimethyl-1-pyrroline N-oxide (DMPO, 98%), cinnamic acid (98%), (E)-3-(4-
methoxyphenyl)acrylic acid (98%), (E)-3-(3,4-dimethoxyphenyl)acrylic acid (98%), and (E)-3-(4-
bromophenyl)acrylic acid (98%) were purchased from Energy Chemical. 2,2,6,6-tetramethyl-4-piperidone
(4-oxo-TMP) was purchased from J&K Scientific. Acetonitrile was purified by a PURE SOL system from
Innovative Technology made in USA, and THF was distilled with Na.
4.2 Instruments.
An LED lamp (395 ± 10 nm, 100 W) from a domestic manufacturer was used as the light source. The
light intensity irradiated on the reaction vial was measured using a light intensity meter (model CEL-
NP2000-10; Au light, China) and the value was found to be ca. 120 mW cm-2. Superoxide was detected
with added DMPO (1.0 mM) by a Shimadzu X320 electron spin resonance (ESR) spectroscope. 1H NMR
and 13C NMR spectra in CDCl3 at 20 °C were obtained using a Bruker 500 MHz Fourier transform NMR
spectrometer. Diastereomeric ratio (d. r.) for all compounds was determined by 1H NMR analysis, based
on the peak area of the isomers. GC-MS measurements were conducted using Shimadzu QP2010 SE
system equipped with a capillary column (HP-5, 30 m × 0.32 mm × 0.5 ꢁm). High-resolution electro-
spray ionization mass spectrometry (HR-ESI-MS) measurements were performed on an Agilent Q-TOF
6510 instrument. High performance liquid chromatography (HPLC) was performed on a Shimadzu LC-
20A system equipped with a photodiode array detector and Venusil XBP silica column (Agela
Technologies), using n-hexane and ethyl acetate as the two eluents: 20% ethyl acetate for 15 min, then to
60% in 15 min, and to 70% in 30 min. Some of the products were isolated by using a domestic semi-
preparative HPLC equipped with an Innoval silica column (5 ꢁm filler size, 30 mm × 250 mm) as follows:
15% ethyl acetate for 10 min, then to 27% in 6 min and maintained for 40 min, and finally to 40% in 70
min.
4.3 Syntheses of the substrates.
(E,E)-1,7-bis(4-methoxyphenyl)hepta-1,6-diene. The ethers were prepared according to procedures
described by Yoon et al.5,31 Alcohol (8.0 mmol in 9.0 mL of THF) was added dropwise into a mixture of
NaH (13.0 mmol) and anhydrous THF (3.0 mL), and the resultant mixture was cooled to 0 °C. Then the
bromide (10.0 mmol in 9.0 mL of THF) was added dropwise. The flask was returned to room temperature,
stirred overnight and then quenched by saturated ammonium chloride (aq). The aqueous phase was
extracted with Et2O twice. The organic phases were mixed together, washed with brine, and dried over
14

ACCEPTED MANUSCRIPT
Na2SO4. Column chromatography was used to obtain the desired pure product.
(E, E)-Diethyl 2,2-bis(p-methoxycinnamyl)malonate. Diethyl malonate (0.80 g, 5.00 mmol) was
added to a suspension of 60% NaH (750 mg, 18.6 mmol) in 15 mL THF. The mixture was stirred for 1 h
and then cooled to 0 °C. A solution of p-methoxycinnamyl bromide (2.85 g, 12.5 mmol) in 15 mL of
DMF was added to it. After 15 h, the reaction was quenched with water and the aqueous layer was
extracted with two portions of Et2O. Mixture of the organic fractions were dried over MgSO4 and
concentrated via rotary evaporation. Column chromatography was used to obtain the desired pure product.
4.4 General procedure for photocatalytic reactions.
In a 10 mL Pyrex reaction vial, 0.15 mmol substrate and 25 mg TiO2 were dispersed in 2.0 mL of
acetonitrile. The vial was sealed with a butyl stopper and purged with O2; an oxygen balloon was used for
continuous supply of O2 during the reaction. Afterwards, the vial was cooled with ice bath, magnetically
stirred and irradiated from the side. After certain intervals, an aliquot of the solution was withdrawn by a
syringe and analyzed by TLC and/or GC-MS. After completion of the reaction, the mixture was filtered
with celite to remove TiO2 and then the filtrate was either quantitatively diluted for HPLC analysis, or
concentrated to produce a residue which was purified by column chromatography for further analysis (e.g.,
GC-MS and NMR). Yield was quantified using HPLC based on the isolated pure compounds as the
authentic samples.
Acknowledgments
We thank the NSFC (Grant 21401117), the Natural Science Foundation of Shandong Province
(ZR2014BQ003), and Shandong University (104.205.2.5) for the financial support.
Appendix A. Supplementary data
Supplementary data related to this article can be found at https://doi.org/10.1016/xxxxxxx.
Corresponding Author
*E-mail: yifeng@sdu.edu.cn
References
15

ACCEPTED MANUSCRIPT
1. Schirmeister T, Oli S, Wu H, Della Sala G, Costantino V, Seo E-J, Efferth T. Marine Drugs
2017;15:63.
2. Yaremenko I A, Syroeshkin M A, Levitsky D O, Fleury F, Terent’ev A O. Med. Chem. Res.
2017;26:170.
3. Rudrapal M, Washmin Banu Z, Chetia D. Med. Chem. Res. 2017:DOI: 10.1007/s00044.
4. Rudrapal M, Chetia D, Singh V. J. Enzyme Inhib. Med. Chem. 2017;32:1159.
5. Parrish J D, Ischay M A, Lu Z, Guo S, Peters N R, Yoon T P. Org. Lett. 2012;14:1640.
6. Gesmundo N J, Nicewicz D A. Beilstein J. Org. Chem. 2014;10:1272.
7. Clennan E L. Tetrahedron 1991;47:1343.
8. Brill Z G, Condakes M L, Ting C P, Maimone T J. Chem. Rev. 2017;117:11753.
9. Ghogare A A, Greer A. Chem. Rev. 2016;116:9994.
10. Balci M. Chem. Rev. 1981;81:91.
11. Krabbe S W, Do D T, Johnson J S. Org. Lett. 2012;14:5932.
12. Rubush D M, Morges M A, Rose B J, Thamm D H, Rovis T. J. Am. Chem. Soc. 2012;134:13554.
13. Gemma S, Martí F, Gabellieri E, Campiani G, Novellino E, Butini S. Tetrahedron Lett. 2009;50:5719.
14. Harris J R, Waetzig S R, Woerpel K A. Org. Lett. 2009;11:3290.
15. Porter N A, Gilmore D W. J. Am. Chem. Soc. 1977;99:3503.
16. Kou J, Lu C, Wang J, Chen Y, Xu Z, Varma R S. Chem. Rev. 2017;117:1445.
17. Lang X, Chen X, Zhao J. Chem. Soc. Rev. 2014;43:473.
18. Chen X, Selloni A. Chem. Rev. 2014;114:9281.
19. Kisch H. Angew. Chem. Int. Ed. 2013;52:812.
20. Lang X, Ma W, Chen C, Ji H, Zhao J. Acc. Chem. Res. 2014;47:355.
21. Griesbeck A, Oelgemöller M, Ghetti F, Eds. CRC Handbook of Organic Photochemistry and
Photobiology, 3rd ed.; CRC Press: Boca Raton, 2012; Vol. 1.
22. Hoffmann M R, Martin S T, Choi W, Bahnemann D W. Chem. Rev. 1995;95:69.
23. Fox M A. Acc. Chem. Res. 1983;16:314.
24. Fox M A, Dulay M T. Chem. Rev. 1993;93:341.
25. Hoffmann N. Chem. Rev. 2008;108:1052.
26. Naito K, Tachikawa T, Cui S-C, Sugimoto A, Fujitsuka M, Majima T. J. Am. Chem. Soc.
2006;128:16430.
27. Bui T D, Kimura A, Ikeda S, Matsumura M. J. Am. Chem. Soc. 2010;132:8453.
28. Li Y, Wen B, Yu C, Chen C, Ji H, Ma W, Zhao J. Chem. - Eur. J. 2012;18:2030.
29. Liu Y, Zhang M, Tung C-H, Wang Y. ACS Catal. 2016;6:8389.
30. Liu Y, Wang B, Qiao X, Tung C-H, Wang Y. ACS Catal. 2017;7:4093.
16

ACCEPTED MANUSCRIPT
31. Ischay M A, Lu Z, Yoon T P. J. Am. Chem. Soc. 2010;132:8572.
32. Marquez C A, Wang H, Fabbretti F, Metzger J O. J. Am. Chem. Soc. 2008;130:17208.
33. Tojo S, Toki S, Takamuku S. J. Org. Chem. 1991;56:6240.
34. Lu Z, Yoon T P. Angew. Chem. Int. Ed. 2012;51:10329.
35. Finkelstein E, Rosen G M, Rauckman E J. J. Am. Chem. Soc. 1980;102:4994.
36. Schrauben J N, Hayoun R, Valdez C N, Braten M, Fridley L, Mayer J M. Science 2012;336:1298.
37. Chen Y-Z, Wang Z U, Wang H, Lu J, Yu S-H, Jiang H-L. J. Am. Chem. Soc. 2017;139:2035.
17