Z. Wang, et al.
MolecularCatalysis470(2019)1–7
bond of the formed compounds, which is attributed to the electron
donation effect from the clusters plasmon [39]. MALDI-TOF spectrum
of the precipitated Co nanoclusters was shown in Fig. 2E. Three main
peaks of 268.1566, 331.1778 and 346.2053 were found, which are
much greater than the atomic weight of Co and molecular weight of
cyclooctene and cyclooctene oxide, indicating the occurrence of the
oligomerization of the reactants or Co species. Based on FT-IR mea-
surement, C16H32O3Co and C16H31O4Co in Fig. S3B’ and Fig. S3C’ are
the metal-organic compounds containing alkene and hydroxyl/carbonyl
group such as alcohol/ketone. Mass fragments and fragmentation pat-
terns indicate that cyclooctene dimer oxide forms to anchor Co species.
The formation pathway was described as follows. Initially, C]C bond
of cyclooctene coordinates with Co salt [40] and α-CeH dissociates to
cyclooctene radical due to the weak bond [32,35,41]. After that, the
free radical interacts with cis-cyclooctene to generate the dimer, and
meanwhile Co cluster forms. Finally, oxygen gas is activated on Co
clusters surface into the reactive species which further oxidizes the
dimer. Water was observed in the mass fragment assigning for the
formed dimer oxide occurs in the interface between hydrous Co salt and
hydrophobic cis-cyclooctene. Therefore, the stabilizer bridges the mass
transportation between hydrophilic Co species and hydrophobic cis-
cyclooctene as an oil-soluble surfactant (Fig. 1A). Fig. 2F shows Co2p
XPS binding energy from 765 to 815 eV for Co clusters loaded on fumed
silica. The peaks around 777.1 and 789.8 are ascribed to Co(2p3/2) and
Co(2p1/2) of metallic Co species [42,43]. The binding energies and the
absence of satellite peaks suggest that metallic Co clusters occur in
these samples. However, it is possible that an oxidic cobalt present in
the sample is reduced by the X ray beam in the XPS. Noticeably, a large
peak around 770.9 eV was observed towards lower energy than metallic
cobalt, which could be ascribed to the electron donation of Co na-
noclusters promoted by the surrounding stabilizer. This feature pro-
motes oxygen gas activation through donating the electron from Co
nanoclusters to π* orbit of O2 [44], which is the key step in the aerobic
oxidation [35]. The fluorescence spectrum of the prepared cobalt
clusters is shown in Fig. S4. The emission scan was carried out at the
excitation wavelength of 410 nm which shows maximum intensity,
while the excitation scan was taken at the emission of 480 nm. For the
filtrate solution with no Co species, the emission peak intensity around
450 nm is low [32]. Thus, the observed spectra could be attributed as
arising from Co containing clusters. The excitation spectrum is sig-
nificantly blue shifted from the band ascribed to surface plasmon re-
sonance of larger Co nanoparticles, attesting to the small size of the Co
containing clusters. The formed surface plasmon is stabilized by the
formed dimer oxide, which assists the electron transfer during the ex-
citation of the light wave [39,45].
Based on the above analysis, mass transportation and electron do-
nation of Co clusters was promoted with the formed cyclooctene dimer
oxide, which might be used for aerobic oxidation of organic compounds
efficiently. The result of solvent-free aerobic oxidation of ethylbenzene
using in situ formed Co clusters catalyst are shown in red label of
Fig. 3A. Co containing nanoclusters show high activity with no induc-
tion period. With the time lasting, the ethylbenzene conversion and the
acetophenone (Fig. 3A, red curve) and 1-phenylethylhydroperoxide
yield (Fig. 3B, red curve) increase. The selectivity of acetophenone is
enhanced from 31% to 63%. The product distribution of ethylbenzene
oxidation is shown in Table S1. The carbon balance is good despite
evaporative losses, less than 5.0% around 50% conversion and
1.5% around 20% conversion. The control experiment, without the
catalyst, showed no products after 50 h of reactions. Similarly, low
conversion of ethylbenzene about 5% and low yield of acetophenone
about 1% were observed when 9 mg of CoCl2 salt was used as the
catalyst (Fig. 3A, blue curve) and interestingly no Co cluster was ob-
servable in the TEM (Fig. S5) after reaction despite the fact that the
solution contains 85 μg/ml of soluble cobalt. This indicates that cobalt
salt is poorly active for ethylbenzene oxidation under our mild reaction
conditions. In contrast, when 3 ml of a solution containing in situ
formed Co clusters derived from cis-cyclooctene oxidation reaction was
added to the ethylbenzene reaction solution (Fig. 3A, red curve), oxi-
dation reaction was observed with the introduction of O2. After 50 h,
ethylbenzene conversion reaches around 50% and the yield of acet-
ophenone is about 30%.
The hydroperoxide concentration in the in situ formed Co clusters
solution is around 0.3 mol/L as determined by triphenylphosphine ti-
tration [46]. When the same amount of tert-butyl hydroperoxide
(TBHP) was added as an initiator to the ethylbenzene oxidation solution
in place of the Co cluster catalyst [32], only 1% ethylbenzene conver-
sion results after 50 h, indicating that hydroperoxide is not an effective
promoter in ethylbenzene oxidation and Co clusters solubilized in the
solution plays the major role in dissociating α-CeH bond to form the
acetophenone of interest. The turnover number (TON) of acetophenone
is about 6000 after 50 h, calculated using the total number of cobalt
atoms in the reaction solution, which is around 110 times higher than
the one using cobalt salt as the catalyst.
The activity and stability of in situ formed Co clusters was further
investigated through cis-cyclooctene addition. When the prepared Co
clusters solution was diluted 10 times by cis-cyclooctene and 3 ml of the
diluted catalyst solution was added to the ethylbenzene solution, a 22%
conversion of ethylbenzene (Fig. 3C, orange curve) and a TON of
20,000 for acetophenone production were obtained after 50 h. This is
significantly higher than expected based on the amount of Co species.
With 100 times dilution by cis-cyclooctene, the conversion of ethyl-
benzene and the yield of acetophenone after 50 h decrease slightly to
19% and 12% respectively (Fig. 3C, purple curve), but the TON of
acetophenone reaches 1.4 × 105, suggesting that there might be other
species in the filtrate solution that are active for ethylbenzene oxidation
[30]. It is noticeable that 1-phenylethylhydroperoxide yield does not
increase after 20 h’ reaction in 100 times dilution sample (Fig. 3D,
purple curve). Previously we have shown that during cis-cyclooctene
oxidation, cyclooctene hydroperoxide is also formed that could initiate
hydrocarbon oxidation [32]. In order to delineate the roles between
cobalt clusters and other organics such as cyclooctene hydroperoxide,
we prepared a separate solution from cis-cyclooctene auto-oxidation
without any Co salt. When 3 ml of the auto-oxidation-generated solu-
tion was added to ethylbenzene without Co, the ethylbenzene conver-
sion is around 13% and the yield of acetophenone is about 6% after
50 h of reaction (Fig. S6). These are lower than when in situ formed
cobalt filtrate was used. Consequently, in situ formed Co clusters played
a significant role in ethylbenzene oxidation and TON of acetophenone
of around 7 × 104 was achieved after substracting the activity from the
organics in the filtrate, which is about 960 times higher than the direct
usage of cobalt salt (Fig. 3A). The excitation and emission spectra after
ethylbenzene oxidation reaction (Fig. S7) show no significant change as
compared with that before reaction (Fig. S4), indicating that the Co
clusters stabilized with the dimer oxide are active and stable under
reaction conditions.
According to the literatures [47–52], the ethylbenzene oxidation
proceeds by first formation of free radicals of ethylbenzene, which traps
a molecule of oxygen to generate a peroxide. Then, the product of
acetophenone is generated through the propagator of 1-phenyl-ethyl-
hydroperoxide. Thus, 1-phenyl-ethylhydroperoxide may be observed in
the oxidation reaction. However, small amount of 1-phenyl-ethylhy-
droperoxide was observed in 100 times dilution sample and the content
does not increase after 20 h’ reaction (Fig. 3D, purple curve). Al-
shammari found that 3 × 10−5 mol of 6-bis(t-butyl)-4-methylphenol
(BHT) radical scavenger was able to completely quench the Au/C cat-
alyzed oxidation of cyclopentene [53]. To test the role of free radicals in
the reaction, Irganox-1076, a radical scavenger was added to remove
the free radicals. When Irganox-1076 (˜ 3 × 10−5 mol) was added to
the ethylbenzene reaction after 25 h, the rate of ethylbenzene conver-
sion and the content of 1-phenyl-ethylhydroperoxide decrease notice-
ably (Fig. 3E, black curve). This phenomenon can be explained by the
scavenger which stops the conversion through phenylethyl radical. But
4