Communication
transition metals, which are known alcohol oxidation catalysts
in the presence of peroxides (Cr, Mn, Fe, Co, Ni, Cu, Zn, Mo, W,
Pb), as well as two alkaline-earth metals (Mg and Sr) for com-
only 2% conversion. Table 1 shows the conversion, selectivity
and yield for the different catalysts. Control experiments con-
firmed that <3% conversion was obtained without any cata-
lyst, and <5% conversion was obtained in the presence of
equivalent amounts of either Cundoped or a commercially avail-
able activated carbon (Ccomm) with a similar surface area
(1500 m2gÀ1). Even when impregnated with Cu or Co, either of
which could be expected to activate oxygen,[25] these carbons
gave much lower conversions compared with N:C.
parison. In each case, the loadings were about 1.3 mmolgÀ1
.
This loading was chosen based on the known nitrogen content
of the support (5.0 mmolgÀ1), giving a nitrogen:metal equiva-
lent ratio of 4:1. Thus, the nitrogens (which are responsible for
oxygen activation[8,9]) would outnumber the metal ions, and
some would remain unoccupied.
For our test reaction, we chose the oxidative dehydrogena-
tion of cinnamyl alcohol (1) to cinnamaldehyde (2, Table 1).
This alcohol substrate can undergo several different oxidations
(epoxidation, allylic alcohol oxidation, ring hydroxylation), as
well as combustion to CO, CO2 and water.[17,22] Moreover, the
aldehyde can react further on nitrogen-doped carbons, giving
esters or acids.[13,23,24] The variety of possible products makes
this a good reaction for testing selective oxidation.
The initial catalyst screening showed that Cu/N:C and Co/
N:C were especially active. Therefore, we focused on these cat-
alysts. First, we measured the intrinsic catalytic activity of
equivalent amounts of copper oxide and cobalt oxide on
Ccomm, as well as on Cundoped (Table 1, entries 18–19 and 21–22).
Cobalt oxide on carbon shows little or no activity, and al-
though copper oxide shows some activity, it is nowhere near
that of the Cu/N:C catalyst (see entries 21 and 5). Kinetic stud-
ies using the initial rates method showed that the reaction in
the presence of the Co/N:C catalyst was three times faster
than that in the presence of the plain N:C support (Figure S1).
To rule out the possibility that carbon is simply not the ideal
support for cobalt oxide, we synthesized an equivalent catalyst
on g-Al2O3.[26] Here, one might expect a much better activity as
the fit between the oxide support and the metal oxide crystals
would enable the formation of smaller (and thus more active)
particles.[26–29] The fact that this material gave <5% conversion
supports our hypothesis that the nitrogen-doped carbon is re-
sponsible for oxygen activation. We also studied the effect of
metal-oxide loading on the catalytic activity. All metal oxides
were prepared at two different loadings, 0.43 mmolgÀ1 and
1.3 mmolgÀ1. The higher loadings (shown in Table 1) gave con-
sistently higher yields by 20–30%. This suggests that the
number of available metal-oxide sites is limiting the reaction in
this loading range.
In a typical reaction, one equivalent of 1 was dissolved in
ethanol and charged to an autoclave under 1 atm of oxygen
(ca. 5 equiv oxygen per substrate) with 20 mg of catalyst
(2.6 mol% of transition metal, based on 1). The mixture was
stirred at 808C for 16 h, and the products were analysed by
gas chromatography (see the Supporting Information for full
experimental details). Control experiments run at 258C gave
Table 1. Catalytic oxidation of cinnamyl alcohol with molecular oxygen.[a]
Entry
Catalyst[b]
Conversion
of 1[c] [%]
Selectivity
of 2[c] [%]
Yield
of 2[d] [%]
1
2
3
none
Ccomm
Cundoped
N:C
2
5
5
49
84
87
94
90
96
93
79
94
97
91
94
95
91
92
93
90
90
91
81
89
92
84
1
4
5
One key question for the performance of every heterogene-
ous catalyst is that of possible leaching. To test for leaching,
we ran a simple filtration experiment (Figure 1) using the “cat-
in-a-cup” concept.[30] We ran the reaction for 5 h, then removed
the catalyst and let the reaction run for another 20 h, and then
re-introduced the catalyst. The stepped reaction profile in
Figure 1 shows that the reaction stops when the catalyst is
taken out. Moreover, the catalyst retains its activity after re-in-
troduction, as shown by the similar reaction rates (0.47 and
0.51 mmhÀ1, respectively).
To understand how this Co/N:C catalyst works, we character-
ized its composition and structure. X-ray photoelectron spec-
troscopy (XPS) showed that Co/N:C contains 7.3 at% nitrogen
and 1.4 at% cobalt (Table 1), both of which give an upper limit
for surface concentrations of 4.06 and 0.88 mmol/m2, respec-
tively. Furthermore, XPS revealed that the binding energies of
the nitrogen atoms (N 1s orbital) are unaffected by the cobalt
impregnation (Figure 2, top). This suggests that there is no sig-
nificant NÀCo coordination, albeit without precluding the
N···Co proximity required for cooperative catalysis. Moreover,
the Co (3/2p) XPS signals are the same for the N:C and the
Cundoped samples (Figure 2, bottom). Thus, the active sites are
not simple metal–nitrogen sites. Earlier studies hypothesized
4
17
56
48
55
80
31
27
22
23
20
19
14
13
16
11
8
15
50
46
51
63
29
26
20
22
19
17
13
12
14
10
6
5[e]
6[e]
7[f]
8[g]
9
Cu/N:C
Co/N:C
Co/N:C
Co/N:C
Ni/N:C
Fe/N:C
Pb/N:C
Mn/N:C
Zn/N:C
Cr/N:C
Mo/N:C
W/N:C
Mg/N:C
Sr/N:C
Co/Cundoped
Co/Ccomm
Co/g-Al2O3
Cu/Cundoped
Cu/Ccomm
10
11
12
13
14
15
16
17
18
19
20
21
22
22
4
4
10
25
3
4
9
21
[a] Reaction conditions: 1 atm O2, 1.0 mmol 1, 20 mg catalyst, 5 mL etha-
nol; stirred in an autoclave, 808C, 16 h. [b] Catalyst notation: [Metal ion]/
[dopant]:[support]. Example: Co/N:C designates a cobalt oxide on nitro-
gen-doped carbon; all metal loadings are 1.3Æ0.1 mmolgÀ1. [c] Based on
GC analysis (n-octane internal standard). [d] Conversionꢁselectivity.
[e] Results are the average of triplicate experiments. [f] 48 h reaction time.
[g] 120 h reaction time.
&
&
Chem. Eur. J. 2016, 22, 1 – 6
2
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!