Aerobic oxidation of cycloalkanes, alcohols and ethylbenzene catalyzed by the novel carbon radical chain promoter NHS (N-hydroxysaccharin)

Article abstract of DOI:10.1002/adsc.200303197

Replacement of Ishii's N-hydroxyphthalimide (NHPI) with the novel carbon radical chain promoter N-hydroxysaccharin (NHS) affords, in combination with metal salts, notably Co, or other additives, selective catalytic autoxidation of hydrocarbons, alcohols and alkylbenzenes under mild conditions (25-100°C, O₂ 1 atm). The effects of solvent, temperature and the nature of the additives were investigated to give an optimised oxidation protocol for the various systems. The NHS/Co combination was more reactive than NHPI/Co in the autoxidation of cycloalkanes. In contrast, the opposite order of reactivity was observed in the autoxidation of ethylbenzene and alcohols. It is suggested, on the basis of bond dissociation energy (BDE) considerations, that this is a result of a change in the rate-limiting step with the more reactive ethylbenzene and alcohol substrates. In the autoxidation of the model cycloalkane, cyclododecane, the best results (90% selectivity to a 4:1 mixture of alcohol and ketone at 24% conversion) were obtained with NHS/Co(acac)₃ in PhCF₃ at 80°C. Competition experiments revealed that, in contrast to what is commonly believed, formation of the dicarboxylic acid by ring opening is not a result of further oxidation of the ketone product. It is suggested that ring opened products are a result of β-scission of the cycloalkoxy radical formed via (metal-catalysed) decomposition of the hydroperoxide. This is suppressed in the presence of NHS (or NHPI) which efficiently scavenge the alkoxy radicals.

Full text of DOI:10.1002/adsc.200303197

FULL PAPERS
Aerobic Oxidation of Cycloalkanes, Alcohols and Ethylbenzene
Catalyzed by the Novel Carbon Radical Chain Promoter NHS
(N-Hydroxysaccharin)
Xavier Baucherel,a, b Luca Gonsalvi,a, c Isabel W. C. E. Arends,a Simon Ellwood,d
Roger A. Sheldona,*
a
Laboratory for Biocatalysis and Organic Chemistry, Department of Biotechnology, Delft University of Technology,
Julianalaan 136, 2628 BL Delft, Netherlands
Fax: ( ‡ 31)-15-2781415, e-mail: r.a.sheldon@tnw.tudelft.nl
b
Current address: Johnson Matthey Catalysts, Billingham, Cleveland TS23 1LB, UK
c
Current address: Istituto di Chimica dei Composti OrganoMetallici, Area di Ricerca CNR di Firenze (ICCOM-CNR),
Via Madonna del Piano, 50019 Sesto Fiorentino, Firenze, Italy
Quest International, Ashford, Kent TN24 0LT, UK
d
Received: November 11, 2003; Accepted: February 3, 2004
Abstract: Replacementof Ishii×s N-hydroxyphthali- ethylbenzene and alcohol substrates. In the autox-
mide (NHPI) with the novel carbon radical chain idation of the model cycloalkane, cyclododecane, the
promoter N-hydroxysaccharin (NHS) affords, in best results (90% selectivity to a 4:1 mixture of
combination with metal salts, notably Co, or other alcohol and ketone at 24% conversion) were obtained
additives, selective catalytic autoxidation of hydro- with NHS/Co(acac)3 in PhCF3 at80 8C. Competition
carbons, alcohols and alkylbenzenes under mild experiments revealed that, in contrast to what is
conditions (25 1008C, O2 1 atm). The effects of commonly believed, formation of the dicarboxylic
solvent, temperature and the nature of the additives acid by ring opening is nota resultof furhter
were investigated to give an optimised oxidation oxidation of the ketone product. It is suggested that
protocol for the various systems. The NHS/Co ring opened products are a result of ˚-scission of the
combination was more reactive than NHPI/Co in cycloalkoxy radical formed via (metal-catalysed)
the autoxidation of cycloalkanes. In contrast, the decomposition of the hydroperoxide. This is sup-
opposite order of reactivity was observed in the pressed in the presence of NHS (or NHPI) which
autoxidation of ethylbenzene and alcohols. It is efficiently scavenge the alkoxy radicals.
suggested, on the basis of bond dissociation energy
À
(BDE) considerations, that this is a result of a change Keywords: alcohols; C H activation; cycloalkanes;
in the rate-limiting step with the more reactive molecular oxygen; oxidation
Introduction
The introduction of electron-withdrawing groups in
the benzene ring ofNHPI was shown tohave abeneficial
The selective oxidation of saturated hydrocarbons is one effect on the catalyst performance in the aerobic
of the most challenging subjects in oxidation chemistry. oxidation of alkylbenzenes[3] and the electrocatalytic
In particular, homogeneous catalytic oxidation of hy- oxidation of alcohols.[4] We reasoned that the use of N-
drocarbons with molecular oxygen is very important in hydroxysaccharin (NHS; 1b),[5] in which one carbonyl
the chemical industry.[1] Recently Ishii and co-workers group (CO) is replaced by the more electron-with-
discovered that N-hydroxyphthalimide (NHPI; 1a) in drawing sulphonyl (SO2) group, could provide an even
combination with cobalt catalyses the autoxidation of more effective carbon radical chain promoter (CRCP).[6]
hydrocarbons under mild conditions (25 1008C, O2 In the course of our study on the oxidation of large ring
1 at m)[.2] The promoting effect of NHPI was explained cycloalkanes with molecular oxygen we found that NHS
on the basis of the mechanism shown in Scheme 1. NHPI was an efficient catalyst in combination with cobalt salts.[7]
is converted into its corresponding phthalimide N-oxyl In this paper we report details of the NHS-catalysed
(PINO) radical (2a) which is able to abstract a hydrogen oxidation of large ring cycloalkanes, ethylbenzene and a
atom from the organic substrate leading to longer wide range of alcohols. Mechanistic considerations based
propagation chains and, hence, higher rates and selectiv- on bond dissociation energies (BDE) provide a plausible
ities compared to conventional autoxidation protocols.
rationale for the different activities of NHS and NHPI.
286
¹ 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/adsc.200303197
Adv. Synth. Catal. 2004, 346, 286 296

Aerobic Oxidation of Cycloalkanes, Alcohols and Ethylbenzene
FULL PAPERS
reaction mixture showed the presence of two different
cyclododecadiones and two hydroxycyclododecanones
as the main by-products. The formation of these by-
products is attributed to competing transannular hydro-
gen abstraction by an intermediate cyclododecylperoxy
radical, analogous to that observed in the autoxidation
of cyclohexylbenzene.[8] Decreasing the reaction tem-
perature either with NHS or NHPI, led, as expected, to a
decrease of conversion together with an increase in
selectivity toward ketone and alcohol (runs 4 7). The
slightly higher activity already observed with the NHS
system at 1008C, became more pronounced at75 8C
(47% and 36% conversion with NHS and NHPI
respectively). At 508C NHPI completely failed to
promote substrate oxidation whilst NHS gave 42%
conversion, in 24 h, with an increase in overall selectivity
(ketone ‡ alcohol) to 61%.
Having shown that NHS is, in combination with
Co(acac)2, an effective catalyst for the oxidation of
cyclododecane into cyclododecanone under mild con-
ditions, we undertook a comprehensive study ofreaction
parameters in order to optimize the yield of ketone.
Several metal co-catalysts were first screened, in combi-
nation with NHS at different temperatures, using acetic
acid as solvent(Table 2). Co(OAc) 2 and Co(acac)2 gave
similar results. At 1008C Co(acac)3 gave similar results
to those obtained with Co(acac)2 but the activity of the
former was much more temperature-dependent as
shown in Figure 1. At50 8C a much longer initiation
period was observed with Co(acac)3. This resultcan be
Scheme 1.
Results and Discussion
Cycloalkane Oxidation
Under the standard conditions described by Ishii et al. explained by the need for a one-electron reduction of
[O2 1 at m, Tˆ 1008C in acetic acid, NHPI 10 mol % and Co(acac)3 into Co(acac)2, which is the actual co-catalyst
Co(acac)2 0.5 mol %],[2] cyclododecane 3 was oxidised (Scheme 2). This presumably involves one-electron
to cyclododecanone 4, cyclododecanol 5 and 1,12- oxidation of the acac ligand which is expected to be
dodecanedioic acid 6 in 29%, 5% and 30% selectivity, much faster at 1008C than at 508C.[9]
respectively, at 58% conversion (Table 1, run 3). Sub-
The Co(acac)2 could subsequently reactwith dioxygen
stitution of NHPI by NHS led to a slightly higher to form a superoxocobalt(III) species which abstracts a
conversion (64%) with comparable selectivity toward hydrogen from NHS (Scheme 2), as proposed by Ishii
the formation of the ketone and alcohol (31% and 8% for the Co/NHPI system.[2e] It is worth noting, however,
respectively) (run 2). GC-MS analysis of the crude that the oxidation of cobalt(II) by dioxygen is thermo-
Table 1. Oxidation of cyclododecane 3 in AcOH in the presence of Co(acac)2.[a]
Run
Co-catalyst
T [8C]
Time [h]
Conv.[b] [%]
Ketone[b, c] 4 [%]
Alcohol[b, c,d] 5 [%]
Diacid[c, e] 6 [%]
1
2
3
4
5
6
7
none
NHS
NHPI
NHS
NHPI
NHS
NHPI
100
100
100
75
75
50
24
6
6
8
8
<4
64
58
47
36
42
0
0
31
29
41
42
47
0
0
8
5
10
8
14
0
0
16
30
12
23
20
0
24
24
50
[a]
Experimental conditions: 0.5 mol % Co(acac)2, 10 mol % co-catalyst (0.3 mmol), 3 mmol 3, 1 at m O, 7.5 mL acetic acid.
Conversions and selectivities are determined by GC using 1,2,4-trichlorobenzene as internal standard.
Based on cyclododecane 3 reacted.
2
[b]
[c]
[d]
[e]
Alcohol 5 ‡ ester 7.
Determined by HPLC using octanoic acid as internal standard.
Adv. Synth. Catal. 2004, 346, 286 296
asc.wiley-vch.de
¹ 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
287

Xavier Baucherel etal.
FULL PAPERS
CoII(acac)2(HOAc)2. The HOAc ligand therein will
probably be very similar to that in the Co(OAc)2 system,
where it has been demonstrated that peroxy radicals,
peroxide, peracids etc. can easily enter the Co(II)
coordination sphere via the highly labile and very
weakly bonded HOAc ligand.[10,11] The replacementof
CoII by either CuII (runs 13 15) or FeIII (runs 16 and 17)
led to a lower activity without any significant increase in
selectivity to ketone.
The autoxidation of cycloalkanes involves the forma-
tion of the corresponding cycloalkyl hydroperoxide as
the primary oxidation product.[1] Its selective decom-
position is the key to a high selectivity toward ketone.
Chromium(VI) complexes are known to be efficient
radical initiators for hydrocarbon autoxidation and
effective catalysts for the decomposition of secondary
alkyl hydroperoxides to ketones.[12] The reaction was
therefore performed in the presence of 0.5 mol %
pyridinium dichromate (runs 18 and 19). The conver-
sion and overall selectivity (ketone ‡ alcohol) were
similar to those observed with Co(acac)2 but the ratio of
ketone to alcohol was greatly improved and reached
Scheme 2.
dynamically very unfavourable [redox potential Co(II)/ 16.8 and 28.5 at100 8C and 508C, respectively. Ruthe-
À
Co(III) around 1.8 V, for O2/O2 E0 ˆÀ 0.3 V]. For nium complexes are also well known catalysts for the
example, Partenheimer[10] reported that the formation decomposition of alkyl hydroperoxides. However, when
of cobalt(III) from cobalt(II) acetate in a refluxing Ru(CH3CN)4Cl2 was used as the co-catalyst the oxida-
oxygenated solution in acetic acid was initiated by trace tion of cyclododecane stopped after 2 hours at about
amounts of adventitious peroxides. We suggest that a 20 25% conversion at all temperatures, albeit with
more likely mechanism for initiation by Co(acac)3 similar selectivity (57 59%) toward the formation of
involves reaction of the initially formed acac ¥ radical ketone and alcohol. The formation of a black precipitate
with oxygen, affording the corresponding acacO2 ¥ spe- indicated that the ruthenium catalyst had decomposed.
cies which is scavenged by the NHS to afford the
From the results summarised in Table 2, CoII salts
corresponding radical (SINO) which enters the catalytic appeared to be the best co-catalysts in acetic acid.
cycle of Scheme 1 (see Scheme 2). In acetic acid as Hence, we subsequently used the combination CoII/
reaction medium the cobalt(II) is very likely present as NHS to study the aerobic oxidation of cyclododecane in
Table 2. Oxidation of cyclododecane 3 in AcOH in the presence of NHS.[a]
Run
Metal complex
T [8C]
Time [h]
Conv.[b] [%]
Ketone[b, c] 4 [%]
Alcohol[b, c,d] 5 [%]
8
9
Co(OAc)2 ¥ 4 H2O
100
75
50
100
50
100
75
100
100
75
100
50
6
8
24
5
24
8
10
6
6
24
6
24
2
2
64
50
45
64
22
55
28
45
31
24
58
45
21
24
33
43
52
30
48
32
34
31
33
38
42
57
32
36
8
11
13
9
29
9
25
14
26
35
3
10
11
12
13
14
15
16
17
18
19
20
21
Co(acac)3
Cu(OAc)2 ¥ 4H2O
Cu(acac)2
Fe(acac)3
(py)2Cr2O7
2
25
23
Ru(CH3CN)4Cl2
100
75
[a]
Experimental conditions, 0.5 mol % metal complex, 10 mol % NHS (0.3 mmol), 3 mmol 3, 1 at m O, 7.5 mL acetic acid.
Conversions and selectivities are determined by GC using 1,2,4-trichlorobenzene as internal standard.
Based on cyclododecane 3 reacted.
Alcohol 5 ‡ ester 7.
2
[b]
[c]
[d]
288
¹ 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
Adv. Synth. Catal. 2004, 346, 286 296

Aerobic Oxidation of Cycloalkanes, Alcohols and Ethylbenzene
FULL PAPERS
Figure 2. Oxidation of cyclododecane 3, in the presence of
NHS and Co(acac)3 or Co(acac)2, in PhCF3 atdifferent
temperatures.
Figure 1. Oxidation of cyclododecane 3, in the presence of
NHS and Co(acac)3 or Co(acac)2, in AcOH atdifferent
temperatures.
different solvents (Table 3). In acetonitrile and benzo- obtained when the Co(acac)2 was replaced by Co(OAc)2
nitrile (runs 22 and 23) conversions were lower than in (run 29). Replacementof Co(acac) with Co(acac)3, in
2
acetic acid, under comparable conditions, and selectiv- contrast, led to a higher reaction rate (35% conversion
ities were poor to moderate. Since ionic liquids are in 4 h) and a selectivity to ketone and alcohol of 64%
attracting increasing attention as media for catalytic and 16%, respectively (run 30). The best result (90%
processes,[13] we tested 1-butyl-3-methylimidazolium selectivity for a 4:1 mixture of ketone and alcohol at
and 1-hexyl-3-methylimidazolium tetrafluoroborates, 24% conversion) was observed with Co(acac)3/NHS in
[bmim][BF4] and [hmim][BF4], respectively, as solvent. PhCF3 at80 8C (run 32).
However, relatively low conversions were observed
(runs 24 and 25).
The course of cyclododecane oxidation in the pres-
ence of NHS and Co(acac)2 or Co(acac)3 at80 8C and
A reasonable activity (24% conversion in 8 h at 1008C is shown in Figure 2. It is interesting to note that,
1008C) together with good selectivity to cyclododeca- depending on the solvent, acetic acid or PhCF3, either
none (53%) ‡ cyclododecanol (26%) was observed in Co(acac)2 or Co(acac)3, respectively, displays the higher
chlorobenzene (run 26). Good results were also ob- activity in combination with NHS (compare Figures 1
served in a,a,a-trifluorotoluene (runs 27 and 28), a and 2).
solvent that is often used for aerobic oxidations.[14] At
We propose that this can be explained by assuming
1008C, 35% conversion was observed in 24 h, affording that, in PhCF3, NHS displaces an acac ligand in
cyclododecanone and cyclododecanol in 54% and 20% Co(acac)3 via nucleophilic substitution at cobalt, to
selectivity, respectively (run 28). Similar results were form a CoIII-NHS complex (Scheme 3). Decomposition
Table 3. Oxidation of cyclododecane 3 with oxygen in the presence of NHS in different solvents.[a]
Run
SoTlav[leynstCo-cat
8C]
Time [h]
Conv.[b] [%]
Ketone[b, c] 4 [%]
Alcohol[b, c] 5 [%]
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
CH3CN
PhCN
[C4mim][BF4]
[C6mim][BF4]
PhCl
PhCF3
PhCF3
PhCF3
PhCF3
PhCF3
PhCF3
PhCF3
PhCF3
PhCF3
PhCF3
Co(acac)2
Co(acac)2
Co(acac)2
Co(acac)2
Co(acac)2
85
24
24
24
24
8
24
24
24
24
4
10
8
8
6
24
28
16
7
11
24
9
16
35
26
35
24
32
20
18
36
35
18
traces
45
53
41
69
54
49
64
72
61
59
40
50
14
24
traces
5
26
33
28
20
30
16
18
9
100
100
100
100
80
‡
n-Bu4N BrÀ
Co(acac)2
Co(acac)2
Co(OAc)2
Co(acac)3
Co(acac)3
Cu(acac)2
Fe(acac)3
(py)2Cr2O7
Cr(acac)3
80
100
100
100
80
100
100
100
100
37
21
6
[a]
Experimental conditions, 0.5 mol % co-catalyst, 10 mol % NHS (0.3 mmol), 3 mmol 3, 9 mL solvent.
Conversions and selectivities are determined by GC using 1,2,4-trichlorobenzene as internal standard.
Based on cyclododecane 3 reacted.
[b]
[c]
Adv. Synth. Catal. 2004, 346, 286 296
asc.wiley-vch.de
¹ 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
289

Xavier Baucherel etal.
FULL PAPERS
Scheme 3.
of the latter, via one-electron transfer, affords the N-
oxyl radical and Co(acac)2. This ligand substitution
reaction is much less favourable in the protic solvent,
acetic acid, than in the apolar PhCF3.
Figure 3. Competition experiments with cyclododecane and
cyclopentadecanone.
The optimum conditions developed for cyclodode-
cane oxidation NHS/Co(acac)3 in PhCF3 at80 8C
were subsequently applied to the oxidation of cyclo- were separately analysed by HPLC. The NHS was
decane and cyclooctane (Table 4). In the oxidation of found largely unchanged (atleast75% could be
cyclododecane (run 37) no oxidative cleavage leading to recovered) which suggests the possibility of recycling
dodecanedioic acid was observed. The only by-products of NHS as a catalyst.
observed were a resultof rtansannular hydrogen
abstraction (see earlier). As a result of ring strain,
caused by transannular interactions, cyclodecane and By-Product Formation in Cycloalkane Oxidation
cyclooctane are more reactive than cyclododecane
towards autoxidation. Thus, cyclodecane gave 31% An interesting question with regard to cycloalkane
conversion in 5 h at100 8C, affording cyclodecanone oxidation is how dicarboxylic acids are formed via
(61%), cyclodecanol (24%) and 2% of 1,10-decanedioic oxidative cleavage of the ring. It is generally assumed
acid (run 38). The oxidation of cyclooctane was even that they are formed via consecutive oxidation of the
faster (run 39): 30% conversion in 1.5 h, affording cycloalkanone product.[1a] We tested this hypothesis by
cyclooctanone (55%), cyclooctanol (22%) and 1,8-oct- performing competition experiments in which a mixture
anedioic (20%). It is interesting to note that in the case of of cyclododecane and cyclopentadecanone, in a molar
oxidation of cyclododecane in PhCF3 as opposed to ratio of 5:1 (corresponding to ca. 17% conversion in a
acetic acid as solvent no oxidative cleavage leading to cycloalkane oxidation) was subjected to aerobic oxida-
dodecanedioic acid was observed.
tion (see Figure 3). The ketone was added either at t ˆ 0
The stability of N-hydroxysaccharin was determined or after the cyclododecane had been allowed to react for
in a separate experiment, by heating NHS in either 1 h, and the oxidation was allowed to proceed for a total
acetic acid or trifluorotoluene up to 1008C for 6 hours of 6 h.
under 1 atm O2. No loss of NHS occurred and the
When the oxidation was conducted with NHS/Co-
identity was verified by NMR. To study the stability of (acac)3 in PhCF3, the conversions of cyclopentadeca-
NHS under the reaction conditions, two reaction none were 25% and 12% when itwas added att ˆ 0 or
mixtures (in either acetic acid or trifluorotoluene) t ˆ 1 h, respectively. In neither case was any formation
Table 4. Oxidation of cycloalkanes in the presence of NHS.[a]
Run
Substrate
Time [h]
Conv.[b] [%]
Selectivity [%][b, c]
Ketone[d]
Alcohol
Diacid[e]
Total
37
38
39
Cyclododecane
Cyclodecane
Cyclooctane
10
5
1.5
24
31
30
72
61
55 (0)
18
24
22
0
2
20
90
87
99
[a]
Cycloalkanes (3 mmol), NHS (0.3 mmol), Co(acac)3 (0.015 mmol), PhCF3 (9 mL), 808C, O2 1 at m.
Conversions and selectivities are determined by GC using 1,2,4-trichlorobenzene as internal standard.
Based on cycloalkane reacted.
Between brackets 1,3-cyclooctadione.
Determined by HPLC.
[b]
[c]
[d]
[e]
290
¹ 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
Adv. Synth. Catal. 2004, 346, 286 296

Aerobic Oxidation of Cycloalkanes, Alcohols and Ethylbenzene
FULL PAPERS
Figure 4. Competition experiments with cyclododecane and cyclopentanedecanol.
of 1,15-pentadecanedioic acid observed, consistent with
the observation of no 1,12-dodecanedioc acid in the
oxidation of cyclododecane (see above). When the
oxidation was performed with NHS/Co(acac)2 in acetic
acid the ketone conversions were 43% and 24%,
respectively. No formation of the dicarboxylic acid was
observed in the latter case while in the former a small
amount( < 5%) was observed. We conclude from these
results that consecutive oxidation of the ketone product
is not the (primary) source of ring-opened by-products.
The observed disappearance of the ketone in the
competition experiments probably involves oxidation
À
atother C H bonds in the ring than those adjacent to the
ketone group.
We subsequently performed competition experiments Scheme 4.
with a mixture of cyclododecane and cyclopentadecanol
(see Figure 4). When cyclopentadecanol alone was
subjected to aerobic oxidation in the presence of NHS/ Alcohol Oxidation
Co(acac)3 in PhCF3 at100 8C for 6 h, cyclopentadeca-
none was obtained in 91% selectivity at 94% conver- The selective oxidation of alcohols to the corresponding
sion. When a 5:1 mixture of cyclododecane/cyclopen- carbonyl compounds is one of the most frequently used
tadecanol was subjected to the same conditions 98% of transformations in organic synthesis. In recent years,
the cyclopentadecanol was converted, affording cyclo- increasing attention has been focused on the selective
pentadecanone in 87% selectivity, together with 9% of aerobic oxidation of alcohols to carbonyl compounds
1,15-pentadecanedioic acid. We propose that, in cyclo- under mild conditions.[16] The use of N-hydroxyphthal-
alkane oxidations, ring-opened products are formed via imide (NHPI) in combination with Co(acac)3 as a
b-cleavage of intermediate cycloalkoxy radicals formed catalyst for the aerobic oxidation of alcohols in acetoni-
via (metal-catalysed) decomposition of the correspond- trile at 758C was described by Ishii.[17] The same group
ing hydroperoxide (Scheme 4). The formation of car- subsequently reported that the Co(acac)3 could be
bonyl compounds via b-cleavage is a well-documented replaced by Co(OAc)2 in the presence of various
reaction of alkoxy radicals,[15] which in the case of cyclic substituted benzoic acids in ethyl acetate at room
alkoxy radicals leads to ring opening. In the presence of temperature.[18] The NHPI-catalysed oxidation of 2-
NHS (or NHPI) efficientscavenging of alkoxy radicals
octanol under these two different conditions gave, in our
suppresses b-cleavage in favour of alcohol formation hands, 93% and 70% yield respectively (Table 5, runs 41
leading to a higher selectivity to alcohol ‡ ketone and 43). The substitution of NHPI by NHS under the
(Scheme 4). In contrast, alcohol autoxidation does not conditions developed by Ishii gave disappointing results
proceed via intermediate alkoxy radicals but primarily (10 and 12% yield respectively, runs 42 and 44). An
via a-hydroxyalkyl radicals, formed via hydrogen ab- increase of the temperature to 758C, however, allowed
À
straction from the a-C H bond. Hence, alcohols are the NHS-catalysed oxidation of 2-octanol in EtOAc in
selectively oxidised to the corresponding ketone with the presence of Co(OAc)2 and benzoic acid to take place
little or no oxidative cleavage. The promising results efficiently, and 2-octanone was obtained with 80% yield
obtained in the oxidation of cyclopentadecanol led us to (run 45). Moreover, the oxidation of 2-octanol could be
explore the NHS/Co(acac)3 catalysed aerobic oxidation successfully carried out under the conditions previously
of a range of alcohols.
described for the oxidation of cycloalkanes, e.g., NHS,
Adv. Synth. Catal. 2004, 346, 286 296
asc.wiley-vch.de
¹ 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
291

Xavier Baucherel etal.
FULL PAPERS
Table 5. Oxidation of 2-octanol catalysed by NHPI or NHS at various temperatures in different solvents.[a]
Run
Conditions
Co-catalyst
T [8C]
Time [h]
Conv.[b] [%]
Sel. to ketone[b] [%]
40
41
42
43
44
45
46c
Co(acac)3, PhCF3
Co(acac)3, CH3CN
Co(acac)3, CH3CN
Co(OAc)2, PhCO2H, EtOAc
Co(OAc)2, PhCO2H, EtOAc
Co(OAc)2, PhCO2H, EtOAc
Co(OAc)2, PhCO2H, EtOAc
NHS
NHPI
NHS
NHPI
NHS
NHS
NHS
100
75
75
25
25
75
75
24
24
24
24
24
24
24
84
93
11
70
12
85
94
94
100
91
100
100
94
95
[a]
Experimental conditions: 2-octanol (4.5 mmol), NHS or NHPI (0.45 mmol), Co-complex (0.022 mmol), PhCO2H
(0.22 mmol), solvent(7.5 mL), O 1 at m.
Conversions and selectivities towards ketone are determined by GC using 1,2,4-trichlorobenzene (200 mg) as internal
standard.
No stirring.
2
[b]
[c]
Table 6. Oxidation of 2-octanol in EtOAc catalysed by Co/NHS and additives.[a]
Run
Metal salt
Co(OAc)2
Additive[b]
T [8C]
Time [h]
Conv.[c] [%]
Sel. To ketone[c] [%]
47
48
49
50
51
52
53
54
56
none
75
75
100
75
75
75
75
100
75
24
24
2
24
24
24
24
5
57
85
89
94
45
90
97
93
78
95
94
89
93
98
92
95
99
95
PhCO2H
PhCO2H
MCBA
PMBA
none
PhCO2H
PhCO2H
PMBA
Co(acac)3
24
[a]
Experimental conditions, see Table 5.
MCBA: 3-chlorobenzoic acid, PMBA: 4-methoxybenzoic acid.
Conversions and selectivities are determined by GC using 1,2,4-trichlorobenzene as internal standard.
[b]
[c]
Co(acac)3 in PhCF3 at100 8C, resulting in 2-octanone in
94% selectivity at 84% conversion (run 40).
Differentsecondary alcohols were oxidised following
the procedure developed for 2-octanol [NHS 10 mol %,
We subsequently studied the effect of different Co(acac)3 0.5 mol %, PhCO2H 5 mol %, in EtOAc at
combinations of metal salt, additive and temperature 1008C] and the results are summarised in Table 7. Cyclic
on the aerobic oxidation of 2-octanol in the presence of aliphatic alcohols gave good to excellent results
NHS (see Table 6). The bestresultwas obtained at75 8C (runs 58 64). The oxidation of 2-adamantanol (en-
in EtOAc, with benzoic acid and Co(acac)3 (97% try 60) and d,l-borneol (entry 61) was very efficient:
conversion at 24 h, 92% selectivity to octanone, both adamantanone and camphor were obtained with 94
run 53). Raising the temperature to 1008C (run 54) and 92% yield, respectively, the latter needing a longer
resulted in 93% conversion in 5 h (selectivity 92%). The reaction time (10 h vs. 3 h for adamantanone) (runs 60
superior results obtained in the presence of benzoic acid and 61). Cyclohexanol and cyclododecanol were atfirst
can be explained as follows. The benzoic acid can oxidised at100 8C leading to a fast reaction and 90 94%
protonate the acetate ligand to give the weak, labile conversion after 5 to 6 hours of reaction (runs 58 and
acetic acid ligand:
59), but the corresponding ketones were obtained with
less than 80% yield (75 and 78%, respectively, for
cyclohexanone and cyclododecanone). The oxidation
was then performed at lower temperature (758C)
leading to a slower reaction rate but still very efficiently
and much more selectively. Cyclohexanol was oxidised
One might expect, therefore, to see an increased with 90% conversion leading to cyclohexanone with
activity with increasing acidity of the benzoic acid, which 87% yield (97% selectivity).
is whatwe observed: conversions decreased in the order
3-chlorobenzoic acid > benzoic acid > 4-methoxyben- alcohols (at100 8C) gave excellentresults (runs 62 64).
zoic acid (Table 6). 1-Phenylethanol and benzhydrol were oxidised within
As expected the oxidation of secondary benzylic
292
¹ 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
Adv. Synth. Catal. 2004, 346, 286 296

Aerobic Oxidation of Cycloalkanes, Alcohols and Ethylbenzene
FULL PAPERS
Table 7. Oxidation of different alcohols catalysed by NHS.[a]
entcondiiotns [NHS 10 mol %, Co(OAc)
2 ¥ 4 H2O
0.5 mol %, m-CPBA (m-chloroperbenzoic acid)
5 mol %, EtOAc, 758C] (runs 65 and 66). Oxidation of
1-octanol led to the formation of octanoic acid in 67%
yield at86% conversion. Benzyl alcohol gave very good
results with complete conversion after 4 hours, benzoic
acid being formed in 100% yield.
Ethylbenzene Oxidation
We also applied our new catalytic system to the
oxidation of ethylbenzene. In our hands oxidation of
ethylbenzene under optimised Ishii×s conditions, namely
Co(acac)2/NHPI, in acetic acid at 1008C under 1 atm of
molecular oxygen, gave 82% conversion in 6 hours,
affording acetophenone in 90% selectivity (Table 8,
run 67). Substitution of NHPI by NHS led to a decrease
in activity. The conversion reached 55% and acetophe-
none was formed in only 66% selectivity (run 68). When
performed in CH3CN at85 8C, the difference of
reactivity between NHPI and NHS was even larger
(runs 69 and 70). These results are comparable to those
obtained in alcohol oxidations, NHPI showing a higher
catalytic activity than NHS. The protocols developed for
the oxidation of cycloalkanes [e.g., Co(acac)3/NHS in
PhCF3 at100 8C] and alcohols [Co(OAc)2/NHS,
PhCO2H in BuOAc at100 8C] were applied to the
oxidation of ethylbenzene. The former conditions were
notsuccessful: conversion remained below 50% even
after 24 hours (Table 8, run 71). When the second set of
conditions was applied, the conversion reached 71%
after 12 hours with high selectivity toward the formation
of acetophenone (96%) along with a small amount (3%)
of phenylethanol (run 72).
[a]
Experimental conditions: 2-octanol (4.5 mmol), NHS or
NHPI (0.45 mmol), Co(acac)3 (0.022 mmol), PhCO2H
(0.22 mmol), butyl acetate (7.5 mL), 1008C, O2 1 at m.
Conversions and yields are determined by GC using 1,2,4-
trichlorobenzene (200 mg) as internal standard.
Using 0.5 mol % Co(OAc)2 and 5 mol % MCBA in
EtOAc as solvent at 758C.
Using 0.5 mol % Co(OAc)2 and 5 mol % m-CPBA(m-
[b]
[c]
[d]
chloroperbenzoic acid) in EtOAc as solvent at 758C.
Mechanistic Considerations: NHS vs. NHPI
2 hours with complete conversion and almost complete
selectivity (runs 62 and 63). The oxidation of 1,2,3,4- In accordance with the mechanism proposed by Ishii
tetrahydro-1-naphthol gave a-tetralone in 86% yield at and co-workers[5] the carbon radical chain promoter
almost complete conversion (run 64). Finally, 1-octanol (NHPI or NHS) is involved in two chain propagation
and benzyl alcohol were oxidised under slightly differ- steps (see Scheme 5).
Table 8. Oxidation of ethylbenzene catalysed by Co/NHS or Co/NHPI.[a]
Run
Catalytic system
Solvent
T [8C]
Time [h]
Conv.[b] [%]
Sel. to acetophenone[b, c] [%]
67
68
69
70
71
72
Co(acac)2/NHPI
Co(acac)2/NHS
Co(OAc)2/NHPI
Co(OAc)2/NHS
Co(acac)3/NHS
Co(OAc)2/NHS, PhCO2H
AcOH
AcOH
CH3CN
CH3CN
PhCF3
100
100
85
85
100
100
6
6
8
8
24
12
82
55
85
27
46
71
90
66
80
42
76
96
BuOAc
[a]
Experimental conditions: 10 mol % NHPI or NHS, 0.5 mol % Co-complex, 5 mol % PhCO2H (when added).
Conversions and yields are determined by GC using 1,2,4-trichlorobenzene as internal standard.
Selectivity based on ethylbenzene reacted.
[b]
[c]
Adv. Synth. Catal. 2004, 346, 286 296
asc.wiley-vch.de
¹ 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
293

Xavier Baucherel etal.
FULL PAPERS
rate of reaction (2) is largely determined by the bond
dissociation enthalpy (BDE) of the substrate and NHPI
À
(1a) or NHS (1b). The BDE of the C H bond of
cycloalkanes (95 kcal/mol) is significantly higher than
À
that of a-C H bond of alcohols or alkyl benzenes (ca.
85 kcal/mol). This difference will be reflected in a larger
activation energy and lower rate for reaction (2) for
cycloalkanes compared to alcohols or alkylbenzenes.
According to a recent study,[19] substitution of alkyl,
aryl or vinyl groups, attached to the nitrogen atom of
hydroxylamines, with acyl groups results in a large
increase in the BDE of the OH bond (from 70 to 88 kcal/
mol). This effect was rationalised on the basis of the
resonance stabilisation energy derived from the extra
mesomeric structures which are not possible for the
dialkylhydroxylamines. Substitution of one of the car-
bonyl groups in NHPI by the more electron-withdraw-
ing sulphonyl moiety is expected to lead to an even
À
higher BDE of the NO H bond. This translates to an N-
oxyl radical which is more difficult to generate but, by
the same token, is more reactive in the abstraction of a
hydrogen atom of the substrate.
Scheme 5.
In the first step the N-oxyl radical, PINO (2a) or SINO
Hence, we propose that in the oxidation of the less
(2b), generated by reaction of CoII/O2 with NHPI (1a) or reactive cycloalkane substrates reaction (2) is rate-
NHS (1b) (see also Scheme 2), abstracts a hydrogen limiting and is faster with SINO (2b) than with PINO
atom from the substrate [Eq. (2)]. The resulting alkyl (2a). In contrast, in the oxidation of the more reactive
radical reacts with dioxygen, in a diffusion-controlled substrates alcohols and alkylbenzenes reaction (2) is
step, to produce the corresponding alkylperoxy radical much more favourable and reaction (4) can become
[Eq. (3)]. The alkylperoxy radical subsequently reacts rate-limiting. Based on the same BDE arguments this
with 1a or 1b to afford the alkyl hydroperoxide, as the reaction is expected to be more favourable for NHPI vs.
primary oxidation product, and regenerate the N-oxyl NHS.
radical 2a or 2b [Eq. (4)].
In NHPI- and NHS-catalysed oxidations the metal co-
The intervention of NHPI or NHS in the radical chain catalyst, usually a cobalt compound, actually functions
process results in an increase in both the rate and as an initiator, i.e., to generate the N-oxyl radical which
selectivity of the autoxidation.[8] In previous studies it then enters the catalytic cycle (Scheme 1). If the goal is
was shown that for the autoxidation of cyclohexylben- to achieve a high yield of hydroperoxide the metal
zene in the presence of NHPI a higher intrinsic catalyst should be replaced with a radical initiator, such
selectivity towards the main product of the reaction was as a peroxide, to avoid metal-catalysed decomposition
observed. This increase ofboth the rate andselectivity of of the hydroperoxide.[8] When the desired product is the
product formation could be explained by the fact that ketone the metal catalyst has a second function: to
NHPI increased the rate of propagation and/or de- catalyse the decomposition of the alkyl hydroperoxide.
creased the rate of termination. Efficient scavenging of
The overall catalytic activity of NHS and NHPI-based
alkylperoxy radicals by 1a or 1b [Eq. (4)] results in a systems is very dependent on the choice of initiator
decrease in the rate of bimolecular termination (metal complex) and solvent. As noted earlier, depend-
[Eq. (5)], thereby affording an increase in the overall ing on the solvent used, and the reaction temperature,
rate. An increase in selectivity in cycloalkane oxidation either Co(acac)2 or Co(acac)3 is the better initiator. We
could be a consequence of the fact that efficient believe that this is caused by a delicate balance of rates
scavenging of alkoxy and alkylperoxy radicals (see of ligand substitution reactions at cobalt (it is well-
Scheme 4) takes place in the presence of high concen- known that ligand substitution at CoIII is a rather slow
trations of NHPI/NHS, thus suppressing cleavage and process at lower temperatures) and one-electron trans-
transannular hydrogen abstraction, respectively.
fer processes leading to the eventual formation of the
In this study we have shown that NHS (1b) is a more actual catalyst, the N-oxyl radical. Thus, the latter can
effective catalyst in the oxidation of cycloalkanes while either be formed via reaction of NHS with an acacO2 ¥
NHPI is more effective in the oxidation of alcohols and radical, formed by dissociation of CoIII(acac)3
alkylbenzenes. We suggest that a plausible explanation (Scheme 2) or by ligand substitution at CoIII(acac)2 of
for this reversal of activity is that it is a consequence of a NHS followed by one-electron transfer from the N-oxy
change in the relative rates of reactions (2) and (4). The ligand to CoIII (see Scheme 3).
294
¹ 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
Adv. Synth. Catal. 2004, 346, 286 296

Aerobic Oxidation of Cycloalkanes, Alcohols and Ethylbenzene
FULL PAPERS
Conclusions
Analysis
The reactions were monitored by gas chromatography using a
CP-WAX 52 CB column (50 m  0.53 mm). A typical temper-
ature program starts at 508C (5 min) with aramp of 88C/min up
to 2208C. The reported yields are thus GC yields and are
determined using 1,2,4-trichlorobenzene as the internal stand-
ard. Molar responses were obtained by calibration of pure
samples against this internal standard. 1,15-Pentadecanoic
acid, 1,12-dodecanedioic acid, 1,10-decanedioic acid and 1,8-
octanedioic acid were analysed by HPLC (Waters Symmetry
C18 reversed phase column, eluent: water/methanol, 50:50,
0.1% TFA) using octanoic acid as an internal standard. The
products reported in the tables, plus the two isomeric
hydroxycyclododecanones and two isomeric cyclododecadi-
enones which have been identified, together account for more
than 90% of the substrate converted.
The NHS/Co combination is an effective catalyst for the
aerobic oxidation of cycloalkanes to the corresponding
cycloalkanones, displaying higher activities than the
previously reported NHPI/Co system. The system was
also applied successfully to the autoxidation of ethyl-
benzene to acetophenone and of a variety of secondary
and primary alcohols to the corresponding ketones and
carboxylic acids, respectively. A mechanistic rationale is
proposed to explain the activities of N-oxyl radicals as
autoxidation catalysts and the different reactivities
observed for NHS and NHPI.
Experimental Section
Acknowledgements
Safety Remarks
This project was financed by Quest International and by the
Dutch Ministry of Economic Affairs under the Innovation
Oriented Projects scheme (IOP Catalysis) and the contributions
are gratefully acknowledged. We thank W. Motherwell for
fruitful discussions. We are also grateful to W. Partenheimer for
his valuable mechanistic discussions.
Aerobic oxidations should always be performed with caution.
In our case with 100% molecular oxygen, we worked above the
explosion limit, and monitored the oxygen uptake by a burette.
Laboratory glassware was used behind safety screens, and the
scale was limited to 10 mL solutions.
References and Notes
Chemicals
The chemicals were obtained from commercial sources and
used as received. N-Hydroxysaccharin (NHS) was synthesised
according to a seven-step synthesis published by Nagasawa
etal. [5] in 1995 with an overall yield of 17%. 1H NMR (DMSO):
d ˆ 7.9 8.1 (m, 4H, Harom), 11.2 (b, 1H, NOH); 13C NMR
(DMSO): d ˆ 121.5, 125.1, 125.8, 134.7, 135.2, 135.7 (6Carom),
[1] a) R. A. Sheldon, J. K. Kochi, Metal-Catalyzed Oxida-
tions of Organic Compounds, Academic Press, New
York, 1981; b) C. L. Hill, Activation and Functionaliza-
tion of Alkanes, Academic Press, New York, 1989.
[2] a) Y. Ishii, T. Iwahama, S. Sakaguchi, K. Nakayama, M.
Takeno, Y. Nishiyama, J. Org. Chem. 1996, 61, 4520
4526; b) Y. Ishii, J. Mol. Catal. A: Chem. 1997, 117, 123
137; c) Y. Ishii, S. Sakaguchi, Catalysis Surveys from
Japan 1999, 3, 27 35; d) Y. Ishii, Eur. Patent EP
0 824 962 A1, 1996; e) Y. Ishii, S. Sakaguchi, T. Iwahama,
Adv. Synth. Catal. 2001, 343, 393 427.
ˆ
157.5 (C O).
Typical Catalytic Run
Experiments were carried out in a two-necked 25-cm3 round-
bottom flask containing a magnetic stirrer, equipped with a
condenser and connected to a gas burette filled with molecular
oxygen. In a typical experiment 504 mg (3 mmol) of cyclo-
dodecane, 60 mg (0.3 mmol) of N-hydroxysaccharin, 5.3 mg
(0.015 mmol) of Co(acac)3 and 200 mg of 1,2,4-trichloroben-
zene were weighed in the flask before addition of 7.5 mL acetic
acid or 9 mL of acetic acid. A series of vacuum/molecular
oxygen purges were applied before the reaction was started.
The resultant mixture was stirred at a temperature of 808C for
10 hours.
For 2-octanol oxidation in a typical experiment 585 mg 2-
octanol (4.5 mmol), 90 mg (0.45 mmol) of N-hydroxysacchar-
in, 8 mg (0.022 mmol) Co(acac)3, 27.5 mg (0.022 mmol) of
benzoic acid and 200 mg of 1,2,4-trichlorobenzene were
dissolved in 7.5 mL solvent, followed by oxygen purges. The
resultant mixture was stirred at a temperature of 1008C for 5 h.
For other substrates conditions are listed with the tables.
[3] B. B. Wentzel, M. P. J. Donners, P. L. Alsters, M. C.
Feiters, R. J. M. Nolte, Tetrahedron 2000, 56, 7797 7803.
[4] K. Gorgy, J-C. Lepretre, E. Saint-Aman, C. Einhorn, J.
Einhorn, C. Marcadal, J.-L. Pierre, Electrochim. Acta
1998, 44, 385 393.
[5] H. T. Nagasawa, S. P. Kawle, J. A. Elberling, E. G. De-
Master, J. M. Fukuto, J. Med. Chem. 1995, 38, 1865
1871.
[6] N. Hirai, N. Sawatari, N. Nakamura, S. Sakaguchi, Y.
Ishii, J. Org. Chem. 2003, 68, 6587 6590.
[7] a) X. Baucherel, I. W. C. E. Arends, S. Ellwood, R. A.
Sheldon, Org. Proc. Res. Dev. 2003, 7, 426 428; b) X.
Baucherel, R. A. Sheldon, WO 02100810, 2002.
[8] I. W. C. E. Arends, M. Sasidharan, A. Kuehnle, M. Duda,
C. Jost, R. A. Sheldon, Tetrahedron 2002, 58, 9055 9061.
[9] A. Onopchenko, J. G. D. Schulz, J. Org. Chem. 1973, 38,
3729.
Adv. Synth. Catal. 2004, 346, 286 296
asc.wiley-vch.de
¹ 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
295

Xavier Baucherel etal.
FULL PAPERS
[10] W. Partenheimer, J. Mol. Catal. A: Chemical 2001, 174,
29 33.
[11] W. Partenheimer, J. Mol. Catal. A: Chemical 2003, 206,
105 119.
[12] J. Muzart, Chem. Rev. 1992, 92, 113 140.
[13] a) T. Welton, Chem. Rev. 1999, 99, 2071 2084; b) R. A.
Sheldon, Chem. Commun. 2001, 2399 2407.
Commun. 1999, 1591; f) K. Yamaguchi, N. Mizuno,
Angew. Chem. Int. Ed. 2002, 41, 4538; g) K. Yamaguchi,
N. Mizuno, New. J. Chem. 2002, 26, 972; for methods
based on Pd and O2, see: h) T. Nishimura, K. Ohe. S. J.
Uemura, J. Am. Chem. Soc. 1999, 121, 2645; i) G.-J. ten
Brink, I. W. C. E. Arends, R. A. Sheldon, Adv. Synth.
Catal. 2002, 344, 355; for methods based on Cu, see:
¬
[14] K. Matsunaka, T. Iwahama, S. Sakaguchi, Y. Ishii,
Tetrahedron Lett. 1999, 40, 2165 2168.
j) I. E. Marko, P. R. Giles, M. Tsukazaki, I. Chelle-
Regnaut, A. Gautier, S. M. Brown, C. J. Urch, J. Org.
Chem., 1999, 64, 2433; k) B. Betzemeier, M. Cavazzine, S.
Quici, P. Knochel, P. Tetrahedron Lett. 2000, 41, 4343;
l) P. Gamez, I. W. C. E. Arends, J. Reedijk, R. A. Shel-
don, Chem. Commun. 2003, 2414.
[15] a) S. Wilsey, P. Dowd, K. N. Houk, J. Org. Chem. 1999,
64, 8801; b) A. L. J. Beckwith, B. P. Hay, J. Am. Chem.
Soc. 1989, 111, 2674(( ‡ Kochi ??)).
[16] a) For methods based on Ru and O2,see e.g.: a) J. E.
B‰ckvall, R. L. Chowdhury, U. Karlsson, J. Chem. Soc.
Chem. Commun. 1991, 473; b) I. E. Marko, P. R. Giles,
M. Tsukazaki, I. Chelle-Regnaut, C. J. Urch, S. M.
Brown, J. Am. Chem. Soc. 1997, 119, 12661; c) B. Hinzen,
R. Lenz, S. V. Ley, Synthesis 1998, 977; d) T. Matsushita,
K. Ebitani, K. Kaneda, Chem. Commun. 1999, 265; e) A.
Dijksman, I. W. C. E. Arends, R. A. Sheldon, Chem.
[17] T. Iwahama, S. Sakaguchi, Y. Nishiyama, Y. Ishii,
Tetrahedron Lett. 1995, 36, 6923 6926.
[18] T. Iwahama, Y. Yoshino, T. Keitoku, S. Sakaguchi, Y.
Ishii, J. Org. Chem. 2000, 65, 6502 6507.
[19] R. Amorati, M. Lucarini, V. Mugnaini, G. F. Pedulli, F.
Minisci, F. Recupero, F. Fontana, P. Astolfi, L. Greci, J.
Org. Chem. 2003, 68, 1747 1754.
296
¹ 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
Adv. Synth. Catal. 2004, 346, 286 296