Ligand and solvent effects on cobalt(I)-catalysed reactions: Alkyne dimerisation versus [2+2+2]-cyclotrimerisation versus Diels-Alder reaction versus [4+2+2]-cycloaddition

Article abstract of DOI:10.1016/j.jorganchem.2005.03.067

The cobalt catalysed conversion of phenyl acetylene led to linear enyne dimerisation products when CoBr₂(dppe) was activated with magnesium in the absence of a Lewis acid. In contrast, in the presence of a Lewis acid the cyclotrimerisation process is favoured. Among several ligand systems and solvents tested the best results were obtained using a catalyst system consisting of a diimine cobalt bromide complex, zinc and zinc iodide in acetonitrile. With 2-5 mol% of the cobalt catalyst at ambient temperatures 1,2,4-triphenylbenzene could be obtained in 99% yield and in excellent regioselectivity (95:5) in 10 min reaction time. Competition experiments of phenylacetylene and isoprene were performed. A preference for the cyclotrimerisation reaction was found for the diimine cobalt complex in acetonitrile, while the Diels-Alder reaction is favoured with the cobalt(dppe) complex in dichloromethane. Also a regioselectively substituted cyclooctatriene product was formed in a [4+2+2]-cycloaddition process and isolated which allows assumptions on the reaction mechanism.

Full text of DOI:10.1016/j.jorganchem.2005.03.067

Journal of Organometallic Chemistry 690 (2005) 5170–5181
www.elsevier.com/locate/jorganchem
Ligand and solvent effects on cobalt(I)-catalysed reactions: Alkyne
dimerisation versus [2+2+2]-cyclotrimerisation versus
Diels–Alder reaction versus [4+2+2]-cycloaddition
*
Gerhard Hilt , Wilfried Hess, Thomas Vogler, Christoph Hengst
Fachbereich Chemie, Philipps-Universita¨t Marburg, Hans-Meerwein-Straße, 35043 Marburg, Germany
Received 29 January 2005; received in revised form 29 March 2005; accepted 29 March 2005
Available online 15 June 2005
Abstract
The cobalt catalysed conversion of phenyl acetylene led to linear enyne dimerisation products when CoBr2(dppe) was activated
with magnesium in the absence of a Lewis acid. In contrast, in the presence of a Lewis acid the cyclotrimerisation process is
favoured. Among several ligand systems and solvents tested the best results were obtained using a catalyst system consisting of a
diimine cobalt bromide complex, zinc and zinc iodide in acetonitrile. With 2–5 mol% of the cobalt catalyst at ambient temperatures
1,2,4-triphenylbenzene could be obtained in 99% yield and in excellent regioselectivity (95:5) in 10 min reaction time. Competition
experiments of phenylacetylene and isoprene were performed. A preference for the cyclotrimerisation reaction was found for the
diimine cobalt complex in acetonitrile, while the Diels–Alder reaction is favoured with the cobalt(dppe) complex in dichlorometh-
ane. Also a regioselectively substituted cyclooctatriene product was formed in a [4+2+2]-cycloaddition process and isolated which
allows assumptions on the reaction mechanism.
Ó 2005 Elsevier B.V. All rights reserved.
Keywords: Cobalt; Cyclotrimerisation; Diels–Alder; Dimerisation; Ligand; Solvent effect
1. Introduction
such transformations, cycloaddition processes, such as
the transition metal catalysed Diels–Alder reaction or
the [2+2+2]-cyclotrimerisation of alkynes, are already
well documented [1]. Over the last couple of years, we
have investigated cobalt(I)-catalysed Diels–Alder reac-
tions between non-activated terminal as well as internal
alkynes and acyclic 1,3-dienes. We found that a catalyst
system consisting of Co(dppe)Br2/ZnI2/Bu4NBH4 or
zinc as reducing agent proved to be highly effective in
the generation of dihydroaromatic compounds under
mild reaction conditions in good to excellent yields [2].
In the earliest report of such transformations, we de-
scribed the [2+2+2]-cyclotrimerisation of an acetylene
carboxylic ester to the corresponding trisubstituted ben-
zene derivative, which was an undesired side product at
that point [3]. Besides the acetylene carboxylic ester, the
cobalt catalyst system exhibits only a very low reactivity
Transition metal complexes with vacant coordination
sites are able to form complexes with unsaturated organ-
ic molecules such as alkenes or alkynes. These organic
molecules can be activated within the ligand sphere of
the transition metal to undergo carbon–carbon bond
formation processes. When the liberation of the organic
product regenerates the active species a catalytic process
becomes possible. Accordingly, organic transformations
with carbon carbon bond formations can be realised
which would not be possible under regular thermal con-
ditions in the absence of the transition metal. Among
*
Corresponding author. Tel.: +49 6421 282 5601; fax: +49 6421 282
5601.
E-mail address: hilt@chemie.uni-marburg.de (G. Hilt).
0022-328X/$ - see front matter Ó 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.jorganchem.2005.03.067

G. Hilt et al. / Journal of Organometallic Chemistry 690 (2005) 5170–5181
5171
for the cyclotrimerisation of other terminal alkynes.
Ph
CoBr2(dppe), Mg, Zn
CH3CN
Therefore, we started an investigation to determine the
factors that would enhance the activity of the very inex-
pensive and easy to handle cobalt catalyst system to-
wards carbon carbon bond formations.
Ph
1
Ph
+
linear and cyclic trimeres
Scheme 2.
2. Results and discussion
fied as (E)-1,4-diphenyl-but-1-en-3-yne (1) (Scheme 2)
[6]. A thermal pre-treatment of the catalyst components
in the absence of phenylacetylene was found to have a
major effect on the reaction time. Prior to substrate
addition the reaction time could be reduced from 20 h
to only 10 min for complete conversion of the substrate.
An only slight effect on the product ratio was found
when the ratio of the reducing agents magnesium and
zinc was altered. Nevertheless, in the absence of zinc,
using magnesium as the sole reducing agent (E)-1,4-di-
phenyl-but-1-en-3-yne (1) was obtained as major prod-
uct after a few minutes reaction time when the catalyst
was briefly heated to reflux prior to the addition of the
starting material (compare Table 1, entries 2 and 7). A
combination of magnesium and zinc as reducing agents
yields trimers as main products. However, the reaction
proceeds without magnesium and with zinc as reducing
agent very slowly (Table 1, entry 8).
For the cobalt-catalysed Diels–Alder reaction of
alkynes with acyclic 1,3-dienes we proposed a cationic
cobalt(I)-species as the active catalyst. This species is
generated upon reduction and halide abstraction pro-
cesses starting from CoBr2(dppe) as a catalyst precursor
(Scheme 1) [4].
We envisaged that with a stronger reducing agent
such as magnesium a second reduction process could
take place to obtain a neutral cobalt(0)(dppe)-species
with different reactivity in the reaction with unsaturated
starting materials such as phenylacetylene.
2.1. Dimerisation process
The first experiments with the CoBr2(dppe) catalyst
proved to be unsuccessful in dichloromethane as sol-
vent, while in an acetonitrile/dichloromethane solvent
mixture small amounts of the dimerisation product
could be isolated [5]. Better results were obtained when
acetonitrile alone was used as solvent. In these reactions
a dimer accompanied by a mixture of linear and cyclic
trimers (confirmed by GC–MS) of phenylacetylene was
isolated. The dimer 1 could be easily separated from
the trimers by column chromatography and was identi-
The formation of linear and cyclic trimers could not
be reduced significantly, so that only moderate yields
(47%) of the dimerisation product 1 could be obtained
as the best result (Table 1, entry 7). Further investiga-
tions towards a synthetically useful chemo- and regio-
chemically pure dimerisation process were not
conducted.
ZnI2
ZnI2
Zn
CoIIBr2(dppe)
CoIBr(dppe)
CoI(dppe)+
-
-
ZnI2Br
ZnI2Br
Scheme 1.
Table 1
Cobalt(I)-catalyzed dimerisation/trimerisation of phenylacetylene in the absence of a Lewis acid
No.
CoBr2(dppe) (mol%)
Mg (mol%)
Zn (mol%)
Solvent
Time
Dimer (1) (%)
Trimersc (%)
1a
2a
3
20
2
100
10
10
40
40
10
10
–
–
CH3CN/CH2Cl2
CH3CN
CH3CN
14 h
20 h
9
32
36
26
34
34
47
7
n.i.
49
53
22
49
45
41
8
b
–
2
10
40
10
40
–
10 min
10 min
10 min
10 min
10 min
10 min
4
2
CH3CN
CH3CN
5
2
6
2
CH3CN
CH3CN
CH3CN
7
2
8
2
40
Reaction conditions: phenylacetylene (2.0 mmol), CH3CN (1.0 mL), room temperature.
a
Without previous heating of the catalyst mixture.
CH2Cl2 (1.0 mL), CH3CN (0.1 mL).
The 1H NMR shows additional olefinic protons. n.i. = not isolated.
b
c

5172
G. Hilt et al. / Journal of Organometallic Chemistry 690 (2005) 5170–5181
Table 2
Cobalt(I)-catalyzed [2+2+2]-cyclotrimerisation of phenylacetylene
using different ligands
2.2. Alkyne trimerisation
In the cobalt-catalysed Diels–Alder reaction of alky-
nes with 1,3-dienes the use of a Lewis acid such as zinc
iodide proved to be essential for the catalytic activity.
When zinc iodide was added to the catalyst mixture con-
taining magnesium and zinc as reducing agent, a good
conversion of the phenylacetylene into the cyclic trimers
2 and 3 was observed (Scheme 3) [7]. The first experi-
ments were conducted with CoBr2(dppe) as catalyst pre-
cursor in dichloromethane and gave in 70% yield a
4.8:1.0 mixture of 2:3 in a relatively short reaction time
at ambient temperatures (Table 2, entry 1).
No.
Catalyst precursor
Time (h)
Yield (%)
Ratio 2:3
1
2
3
4
5
6
CoBr2(dppe)
3
70
57
4.8:1.0
5.0:1.0
1.0:1.5
–
CoBr2(PPh3)2
CoBr2(dppm)a
CoBr2(dppf)a
CoBr2(Pc-Hex3)2
CoBr2[P(o-tol)3]2
20b
16
8c
20b
16
25c,d
88
a
a
8.3:1.0
–
20
Traces
7
4
96e
1.0:2.9e
Ph
S
Br2Co
S
Encouraged by this result obtained with this catalyst
system, we subsequently investigated a small number of
different ligands on the cyclotrimerisation reaction to
identify general trends of the reaction. In all cases, the
use of the Lewis acid zinc iodide favoured the formation
of the cyclotrimerisation products dramatically, so that
only traces of previously observed linear dimerisation
or trimerisation products could be found. The use of
electron rich ligands such as tricyclohexylphosphine
(Table 2, entry 5) turned out to have advantages under
the applied reaction conditions, whereas the use of other
bidentate or monodentate phosphine ligands (Table 2,
entries 1–4 and 6) were contra productive [8]. Although
phosphine ligands are among the most prominent and
versatile ligands used in transition metal catalysed reac-
tions, other donor based and readily available ligands
were tested. Very interesting results were obtained when
ethylene diphenylthioether or cyclohexyl diimine com-
plexes (Table 2, entries 7 and 8) were tested [9]. The reac-
tion times shortened dramatically and after complete
conversion the isolated yields were excellent (>94%).
Particularly, interesting was the observation that the iso-
meric ratio of the unsymmetrical product (2) and the
symmetrical product (3) is noticeably modified by these
different types of ligands. While the symmetrical arene 3
is obtained as main product by the disulfide complex
(Table 2, entry 7), the diimine and most phosphine com-
plexes predominantly yield the 1,2,4-substituted product
(2). Again, a thermal pre-treatment of the system com-
ponents in solution prior to substrate addition decreases
the reaction time drastically.
Ph
8
5
94
5.8:1.0
c-Hex
N
Br2Co
N
c-Hex
Reaction conditions: phenylacetylene (2.0 mmol), catalyst (5 mol%),
magnesium (25 mol%), zinc (10 mol%), zinc iodide (10 mol%), CH2Cl2
(1.0 mL), room temperature.
a
The catalyst was generated in situ.
Incomplete conversion.
b
c
The enyne 1 was detected in trace amounts.
The 1H NMR shows additional olefinic protons, therefore the ratio
of regioisomeres can not be evaluated.
d
e
The product contained traces of the ligand, so that the yield was
corrected by integration of the 1H NMR spectrum.
1H NMR spectroscopy. Although, a large number of
phosphine ligands are known, slight modifications of
the ligand often result in tedious syntheses accompanied
with problems related to the air sensitivity of phospho-
rus(III) compounds. In contrast, the chemistry of disul-
fide or diimine ligands is comparably simple and
modifications can be easily prepared on a large scale
using inexpensive commercially available starting
materials.
2.3. Solvent effect on the alkyne trimerisation
Phenylacetylene proved to be a good choice as sub-
strate because the ratio of the regioisomereic cyclotri-
merisation products 1,2,4-triphenylbenzene (2) and
1,3,5-triphenylbenzene (3) can be easily determined via
Due to an easy synthetic approach to diimines we
chose these ligands for our next set of experiments. In
the course of our investigations magnesium was found
Ph
Ph
CoBr2(ligand)
Mg, Zn, ZnI2
Ph
+
Ph
CH2Cl2
Ph
Ph
Ph
2
3
Scheme 3.

G. Hilt et al. / Journal of Organometallic Chemistry 690 (2005) 5170–5181
5173
Ph
Ph
CoBr2(c-Hex-diimine)
Ph
Zn, ZnI2
+
Ph
solvent
Ph
Ph
Ph
2
3
Scheme 4.
to be not necessary for the generation of the active
species to undergo the cyclotrimerisation process.
Additionally, the presence of magnesium as reducing
agent does not influence the isomeric ratio significantly.
Having found a strong ligand dependency on the iso-
meric ratio we assumed that the ability of solvents to
coordinate to the low valent cobalt catalyst might influ-
ence the product ratio as well. Therefore, the reaction of
phenylacetylene with the bis-cyclohexylethylene diimine
cobalt bromide complex (Scheme 4) was performed in
various solvents to test this hypothesis.
Table 3
Solvent dependency of the cobalt(I)-catalyzed [2+2+2]-cyclotrimerisa-
tion of phenylacetylene
No.
Solvent
Time
Yield (%)
Ratio 2:3
1
2
Acetone
DMSO
DMF
5 h
39
3
3.0:1.0
30.5:1.0
–
15 h
20 h
20 ha
15 h
4 h
3
Traces
7
4
Ethanol
EtOAc
2.7:1.0
4.3:1.0
5.8:1.0
5.3:1.0
3.6:1.0
–
5
71
94
95
26b
0
32c
96
81
32
81
60c
13c
0
6
CH2Cl2
ClCH2CH2Cl
Et2O
7
15 h
20 ha
15 h
20 h
12 hd
15 h
15 h
15 h
20 h
20 h
20 ha
10 min
30 min
15 h
10 min
8
9
Bu2O
Dioxane
THF
Indeed, solvent effects on the catalyst system are dras-
tic. An impact on yield, reaction time and isomeric ratio
is found. Highly polar or protic solvents such as DMF,
DMSO and ethanol gave very low conversions of the
starting material, while relatively unpolar solvents such
as dibutyl ether or toluene gave no conversion of the
starting material based on the low solubility of the cat-
alyst system in these solvents. Among several oxygen
containing solvents the most interesting behaviour was
observed in THF where a relatively good yield was com-
bined with a shifted regioselectivity slightly favouring
the symmetrical isomer. The most intriguing behaviour
was, however, observed in acetonitrile as solvent where
the addition of the starting material led to a very fast
conversion. Surprisingly, in acetonitrile a very short
reaction time was accompanied with an excellent regi-
oselectivity in favour of the unsymmetrical product 2.
Further increase of the sterical hindrance of the nitrile
(Table 3, entries 18–21) led to somewhat lower reactions
rates and a lower selectivity compared to the reaction in
acetonitrile. Nevertheless, from a subjective point of
view, the reaction in benzonitrile was probably the fast-
est observed and the most exothermic. However, the reg-
ioselectivity was not as distinct as in acetonitrile. Also,
the separation of the desired products from benzonitrile
was somewhat more difficult. The dependency of the
regiochemistry of the products from the solvent used
let us speculate that a solvent molecule acts as a ligand
on the cobalt centre during the bond formation process.
To underline this solvent dependency and a coordina-
tion of acetonitrile, the trimerisation reaction was per-
formed in a solvent system consisting of THF and
acetonitrile with various mixing ratios (Fig. 1). To 5
mol% of the cobalt catalyst, an increasing number of
acetonitrile were added. The result is a very steep decay
of the ratio between 2 and 3 starting from pure THF
10
11
12
13
14
15
16
17
18
19
20
21
2.5:1.0
1.0:1.3
1.1:1.0
4.5:1.0
2.4:1.0
3.8:1.0
19.9:1.0
–
2-Methyl-THF
Furan
THP
MeOCH2CH2OMe
Pyridine
Toluene
CH3CN
EtCN
99
95
53
95
19.2:1.0
10.8:1.0
7.1:1.0
11.8:1.0
t-BuCN
PhCN
Reaction conditions: phenylacetylene (2.0 mmol), cobalt complex (5
mol%), zinc (10 mol%), zinc iodide (10 mol%), solvent (1.0 mL), room
temperature.
a
Incomplete conversion.
The 1H NMR shows additional olefinic protons.
Also 1,4-diphenylbuta-1,3-diyne was isolated in trace amounts
b
c
(<2%).
d
Without previous heating of the catalyst mixture.
60
50
40
30
20
10
0
0
20
40
60
80
100
Pure THF
Pure Acetonitrile
[%] Acetonitrile in THF
Fig. 1. Ratio of the symmetrical trimer subject to the composite of the
used solvent mixture.

5174
G. Hilt et al. / Journal of Organometallic Chemistry 690 (2005) 5170–5181
(2:3 = 1.0:1.3) when small amounts of acetonitrile are
present. A lesser exponential decay is observed when
larger amounts of acetonitrile are added until the reac-
tion in pure acetonitrile gave 2:3 in a 19.2:1.0 ratio.
Accordingly, even small amounts of acetonitrile have
a profound effect on the regioselectivity and an incorpo-
ration of acetonitrile as a ligand can be presumed.
Accompanied with this coordination of the nitrile sol-
vent a reorganisation of the ligands at the cobalt centre
can be envisaged. However, predictions upon the geom-
etry and the orientations of the ligands within the tran-
sition state are highly speculative at this point.
Three solvents were chosen to be most interesting for
the next reactions. The reactivity in acetonitrile is out-
standing and an excellent ratio in favour of the unsym-
metrical product 2 is obtained (Table 3, entry 18). In
tetrahydrofurane this ratio is reversed with the diimine
cobalt complex (Table 3, entry 11), the same result is ob-
tained with the disulfide complex in dichloromethane
(Table 2, entry 7). All three solvents provide good to
excellent yields and no traces of olefinic protons, indicat-
ing open chain trimers or oligomers.
was found for the disulfide complex (Table 4, entries
7–9). While the data for the reaction in acetonitrile are
excellent (96%, 2:3 = 16.0:1.0), in the weakly coordinat-
ing solvent dichloromethane, the regioselectivity was in-
versed to give the symmetrical product as main isomer
(2:3 = 1.0:2.9). The reactivity of the disulfide complex
in CH2Cl2 was higher compared to the reaction con-
ducted in THF. This behaviour is different to the diimine
complexes which exhibit a higher reactivity in THF. In
these reactions (Table 4, entries 10–33) we found a pro-
found effect of the steric properties on the reactivity and
the selectivity of the complexes. A steric increase of the
nitrogen-substituents leads to an improvement of the
isomeric ratio in acetonitrile in favour of the unsymmet-
rical product.
While the n-butyl diimine complex produces a ratio
of 2:3 = 7.0:1.0, the ratio is gradually increased with
the ligands iso-propyl, cyclo-pentyl, cyclo-hexyl, Mesityl
and tert-butyl. For the latter t-butyl diimine complex an
excellent ratio of 29.0:1.0 is obtained. In principle a de-
creased reactivity of the complexes is observed when the
steric bulk of the ligand is increased. Although all com-
plexes are very reactive, the best combination of yield,
reactivity and regioselectivity is obtained with the dic-
yclohexyl diimine complex (Table 4, entry 19). In con-
trast to all other diimine ligands tested this particular
complex shows a drastic change of the regioselectivity
in different solvents. In THF the regioselectivity is chan-
ged favouring the symmetrical product 3 (Table 4, entry
21). With the corresponding complex having two addi-
tional methyl groups in the backbone of the diimine
motif (Table 4, entries 31–33) such a dramatic change
of the reaction characteristics was not found. The
hypothesis that the conformational flexibility of the
cyclohexyl ring could have an influence on the regiose-
lective formation of the intermediates during the cyclo-
trimerisation processes can not be ruled out with the
present set of experiments. Among the other catalyst
systems tested we would like to emphasise that also pyr-
idine type ligands (Table 4, entries 37–45) and a tertiary
amine ligand (Table 4, entry 49) was compatible with
the reaction and gave also promising results. Surpris-
ingly, the reaction can be performed even with CoBr2
in acetonitrile without any additional ligands. The reg-
ioisomeric ratio in this case is 18.6:1.0 clearly indicating
an important part of this solvent in the mechanism as
2.4. Ligand effect on the alkyne trimerisation
For a further increase of the reactivity and regioselec-
tivity as well as for a more comprehensive understand-
ing of the effects induced by the ligand system, the
diimine motif was further modified. Subsequently, two
cobalt phosphine complexes and several cobalt imine
type complexes were tested in acetonitrile, tetrahydrofu-
rane and dichloromethane (see Scheme 5).
An overall trend can be identified from the data sum-
marised in the table: the reactivity of most complexes
follows mainly the order CH3CN ꢀ THF > CH2Cl2.
Unfortunately, other parameters such as yield and reac-
tivity differ even within a class of complexes. Small
changes in the ligand sphere altered the results, so that
a general discussion is impropriate. However, the posi-
tive effects of the more electron rich phosphine ligands
were also observed in the other tested solvents. Particu-
larly in Table 4, entry 4 the results for the CoBr2(Pc-
Hex3)2 complex are excellent regarding the reactivity
of the catalyst system, the isolated yield as well as the
regioselective formation of the unsymmetrical product.
The greatest difference in terms of the regioselectivity
Ph
Ph
CoBr2(ligand)
Zn, ZnI2
Ph
+
Ph
CH3CN, THF or CH2Cl2
Ph
Ph
Ph
2
3
Scheme 5.

G. Hilt et al. / Journal of Organometallic Chemistry 690 (2005) 5170–5181
5175
Table 4
Cobalt(I)-catalysed [2+2+2]-cyclotrimerisation of phenylacetylene using different ligands in three different solvents
No.
Catalyst precursor
Solvent
Time
Yield (%)
Ratio 2:3
1
2
3
CH3CN
CH2Cl2
THF
1 h
57
66
51b,c
14.3:1.0
6.0:1.0
3.7:1.0
PPh3
Br2Co
16 h
16 h
PPh3
a
4
5
6
CH3CN
CH2Cl2
THF
30 min
16 h
16 h
91
88
40
21.8:1.0
8.3:1.0
3.5:1.0
Pc-Hex3
Br2Co
Pc-Hex3
7
8
9
CH3CN
CH2Cl2
THF
30 min
5 h
16 h
96
16.0:1.0
1.0:2.9
1.8:1.0
Ph
96d
42
S
Br2Co
S
Ph
10
11
12
CH3CN
CH2Cl2
THF
15 min
16 h
16 h
75
88
58e
7.0:1.0
4.5:1.0
2.4:1.0
n
Bu
N
Br2Co
Bu
N
n
13
14
15
CH3CN
CH2Cl2
THF
15 min
16 h
4 h
90
87
69
11.9:1.0
4.6:1.0
7.4:1.0
i
Pr
N
N
Br2Co
Pr
i
16
17
18
CH3CN
1 h
99
94
75
16.4:1.0
2.7:1.0
3.4:1.0
c-Pen
g
CH2Cl2
THF
15 h
3 h
N
Br2Co
N
c-Pen
19
20
21
CH3CN
CH2Cl2
THF
10 min
4 h
12 h
99
94
96
19.2:1.0
5.8:1.0
1.0:1.3
c-Hex
N
Br2Co
c-Hex
N
22
23
24
CH3CN
CH2Cl2
THF
15 min
16 h
1.5 h
93
81
92
13.6:1.0
4.9:1.0
16.7:1.0
c-dodecyl
Br2Co
N
N
c-dodecyl
25
26
27
CH3CN
CH2Cl2
THF
30 min
16 h
16 h
90
81
85c
29.0:1.0
6.0:1.0
13.7:1.0
tBu
N
Br2Co
N
tBu
(continued on next page)

5176
G. Hilt et al. / Journal of Organometallic Chemistry 690 (2005) 5170–5181
Table 4 (continued)
No.
Catalyst precursor
Solvent
Time
Yield (%)
Ratio 2:3
28
29
30
CH3CN
CH2Cl2
THF
30 min
16 h
16 h
85
31
22
21.3:1.0
2.6:1.0
6.3:1.0
Mes
N
Br2Co
N
Mes
31
32
33
CH3CN
CH2Cl2
THF
30 min
16 h
16 h
94
75
66
21.3:1.0
5.3:1.0
5.6:1.0
c-Hex
N
Br2Co
N
c-Hex
34
35
36
CH3CN
CH2Cl2
THF
1 h
99
0
19.0:1.0
–
3.5:1.0
15 h
3 h
36
O
N
Br2Co
N
O
37
38
39
CH3CN
CH2Cl2
THF
1 h
99
92
86
31.0:1.0
17.6:1.0
9.4:1.0
15 h
3 h
N
Br2Co
N
40
41
42
CH3CN
CH2Cl2
THF
1 h
99
0
22.3:1.0
–
12.3:1.0
15 h
3 h
N
99
Br2Co
N
43
44
45
CH3CN
CH2Cl2
THF
1 h
99
37
21.3:1.0
8.6:1.0
–
15 h
3 h
N
Traces
Br2Co
N
46
47
48
CH3CN
CH2Cl2
THF
1.5 h
10 h
10 h
56
14.9:1.0
BnO
n.i.
n.i.
–
–
N
Br2Co
N
BnO
49
50
51
CH3CN
CH2Cl2
THF
1 h
99
0
Traces
27.0:1.0
N
Br2Co
N
15 h
3 h
–
–

G. Hilt et al. / Journal of Organometallic Chemistry 690 (2005) 5170–5181
5177
Table 4 (continued)
No.
Catalyst precursor
CoBr2(salen)
Solvent
Time
Yield (%)
Ratio 2:3
52
53
54
CH3CN
CH2Cl2
THF
1 h
Traces
Traces
Traces
–
–
–
15 h
3 h
55
56
57
CoBr2
CH3CN
CH2Cl2
THF
30 min
16 h
16 h
99
0f
4
18.6:1.0
–
–
Reaction conditions: phenylacetylene (2.0 mmol), cobalt complex (5 mol%), zinc (10 mol%), zinc iodide (10 mol%), solvent (1.0 mL), room
temperature.
a
The catalyst was generated in situ.
In addition the enyne 1 was isolated in 27% yield.
b
c
The 1H NMR shows additional olefinic protons.
d
The product contained traces of the ligand, so that the yield was corrected by integration of the 1H NMR spectrum.
Traces of the enyne 1 were detectable by GC–MS.
e
f
When CoBr2(CH3CN)x was used after 16 h 75%, of the arene were isolated (2:3 = 5.3:1.0).
Without previous heating of the catalyst mixture. n.i. = not isolated, very low conversion.
g
outlined above in the investigation with different nitrile
type solvents. In dichloromethane the poorly soluble
anhydrous CoBr2 alone gave no conversion. When
anhydrous cobalt bromide was stirred in acetonitrile,
the solvent removed an acetonitrile adduct
[CoBr2(CH3CN)x] was obtained [10]. This complex
was used as a pre-catalyst in dichloromethane to yield
75% of the trimerisation products (2:3 = 5.3:1.0) which
were isolated after 16 h reaction time. In THF both,
the anhydrous cobalt bromide as well as the
[CoBr2(CH3CN)x] gave very low conversions (>5%).
These effects also suggest that the active catalyst species
in acetonitrile could at least be coordinated partially by
a nitrile ligand.
In summary, acetonitrile is the solvent of choice for
the formation of the 1,2,4-substituted product. The reac-
tivity in this solvent is extraordinary and the products
are isolated in high yields. Additionally, several classes
of donor ligands (electron rich phosphines, imines,
amines, pyridines and sulfides) can be used to obtain
interesting results. Furthermore, a large number of
combinations of these substructures could be envisaged
and most likely easily be synthesised. However, the not
uniform tendencies regarding isomeric ratios and yields
in dichloromethane and tetrahydrofurane where the
reaction proceeds generally slower can not easily be
understood.
selective for the cyclotrimerisation of terminal and inter-
nal alkynes. From a synthetical point of view, it was also
important to show changes in chemoselectivity. There-
fore, competition experiments with phenylacetylene
and isoprene were conducted. The three reactions in
Scheme 6 gave an interesting result, which will be of
some relevance when the reaction mechanisms are taken
into account [11].
The important observation for the synthetically
interested chemist is that the chemoselectivity was in-
deed strongly altered by changing the ligand and the
solvent system. Although, the cyclotrimerisation reac-
tion is performed in acetonitrile no corresponding pyr-
idine derivative could be observed (GC–MS) [12]. The
third reaction in Scheme 6 is interesting from a mech-
anistic point of view because of the formation of the
cyclooctatriene derivative 5 by a formal [4+2+2]-cyclo-
addition process in 31% yield [13,14]. Either the forma-
tion of 5 represents an alternative reaction pathway
during the cyclotrimerisation reaction, or 5 is formed
as a side product in the Diels–Alder reaction. In the
first case a reaction of isoprene with a 3,4-diphenyl
substituted cobaltacyclopentadiene derivative must be
proposed. Such an intermediate is not believed to be
of any relevance in the cyclotrimerisation processes
investigated thus far [9]. The second possible route
for the formation of 5 is the insertion of phenylacety-
lene into a cobaltacycle intermediate during the forma-
tion of the Diels–Alder product (Scheme 7) [15]. From
the proposed cobaltacycle intermediate 6 a reductive
elimination leads to the Diels–Alder product 4 which
was also observed in 64% yield with the para to meta
regioselectivity of 4.8:1.0 [3]. Based on the observation
that the cyclooctatriene derivative shows a 1,2-diphe-
nyl substitution pattern, one has to conclude, that
the insertion process goes via intermediate 6 with a
2.5. Competition experiments and mechanistic aspects
The investigations so far led from a catalyst system
consisting of CoBr2(dppe), Zn and ZnI2 in dichloro-
methane to a system consisting of CoBr2(diimine), Zn,
ZnI2 in acetonitrile. While the first catalyst system is well
suited to initiate Diels–Alder reactions of non-activated
starting materials, the second system seems to be very

5178
G. Hilt et al. / Journal of Organometallic Chemistry 690 (2005) 5170–5181
Ph
CoBr2(dppe)
Zn, ZnI2
Ph
+
+
CH2Cl2, 20˚C
95%
4
Ph
CoBr2(c-Hex-diimine)
Zn, ZnI2
Ph
+
+
2 / 3
CH3CN, 20˚C
85%
4%
4
Ph
Ph
Ph
CoBr2(c-Hex-diimine)
Zn, ZnI2
Ph
+
2 / 3
+
CH2Cl2, 20˚C
<1%
31%
64%
4
5
Scheme 6.
Co
Lx
Ph
Ph
Ph
Co
path A
Lx
7
Ph
Co
Lx
Ph
Ph
4
6
Ph
5
path B
Co
Lx
Ph
Ph
8
Scheme 7.
phenyl substituent next to the cobalt centre. From this
species, the formation of the cyclooctatriene 5 has to
proceed by an insertion either in the vinyl-cobalt bond
(path A) via 7 or into the allylic substructure on the
side of the isoprene (path B) via 8. The formation of
5 can be rationalised by reductive elimination from
the intermediates 7 and 8. At this point we are not
able to differentiate whether path A or path B is the
more favourable reaction pathway.
For the cyclotrimerisation process and the effects ob-
served in different nitrile type solvents, effects based on
sterically modifications on the diimine ligands as well
as the regioselectivity changes in the three tested solvents
CH3CN, CH2Cl2 and THF one can speculate that a sol-
vent molecule could be present in the ligand sphere dur-
ing the reaction pathway. A definite interpretation of the
data or a plausible unifying transition state model
explaining all observed effects can not be given at the
present time.
3. Summary
Within this investigation, we were able to show that
additives influenced the reactivity of cobalt phosphine
and cobalt diimine type complexes dramatically. Firstly,
the dimerisation process of phenylacetylene can be catal-
ysed in acceptable yields with CoBr2(dppe) as catalyst
precursor using magnesium as reducing agent. Secondly,
the addition of a Lewis acid and the substitution of mag-
nesium towards zinc as reducing agent favoured the
cyclotrimerisation reaction of phenylacetylene in aceto-
nitrile. The best results for the cyclotrimerisation towards
the unsymmetrical product 2 were obtained for diimine
type ligands. In contrast, when diphenylethylene disulfide
was used as ligand the regioselectivity was inversed and
the symmetric trimer 3 is obtained as major product.
Diimine cobalt complexes gave excellent results in aceto-
nitrile as solvent. Furthermore, the investigation showed
that the solvent has a profound effect which can only be

G. Hilt et al. / Journal of Organometallic Chemistry 690 (2005) 5170–5181
5179
rationalised when a coordination of the nitrile is assumed
in the transition state. Thirdly, the chemoselectivity of
the complexes could be demonstrated in competition
experiments where either the Diels–Alder product 4 or
the cyclotrimerisation product 5 was formed predomi-
nantly. The competition experiment in dichloromethane
led to the formation of an unprecedented [4+2+2]-cycli-
sation product. The regiochemistry of this cyclooctatri-
ene derivative 5 gave a valuable insight into the
regioselective formation of the intermediate in the cobalt
catalysed Diels–Alder reaction.
fied by flash chromatography using pentane/CH2Cl2
100:1 as eluent. (E)-1,4-Diphenyl-but-1-en-3-yne (1)
and an inseparable mixture of cyclic and acyclic trimers
of phenylacetylene were isolated. The spectral data of
the products isolated in pure form are consistent with
the literature [5].
4.4. Representative procedure for the cyclotrimerisation
reaction
A suspension of the cobalt complex (0.1 mmol, 5.0
mol%), zinc dust (13 mg, 0.2 mmol, 10.0 mol%) and
anhydrous ZnI2 (64 mg, 0.2 mmol, 10.0 mol%) in anhy-
drous solvent (CH3CN, CH2Cl2 or THF) (1.0 mL) un-
der argon atmosphere was shortly heated to reflux and
allowed to cool to room temperature. After 15 min phe-
nylacetylene (0.22 mL, 2.0 mmol) was added and the
mixture stirred at room temperature. The reaction was
monitored by thin layer chromatography. After com-
plete conversion the brown reaction mixture was passed
through a pad of silica using CH2Cl2 as eluent. The sol-
vents were removed and the crude product purified by
flash chromatography using pentane/CH2Cl2 100:1 as
eluent. The ratio of regioisomeres 2:3 was verified by
4. Experimental
4.1. General information
All manipulations were carried out using standard
Schlenk techniques under an atmosphere of argon or
nitrogen. All solvents for use in an inert atmosphere were
purified by standard procedures and distilled under nitro-
gen immediately prior to use. Other chemicals were ob-
tained from commercial sources and used without
1
further purification. H and 13C NMR spectra were re-
1
corded on Bruker Physics ARX 300 and Bruker Physics
ARX 400 spectrometer in CDCl3. The spectra are cali-
brated to the residual protons of the solvent (1H) or to
the 13-carbon signals of the solvent (13C). Mass spectra
were recorded on a Varian MAT CH 7a, a Finnigan
MAT 95S or a Micromass VG AutoSpec spectrometer.
integration of the H NMR.
4.5. Competition experiments
Following the general procedure for the trimerisation
reactions, bis-cyclohexyliden ethane diimine cobalt dibr-
omide (44 mg, 0.1 mmol, 5.0 mol%), zinc dust (13 mg,
0.2 mmol, 10.0 mol%) and anhydrous zinc iodide (64
mg, 0.2 mmol, 10.0 mol%) were suspended in dry solvent
(1.0 mL) under argon atmosphere, shortly heated to re-
flux and allowed to cool to room temperature. After 15
min isoprene (0.24 mL, 2.4 mmol) and phenylacetylene
(0.22 mL, 2.0 mmol) were added and the mixture stirred
at room temperature. The reaction was monitored by
thin layer chromatography (CH3CN 0.5 h, CH2Cl2 5.5
h, THF 5.0 h). After complete conversion the brown
reaction mixture was passed through a pad of silica
using CH2Cl2 as eluent. The solvents were removed
and the crude product was purified by flash chromatog-
raphy using pentane/CH2Cl2 100:1 as eluent.
4.2. Synthesis of diimine complexes
The ligands were synthesised as previously described in
the literature or by adapting known procedures [16]. The
complexes were synthesised by suspending anhydrous
CoBr2 (2.18 g, 10.0 mmol) and the ligand (10.0 mmol) in
anhydrous THF (40 mL) at room temperature [17]. After
stirring overnight the solvent was removed in vacuo and
the solid product was used without further purification.
4.3. Representative procedure for the di- and trimerisation
reaction
A suspension of CoBr2(dppe) (62 mg, 0.1 mmol, 5.0
mol%), zinc dust and magnesium chips in anhydrous
CH3CN (1.0 mL) under argon atmosphere was shortly
heated to reflux, allowed to cool to room temperature
and stirred for 30 min. The reactions without zinc dust
were again shortly heated to reflux, allowed to cool to
room temperature and stirred for additional 30 min.
Phenylacetylene (0.22 mL, 2.0 mmol) was added to the
brown suspension and the mixture stirred at room tem-
perature for 10 min. The brown reaction mixture was
passed through a pad of silica using CH2Cl2 as eluent.
The solvents were removed and the crude product puri-
(1Z,3Z,6Z)-6-Methyl-2,3-diphenyl-cycloocta-1,3,6-
trien (5): 1H NMR (400 MHz, CDCl3): d = 7.43 (dt, 4 H,
J = 7.2, 1.2 Hz), 7.29 (t, 4 H, J = 7.5 Hz), 7.22 (dt, 2 H,
J = 7.3, 1.2 Hz), 6.24 (t, 1 H, J = 7.8 Hz), 6.20 (t, 1 H,
J = 7.8 Hz), 5.56 (tq, 1 H, J = 6.0, 1.2 Hz), 2.87 (bs, 4
H), 1.87 (d, 3 H, J = 1.2 Hz). 13C NMR (100 MHz,
CDCl3): d = 141.1, 140.6, 138.9, 135.1, 128.4, 127.1,
127.0, 126.8, 126.7, 123.9, 123.4, 120.2, 34.0, 28.3,
27.3. MS (EI): m/z = 272 (47, M+), 257 (23), 244 (100),
229 (51), 217 (24), 202 (14), 178 (17), 165 (23), 115
(19), 91 (17). HRMS calculated for C21H20:
m/z=272.1565, found: m/z = 272.1559.

5180
G. Hilt et al. / Journal of Organometallic Chemistry 690 (2005) 5170–5181
P. Biagini, T. Funaioli, G. Fachinetti, F. Laschi, P.F. Zanazzi, J.
Acknowledgements
Chem. Soc., Chem. Commun. (1989) 405.
[8] M.-S. Wu, D.K. Rayabarapu, C.-H. Cheng, Tetrahedron 60
(2004) 10005;
We thank the German science foundation (DFG) for
the financial support. We also thank Prof. F. Glorius,
Philipps-Universita¨t Marburg, for the donation of the li-
gand used in Table 4, entries 34–36.
ˇ ´
P. Turek, M. Kotora, I. Tislerova, K. Hocek, I. Votruba, I.
´
Cısarova, J. Org. Chem. 69 (2004) 9224;
´ ˇ
ˇ
L. Dufkova, I. Cısarova, P. Stepnicka, M. Kotora, Eur. J. Org.
´
´
ˇ
´
ˇ
Chem. (2003) 2882;
F. Slowinski, C. Aubert, M. Malacria, Adv. Synth. Catal. 343
(2001) 64.
[9] For the synthesis of the cobalt complexes see: M.C. Barral, E.
References
´
Delgado, E. Gutierrez-Puebla, R. Jimenez-Aparicio, A. Monge,
C. del Pino, A. Santos, Inorg. Chim. Acta 74 (1983) 101;
[1] For reviews see: M. Malacria, C. Aubert, J.L. Renaud, in: M.
Lautens, B.M. Trost (Eds.), Science of Synthesis: Houben–Weyl
Methods of Molecular Transformations, vol. 1, Thieme, Stuttgart,
2001, pp. 439–530;
´
´
M.J. Camazon, A. Alvarez-Valdes, J.R. Masaguer, M.C. Nav-
arro-Ranninger, Transition Met. Chem. 11 (1986) 334;
T.V. Laine, M. Klinga, A. Maaninen, E. Aitola, M. Leskela¨,
Acta Chem. Scand. 53 (1999) 968.
S. Saito, Y. Yamamoto, Chem. Rev. 100 (2000) 2901;
I. Ojima, M. Tzamarioudaki, Z. Li, R.J. Donovan, Chem. Rev.
96 (1996) 635;
[10] J.J. Banewicz, J.A. Maguire, P.O. Smith, D. Lussky, D.W.
Sudbrink, D.E. Allen, J. Chem. Thermodyn. 13 (1981) 629;
V. Gutmann, O. Bohunovsky, Monatsh. Chem. 99 (1968)
740;
M. Lautens, W. Klute, W. Tam, Chem. Rev. 96 (1996) 49;
D.B. Grotjahn, in: E.W. Abel, F.G.A. Stone, G. Wilkinson, L.
Hegedus (Eds.), Comprehensive organometallic chemistry II, vol.
12, Pergamon Press, Oxford, 1995, pp. 741–770;
N.E. Schore, in: B.M. Trost, I. Fleming, L.A. Paquette (Eds.),
Comprehensive organic synthesis, vol. 5, Pergamon Press, Oxford,
1991, pp. 1129–1162;
B.J. Hathaway, D.G. Holah, J. Chem. Soc. (1964) 2400.
[11] For the reaction mechanism of cyclotrimerisation reactions see:
N.E. Schore, Chem. Rev. 88 (1988) 1081;
K.P.C. Vollhardt, Angew. Chem. 96 (1984) 525;
K.P.C. Vollhardt, Angew. Chem. Int. Ed. 23 (1984) 539;
H. Bo¨nnemann, Angew. Chem. 90 (1978) 517;
K.P.C. Vollhardt, Angew. Chem. 96 (1984) 525;
K.P.C. Vollhardt, Angew. Chem. Int. Ed. 23 (1984) 539;
For recent examples of cobalt catalysed cyclotrimerisations of
alkynes see: L. Yong, H. Butenscho¨n, Chem. Commun. (2002)
2852;
H. Bo¨nnemann, Angew. Chem Int. Ed. 17 (1978) 525;
H. Yamazaki, Y. Wakatsuki, J. Organomet. Chem. 139 (1977) 157.
[12] H. Bo¨nnemann, Angew. Chem. 97 (1985) 264;
H. Bo¨nnemann, Angew. Chem. Int. Ed. 24 (1985) 248;
H. Bo¨nnemann, W. Brijoux, in: A.R. Katritzky (Ed.), Advances
in Heterocyclic Chemistry, vol. 48, Academic Press, San Diego,
1990, pp. 177–222;
T. Sugihara, A. Wakabayashi, Y. Nagai, H. Takao, H. Imagawa,
M. Nishizawa, Chem. Commun. (2002) 576;
M.S. Sigman, A.W. Fatland, B.E. Eaton, J. Am. Chem. Soc. 120
(1998) 5130.
[2] For leading references see: G. Hilt, S. Luers, K.I. Smolko, Org.
¨
L.V.R. Bon˜aga, H.-C. Zhang, B.E. Maryanoff, Chem. Commun.
(2004) 2394;
Lett. 7 (2005) 251;
G. Hilt, S. Luers, K. Harms, J. Org. Chem. 69 (2004) 624;
¨
L.V.R. Bon˜aga, H.-C. Zhang, D.A. Gauthier, I. Reddy, B.E.
Maryanoff, Org. Lett. 5 (2003) 4537;
´ ´
J.A. Varela, L. Castedo, M. Maestro, J. Mahıa, C. Saa, Chem.
G. Hilt, K.I. Smolko, Angew. Chem. 115 (2003) 2901;
G. Hilt, K.I. Smolko, Angew. Chem. Int. Ed. 42 (2003) 2795.
[3] G. Hilt, F.-X. du Mesnil, Tetrahedron Lett. 41 (2000) 6757.
[4] H. Brunner, A. Reimer, Bull. Chem. Soc. Fr. 134 (1997) 307;
M. Lautens, W. Tam, J.C. Lautens, L.G. Edwards, C.M.
Crudden, A.C. Smith, J. Am. Chem. Soc. 117 (1995) 6863;
O. Pardigon, A. Tenaglia, G. Buono, J. Org. Chem. 60 (1995)
1868;
Eur. J. 7 (2001) 5203;
G.P. Chiusoli, M. Costa, Z. Zhou, Gazz. Chim. Ital. 122 (1992)
441.
[13] Investigations for the expansion of the [4+2+2] cycloaddition
reaction to other 1,3-dienes and alkynes are underway.
[14] P.A. Evans, E.W. Baum, A.N. Fazal, M. Pink, Chem. Commun.
(2005) 63;
O. Pardigon, G. Buono, Tetrahedron: Asymmetry 4 (1993) 1977;
I.-F. Duan, C.-H. Cheng, J.-S. Shaw, S.-S. Cheng, K.F. Liou, J.
Chem. Soc., Chem. Commun. (1991) 1347.
B. Ma, J.K. Snyder, Organometallics 21 (2002) 4688;
S.R. Gilbertson, B. DeBoef, J. Am. Chem. Soc. 124 (2002)
8784;
[5] Other phosphine ligands were not tested; compare [6].
[6] For dimerisation processes of alkynes see: K. Komeyama, K.
Takehira, K. Takaki, Synthesis (2004) 1062;
P.A. Evans, J.E. Robinson, E.W. Baum, A.N. Fazal, J.
Am. Chem. Soc. 124 (2002) 8782;
Correction: P.A. Evans, J.E. Robinson, E.W. Baum, A.N.
Fazal, J. Am. Chem. Soc. 125 (2003) 14648;
L.D. Field, A.J. Ward, J. Organomet. Chem. 681 (2003) 91;
´
A.M. Trzeciak, J.J. Ziołkowski, Pol. J. Chem. 77 (2003) 749;
S. Pasynkiewicz, A. Pietrzykowski, E. OleEdzka, J. Wang, Pol. J.
Chem. 77 (2003) 701;
Y. Chen, J.K. Snyder, J. Org. Chem. 63 (1998) 2060;
A. Greco, A. Carbonaro, G. DallÕAsta, J. Org. Chem. 35
(1970) 271.
W.A. Herrmann, V.P.W. Bohm, C.W.K. Gstottmayr, M. Gro-
sche, C.-P. Reisinger, T. Weskamp, J. Organomet. Chem. 617
(2001) 616;
[15] For the mechanism of the cobalt catalysed cycloaddition reactions
see also: M.-S. Wu, D.K. Rayabarapu, C.-H. Cheng, Tetrahedron
60 (2004) 10005;
L.D. Field, A.J. Ward, P. Turner, Aust. J. Chem. 52 (1999) 1085;
H.-Y. Rhyoo, B.Y. Lee, H.K.B. Yu, Y.K. Chung, J. Mol. Catal.
92 (1994) 41;
O. Pardigon, A. Tenaglia, G. Buono, J. Mol. Catal. A 196 (2003)
157;
B. Ma, J.K. Snyder, Organometallics 21 (2002) 4688;
M. Lautens, W. Tam, J.C. Lautens, L.G. Edwards, C.M.
Crudden, A.C. Smith, J. Am. Chem. Soc. 117 (1995) 6863, The
R. Herrmann, A.J.L. Pombeiro, Monatsh. Chem. 119 (1988) 583.
[7] G. Hilt, T. Vogler, W. Hess, F. Galbiati, Chem. Commun. (2005)
1474;
cyclooctatriene product 5 could not be observed when
CoBr2(dppe) was used as catalyst precursor in dichloromethane
M.-S. Wu, M. Shanmugasundaram, C.-H. Cheng, Chem. Com-
mun. (2003) 718.

G. Hilt et al. / Journal of Organometallic Chemistry 690 (2005) 5170–5181
5181
during our investigations on the cobalt catalysed Diels–Alder
reactions of non-activated starting materials [2], not even in traces
when an excess of the alkyne was used.
A. Alexakis, N. Lensen, J.-P. Tranchier, P. Mangeney, J. Feneau-
Dupont, J.P. Declercq, Synthesis (1995) 1038;
N. De Kimpe, L. DÕHondt, E. Stanoeva, Tetrahedron Lett. 32
(1991) 3879;
[16] G.A. Grasa, M.S. Viciu, J. Huang, S.P. Nolan, J. Org. Chem. 66
(2001) 7729;
J.M. Kliegman, R.K. Barnes, Tetrahedron 26 (1970) 2555.
[17] M.C. Barral, E. Delgado, E. Gutierrez-Puebla, R. Jimenez-
Aparicio, A. Monge, C. Del Pino, A. Santos, Inorg. Chim. Acta
74 (1983) 101.
M. Haaf, A. Schmiedl, T.A. Schmedake, D.R. Powell, A.J.
Millevolte, M. Denk, R. West, J. Am. Chem. Soc. 120 (1998)
12714;