G. Achonduh et al. / Tetrahedron 71 (2015) 1241e1246
1243
other hand, the use of pure toluene as solvent (Table 1, entries
13e17) proved to be beneficial irrespective of the CO/H2 pressure
employed. Substituting the acid co-catalyst 2-ethylhexanoic acid,
with formic acid or acetic acid did not significantly alter the out-
come of the reaction (Table 1, entries 16 and 17). Based on the data
from Table 1, it was evident that the phosphaadamantane ligand L1
gave significant amounts of branched products and other isomers
while furnishing less of the desired linear aldehyde. It could be that
this ligand is too bulky and less basic to efficiently coordinate to the
cobalt center in order to influence the regioselectivity.
To elaborate the influence of ligand structure on the reactivity of
the catalyst, the remaining phosphines from Fig. 1 were employed
alongside other known ligands such as tri(p-tolyl)phosphine [P(p-
tol)3] and 1,10-bis(diphenylphosphino)ferrocene [dppf] as shown in
Table 2. Based on the results in Table 1, and the ease of isolation of
the hydroformylation products as alcohols, we selected the reaction
conditions from Table 1, entry 15 to carry out the ligand study in
Table 2.
To expand the scope of this protocol to other olefins, we used the
best conditions above where Lim-10 is the ligand of choice. As
shown in Table 3, a variety of alkenes were screened.
In all cases, full conversion was achieved. However, while
moderate linearity was obtained for most alkenes (Table 3, entries
1e3 and 5); the best selectivity was obtained with allylcyclohexene
(Table 3, entry 4). The results show that the protocol tolerates al-
kenes with various structural complexity.
Although the reaction conditions developed so far seemed ap-
propriate, it was desirable to know if the amount of cobalt and li-
gand could be further reduced to 2.5 mol % and 5 mol %,
respectively, given that the stoichiometric relationship between
cobalt and phosphine will still be maintained in both cases. Also, it
was equally useful to find out if the use of 2-ethylhexanoic acid and
KOH/EtOH was necessary or if such a combination could be
replaced by a simpler one such as sodium acetate (NaOAc). Further
optimization studies were conducted (Table 4). In this case, NaBH4
(1.2 equiv relative to the alkene) and MeOH (5 mL) were added to
Table 2
Ligand studya
Entry
Ligand
Conv.b [%]
Selectivityc [L/B/I]
1
2
3
4
5
6
7
8
CYTOPÒ 292
Lim-benzyl
Lim-MeCy
Lim-4
100
100
100
94
58:22:20
69:17:14
66:18:16
68:16:16
67:18:15
77:14:09
73:16:11
69:15:16
73:15:12
d
Lim-6
96
Lim-10
Lim-18
L2
L3
P(p-tol)3
Dppf
100
100
100
100
5
20
4
100
9
10
11
12
13
69:23:08
d
PBu3
P(tBu)3
63:21:16
a
All reactions were performed using a stainless steel autoclave containing a glass liner; 1-decene (2.0 mmol), Co2(CO)8 (5 mol %), ligand (10 mol %), 2-ethylhexanoic acid
(10 mol %), KOH/EtOH (2.5 mol %), and CO/H2 (600 psi) in 5 mL toluene at 110 ꢀC for 22 h.
b
Percent conversion determined by H NMR spectroscopy based on alkene consumed.
Linear/branched/isomers ratio determined by H NMR.
c
When the phosphaadamantane ligand CYTOPÒ 292, which was
previously shown to be highly selective in Rh-catalyzed hydro-
formylation6b was examined, it gave good conversion, but the se-
lectivity for the linear aldehyde in this case was lower than
previously obtained with L1 (Table 1). A thorough examination of
bicyclic phosphine ligands derived from limonene (Table 2, entries
2e7) showed similarity in reactivity and selectivity. However, the
best selectivity for the linear aldehyde was obtained with Lim-10.
These findings are in line with earlier reports in which Phoban and
limonene-derived ligands were evaluated by means of kinetic
studies involving the hydroformylation of linear alkenes at high
temperatures (ꢁ170 ꢀC) and syn gas pressures (ꢁ60 bars).13 Other
ligands L2 and L3 were equally promising with selectivities close to
that of the limonene series. Dppf and P(p-tol)3 were found to be less
reactive under these reaction conditions. Contrary to the successes
of Slaugh and Mullineaux who originally investigated the catalytic
activity of PBu3 in hydrofromylation and achieved linearity ratios of
up to 8:1,8 our reaction conditions gave very little success with PBu3
(Table 2, entry 12). However when P(tBu)3 was employed, full
conversion occurred with lower selectivity for the linear product
compared to Lim-10 (Table 2, entry 13).
the crude reaction mixture and stirred for an additional hour at
room temperature. Isolated yields were obtained after flash column
chromatography on silica gel.
The results showed that the use of an additive is not necessary
to achieve high yields (Table 4, entries 3, 6, and 7). The results
also demonstrated that the combination of 2-ethylhexanoic acid
and KOH/EtOH can be replaced by a cheap and simple reagent
such as NaOAc (Table 4, entries 2, 4, and 8). However, it was
observed that there was a drop in linearity when no additive was
employed (Table 4, entries 3, 6, and 7). Thus, it can be argued that
the additives do have some effect on linearity, although the
overall outcome of the reaction is largely dependent on the
combined effect of both ligand and additives. On the other hand,
reducing the amount of cobalt and phosphine ligand to 2.5 mol %
and 5.0 mol %, respectively, proved more beneficial to the re-
action as improved yields were obtained (compare entries 1e4 to
entries 5e8). Although such an observation seems to contradict
the normal convention where an increase in catalyst loading
generally leads to a higher yield; in this case one could speculate
that at a higher catalyst loading, there is likely formation of other
active species in solution that leads to an undesirable pathway,