M.R. Torviso, et al.
MolecularCatalysis489(2020)110935
diazoacetate was not complete even after 24 h. At the end of the re-
action, the catalysts were separated by centrifugation and another
portion of diazoacetate was added to the solution to check the leaching
of active copper. No further cyclopropanation reaction was detected in
solution (< 2% increase in yield), showing that the contribution of the
homogeneous reaction is negligible and hence the heterogeneous
character of all the catalysts. Some of the recovered catalysts were re-
used under the same conditions.
The catalytic activity of PCuW11 and PCuW11-SR was very low
(entries 1 and 3), with very poor yields of cyclopropanes (< 8%) and
incomplete conversion of ethyl diazoacetate. The modification with the
chiral ligand did not significantly modified this behavior (entries 2 and
4), whereas the enantioselectivity obtained was only moderate (40–42
% ee for trans cyclopropanes). The increase in the amount of catalyst
(entry 5) produced an increase in the enantioselectivity for trans cy-
clopropanes, and also a slight increase in yield although it remained
very low. It is noteworthy, the selectivity is maintained in the reuse,
showing the stability of the copper sites formed. Another proof for the
stability of the lacunary species comes from the chemical analysis of
Box/PCuW11 after reaction (Table 1, entry 2). It can be seen that the
used catalyst possesses the same Cu/PW ratio (0.96) but a significant
decrease in the content of both copper and tungsten (0.21 mmol Cu/g
and 2.41 mmol W/g) due to the weight gain associated with the ad-
sorption of ligand and reaction products and by-products.
These results contrast with those obtained with catalysts prepared
by cation exchange. Box-Cu/PW12 (entry 6) leads to total conversion of
ethyl diazoacetate, with a fair chemoselectivity (38 % yield) and better
enantioselectivity (55 % ee for trans cyclopropanes), even with only 0.2
% catalyst. The immobilization on silica produces a catalyst with lower
catalytic activity but the same enantioselectivity (entry 7), that it is
fully recoverable with the same performance at least in two further
reaction cycles. It is necessary to use an amount of this catalyst as low
as 0.13 % (entry 8) to get results as bad as those obtained with the
lacunary catalysts. The catalyst with lower loading of Cu per PW12
(entry 9) leads to significant lower enantioselectivity, that it is not
improved even using 1% catalyst (entry 10), showing that it is due to a
structural effect of the catalyst and the full exchange on the hetero-
polyanion is required to get good results.
difficult reduction of PCuW11 would explain the lower activity, and an
experiment of comparative reductions were carried out using another
typical reductant for cyclopropanation reaction, phenylhydrazine [48],
to detect in solution the corresponding oxidized products. Several metal
oxides and salts have been used for this kind of reaction, such as Ag2O
[49], MnO2 [50] or Pb(OAc)4 [51]. It is described that on oxidation,
phenylhydrazine leads to molecular nitrogen and phenyl radicals that
evolve depending on the reaction medium. In this work we decided to
test three different copper(II) salts, in order to check the effect of the
anion in a polar non-protic solvent such as acetonitrile. The results are
The reaction with Cu(OAc)2 led to biphenyl as main product, with
phenol and azobenzene as the main by-products. In the case of CuCl2,
the chloride participates in the reaction and chlorobenzene is the main
product, with several by-products. Finally, the most typical cyclopro-
panation catalyst, Cu(OTf)2, led to azobenzene as the only product,
with no detectable by-products. These results indicate that the anion
influences the course of the phenylhydrazine oxidation. Additionally, in
all the cases it was observable a color change and/or the formation of a
precipitate. The addition of phenylhydrazine to PCuW11, even in a
larger excess did not show any observable change in the medium and
the analysis of the reaction mixture did show only the starting phe-
nylhydrazine with no detectable products. On the contrary, the addition
of phenylhydrazine to Box-Cu/PW12-SR did produce a color change of
the solid, from light greenish blue to grey. The analysis of the reaction
mixture showed a good part of the unreacted phenylhydrazine, that had
been added in a larger excess, together with comparable amounts of
several of the expected products: biphenyl, phenol, biphenylamine and
azobenzene. This result confirms the difference in the reductability of
PCuW11 and Box-Cu/PW12-SR and explains the very low activity of
PCuW11 as cyclopropanation catalyst. This effect on the catalytic ac-
tivity of copper had been already detected with strongly coordinative
anions, such as chloride, in solution [52] and the structure of the la-
cunar species can be considered as an anion of this type for copper. On
the other hand, the addition of a nitrogenated ligand has a positive
effect for reactions with ethyl diazoacetate, as seen in the case of car-
bene insertions [53], which is also in good agreement with the effect
observed in the Box/PCuW11 catalysts.
Regarding the structure of the catalysts, the PCuW11 lacunary spe-
cies has been described with anilinium cations as counter-ions [42],
although the orientation disorder in the crystal makes it difficult to
assign the Cu position. However, it is considered that the terminal
oxygen bound to copper corresponds to a coordinated water molecule,
as described also for other lacunary species, such as PAlW11 [43]. When
the box ligand is added, it would coordinate as a monodentate ligand by
substituting the water molecule. Considering the crystal structure of a
free box ligand with substituted phenyl groups [44], the structure of
Box/PCuW11 could be represented as in Fig. 2A. On the other hand, the
crystal structures of several combinations of polyoxometallates and
copper(II) complexes have been also reported. In some cases, one
terminal oxygen participates in the coordination sphere of copper,
whereas in other cases the complex acts as a cation. Even both roles can
be played by the same type complex in one single crystal structure [45].
Taking the structure of Box-CuBr2 as model [46], two structures for
Box-Cu/PW12 can be hypothesized, coordinated to terminal oxygen
(Fig. 2B) or as free cation, far from any coordinating atom of PW12
(4.5 Å from the closest terminal oxygen, Fig. 2C).
However, the hypothesis of a monodentate Box-Cu complex in the
case of Box/PCuW11 is not compatible with the moderate enantios-
electivity obtained (Table 2, entry 2), that should be much lower,
comparable to that obtained with monodentate oxazoline ligands [47].
Moreover, this type of structure would not either explain the lower
activity of Box/PCuW11, as shown by the incomplete consumption of
ethyl diazoacetate. In any case, all the copper(II) complexes must be
considered pre-catalysts, whereas the true catalysts are formed by the
Cu(II) to Cu(I) reduction promoted by ethyl diazoacetate. A more
Those catalytic results seem to indicate that the lacunary species
does not behave in the same way as the Box-Cu/PW12 catalysts pre-
pared by cation exchange. Thus, the identification of lacunary species
in the preparation of Box-Cu/PW12 based in IR and NMR spectra may be
wrong. Some experiments were carried out to check this hypothesis and
to confirm the nature of the sites in Box-Cu/PW12. The carbon and ni-
trogen analysis of the neutralized PW12 prepared according to our
previous paper with Et3N/PW12 molar ratio of 4.8 (1.2 eq of amine)
(Fig. 3b) show that not all the protonic sites are neutralized by trie-
thylamine. Thus, the Keggin PW12 was treated with higher amounts of
triethylamine (Et3N/PW12 molar ratios of 30 and 300, that is 10 and
100 eq). As can be seen, the excess of base produces a decrease in the
intensity of the bands assigned to the lacunary species, that appear first
as shoulders (Fig. 3c) and finally disappear when 100 equivalents of
amine are used (Fig. 3d). This result seems to indicate that in fact, a
lacunary species is not formed during the neutralization process, and
the bands assigned in principle to this kind of species correspond to the
partially neutralized Keggin species, that loses its symmetry due to the
presence of protons and triethylammonium cations in different
amounts. In fact, the loss of symmetry is considered the origin of the
changes in frequency observed in lacunary species [41]. Thus, the loss
of symmetry in the external part of the Keggin species is also able to
modify the IR spectrum in the same way as the change in the internal
structure of the Keggin polyoxometallate.
However, the NMR spectra in solution, where a fast exchange would
presumably take place, should show the features of a Keggin structure,
considered as an average of the partially neutralized species. Then, the
31P NMR spectra of Keggin PW12 and the different neutralized species
4