Mechanism of formation of the peroxocarbonate complex (PCy3)2Ni(CO4) from solid (PCy3)2Ni(CO2) and dioxygen: An example of solid-state metallorganic reaction involving CO2 deco-ordination and reinsertion into the O-O bond of (PCy3)2Ni(O2)

Article abstract of DOI:10.1016/S0020-1693(01)00737-X

Solid (PCy₃)₂Ni(CO₂) (1) reacts with dioxygen to afford the peroxocarbonate complex (PCy₃)₂Ni(CO₄) (3). The use of labelled ¹³CO₂, C¹⁸O₂, ¹⁸O₂ coupled with a FTIR study of both the gas-phase in equilibrium with the solid and the solid resulting complex, allows to propose the reaction mechanism that implies CO₂ deco-ordination, O₂ co-ordination, and CO₂ insertion into the O-O bond of the newly formed, reactive (PCy₃)₂Ni(O₂) complex 2. A normal mode analysis substantiates the band assignment and the proposed mechanism. Peroxocarbonate (3) exhibits a ¹³C resonance at 166.6 ppm and a single ³¹P signal at 43.1 ppm below 200 K. The reactivity of the peroxo-group as one-oxygen transfer agent prevents the spectroscopic characterisation of 3 in solution at room temperature. An out-of-sphere phosphine can be easily oxidised. A gaseous olefin, mono-ene (ethylene) or diene (allene), added to 3, makes easier the deco-ordination process of one phosphine ligand, that is then oxidised to phosphine oxide. The gaseous olefin itself is not oxidised until free phosphine is present in the medium. In solution, styrene is oxidised through a two- or one-oxygen transfer pathway, according to the reaction condition.

Full text of DOI:10.1016/S0020-1693(01)00737-X

www.elsevier.com/locate/ica
Inorganica Chimica Acta 330 (2002) 63–71
Mechanism of formation of the peroxocarbonate complex
(PCy3)2Ni(CO4) from solid (PCy3)2Ni(CO2) and dioxygen: an
example of solid-state metallorganic reaction involving CO2
deco-ordination and reinsertion into the OꢀO bond of
(PCy3)2Ni(O2). Reactivity of the peroxocarbonate complex towards
olefins in the solid state and in solution
Michele Aresta a,*, Immacolata Tommasi a, Angela Dibenedetto a, Monique Fouassier b,
Joelle Mascetti b
a Department of Chemistry and METEA, Research Center, Uni6ersity of Bari, Campus Uni6ersitario, 70126 Bari, Italy
b Laboratoire de Physico-Chimie Moleculaire, Uni6ersite` de Bordeaux I 33405 Talence, Cedex, France
Received 28 June 2001; accepted 25 September 2001
Dedicated to Professor L.M. Venanzi
Abstract
Solid (PCy3)2Ni(CO2) (1) reacts with dioxygen to afford the peroxocarbonate complex (PCy3)2Ni(CO4) (3). The use of labelled
13CO2, C18O2, 18O2 coupled with a FTIR study of both the gas-phase in equilibrium with the solid and the solid resulting complex,
allows to propose the reaction mechanism that implies CO2 deco-ordination, O2 co-ordination, and CO2 insertion into the OꢀO
bond of the newly formed, reactive (PCy3)2Ni(O2) complex 2. A normal mode analysis substantiates the band assignment and the
proposed mechanism. Peroxocarbonate (3) exhibits a 13C resonance at 166.6 ppm and a single 31P signal at 43.1 ppm below 200
K. The reactivity of the peroxo-group as one-oxygen transfer agent prevents the spectroscopic characterisation of 3 in solution at
room temperature. An out-of-sphere phosphine can be easily oxidised. A gaseous olefin, mono-ene (ethylene) or diene (allene),
added to 3, makes easier the deco-ordination process of one phosphine ligand, that is then oxidised to phosphine oxide. The
gaseous olefin itself is not oxidised until free phosphine is present in the medium. In solution, styrene is oxidised through a two-
or one-oxygen transfer pathway, according to the reaction condition. © 2002 Published by Elsevier Science B.V.
Keywords: Peroxocarbonates; Mechanism; Solid-state metallorganic; Olefin; Oxidation
1. Introduction
[4,8] using the impulse oscillation model (IOM) [3].
Spectroscopic studies based on the use of labelled CO2
and O2 (13CO2, C18O2, 18O2) complexes coupled with a
normal mode analysis and a calculation of vibrational
frequencies, have clearly demonstrated that carbon
dioxide inserts into the OꢀO bond of the dioxygen
complex with formation of the peroxocarbonate moi-
ety. The correct coupling of labelled CO2 and O2 spe-
cies brings to the formation of asymmetric 18Oꢀ16O
peroxo-groups that we have used for mechanistic stud-
ies on the oxygen transfer to an oxophile [2,4–6,8],
showing that the oxygen atom linked to the metal is
transferred. We have reported long ago that the carbon
The formation of transition metal peroxocarbonates
from dioxygen complexes and carbon dioxide is known
since long [1]. Recently we have reported experimental
evidence on the reaction mechanism of a Rh–dioxygen
complex with carbon dioxide [2] also supported by a
theoretical normal mode analysis and modelling study
* Corresponding author. Tel.: +39-080-544 2430; fax: +39-080-
544 2429.
E-mail address: aresta@metea.uniba.it (M. Aresta).
0020-1693/02/$ - see front matter © 2002 Published by Elsevier Science B.V.
PII: S0020-1693(01)00737-X

64
M. Aresta et al. / Inorganica Chimica Acta 330 (2002) 63–71
dioxide Ni-complex (PCy3)2Ni(CO2) (1) is able to react
with dioxygen and to afford peroxocarbonate (3),
(PCy3)2Ni(CO4) [7]. The reaction pathway was not es-
tablished at that time.
We have now re-examined this reaction in order to
ascertain if dioxygen were able to insert into the NiꢀC
bond of 1, a reaction that appears to be spin forbidden.
The reaction has been carried out in a heterogeneous
system (solid complex–gas dioxygen) in order to elimi-
nate the solvent influence.
The experimental results suggest that the reaction
takes place through a carbon dioxide deco-ordination,
dioxygen co-ordination to Ni and CO2 insertion into
the OꢀO bond.
The results of this study show that the formation of
the peroxo-moiety requires an activated dioxygen
molecule that interacts with carbon dioxide in an inter-
molecular mode [8]. The incoming hetero-cumulene at-
tacks one oxygen of the co-ordinated O2 and this causes
the OꢀO bond splitting, more than the NiꢀO, or in
general the MꢀO bond cleavage [2].
2.1.2. Preparation of (PCy3)2Ni(CO4) (3)
Solid (PCy3)2Ni(CO2) (200 mg, 0.3 mmol) was al-
lowed to react with O2 at 278 K. The FTIR spectrum of
the gas evolved, in equilibrium with the solid, during
the reaction showed the presence of CO2 (asymmetric
stretching OꢁCꢁO mode at 2340 cm−1). After 8 h the
original yellow colour disappeared and a pink solid was
formed that was used for IR analysis. At this time the
disappearence of the CO2 in gas phase in equilibrium
with the solid was observed.
IR (nujol, KBr disk): 1685 cm−1 (wCꢁO), 1250 cm−1
(wCꢀO), 995 cm−1 (wOꢀO), 610 cm−1 (wNiꢀO+
wCꢀO).
If the equilibrium gas was pumped off at regular
intervals of time and replaced by fresh O2, the yield of
peroxocarbonate was lowered and new species were
formed. This procedure eliminates the released CO2
that cannot re-insert into the Ni(O2) complex that is
formed and accumulates in the reaction vessel.
2.1.3. Preparation of (PCy3)2Ni(13CO4)
The reactivity of the formed peroxocarbonate has
been studied both in the solid state and in solution.
Interestingly, any agent able to displace a phosphine
ligand (solvent, additional ligand) causes the oxygen
transfer from the peroxocarbonate to the external
phosphine.
Solid (PCy3)2Ni(13CO2) (100 mg, 0.15 mmol) was
reacted with 16O2 at 278 K. The gas in equilibrium with
the solid phase showed the presence of 13CO2 (asym-
metric stretching Oꢁ13CꢁO mode at 2282 cm−1 in the
FTIR spectrum). After 8 h the yellow starting complex
was converted into a pink solid that was isolated and
analysed. At this time 13CO2 was not present in gas
phase in equilibrium with the solid.
IR (nujol, KBr disk): 1640 cm−1 (wCꢁO), 1230 cm−1
(wCꢀO), 995 cm−1 (wOꢀO), 760 cm−1 k(CꢁO).
31P and 13C NMR spectra of complex
(PCy3)2Ni(13CO4) were measured in CH2Cl2 (C3H6O-d6
as deuterium reference was placed into a coaxial glass
tube device) at 203 K.
2. Experimental
2.1. Materials
Air sensitive compounds were handled under Ar with
a vacuum–inert gas manifold and glovebox. FTIR
spectra were recorded by using a Bruker 113V Fourier
transform interferometer.
31P NMR: 43.2 ppm, s (225 K).
13C NMR: resonance at 166.6 ppm due to the 13C
carbon of the peroxocarbonate group (213 K).
Frequencies are accurate to 91 cm−1. Solid samples
were studied as Nujol mull (Nujol was previously dried
on sodium wire and bubbled with Ar).
2.1.4. Preparation of (PCy3)2Ni(18O2CO2)
Solid (PCy3)2Ni(CO2) (120 mg, 0.18 mmol) was re-
acted with 18O2 at 278 K. The gas in equilibrium with
the solid phase showed the presence of CO2 (asymmet-
ric stretching OꢁCꢁO mode at 2340 cm−1 in the FTIR
spectrum). After 9 h the yellow starting complex was
converted into a pink solid that was isolated and
analysed. At this time the gas phase in equilibrium with
the solid did not show the presence of CO2.
Anal. Calc. for C37H66Ni16O218O2P2: C, 63.53; H, 9.51;
Ni, 8.39; P, 8.85. Found: C, 63.50; H, 9.49; Ni, 8.40; P,
8.85%.
GC and MS analyses were performed with a DANI
GC and a Shimadzu GC–MS. NMR spectra were
measured with Varian 200 XL or Bruker 500 MHz
equipments. All solvents were dried and distilled under
dinitrogen.
(PCy3)2Ni(CO2) (1) was prepared as described in the
literature [9].
2.1.1. Preparation of (PCy3)2Ni(C18O2) (1b),
(PCy3)2Ni(13CO2) (1c)
A solution of [(PCy3)2Ni]2N2 [9] (0.50 g, 0.4 mmol) in
C6H5CH3 was reacted with C18O2 (1b) or 13CO2 (1c) at
250 K. The yellow solid that separated was filtered off,
washed with cold (250 K) C6H5CH3, and dried under
vacuum.
2.1.5. Preparation of (PCy3)2Ni(16O2C18O2)
Solid (PCy3)2Ni(C18O2) (150 mg, 0.224 mmol) was
reacted with 16O2 at 278 K. The gas in equilibrium with
the solid phase showed the presence of C18O2 (asym-

M. Aresta et al. / Inorganica Chimica Acta 330 (2002) 63–71
65
metric stretching 18OꢁCꢁ18O mode at 2265 cm−1 in the
FTIR spectrum). After 8 h the yellow starting complex
was converted into a pink solid that was isolated and
analysed. At this time C18O2 was not present in the gas
phase in equilibrium with the solid.
2.1.10. Reaction of (PCy3)2Ni(CO4) with allene
Solid (PCy3)2Ni(CO4) (0.05 g, 0.07 mmol) was stirred
at 270 K under C3H4 atmosphere. After few minutes
the original pink solid became light greenish. The IR
spectrum once again showed the formation of carbon-
ate and phosphine oxide.
Anal. Calc. for C37H66Ni16O218O2P2: C, 63.53; H, 9.51;
Ni, 8.39; P, 8.85. Found: C, 63.30; H, 9.39; Ni, 8.30; P,
8.90%.
IR (nujol, KBr disk): 1620 cm−1 w(CꢁO), 1160 cm−1
w(PꢁO).
At 298 K the reaction goes in 2 min.
2.1.6. Synthesis of (PCy3)2Ni(18O213CO2)
Solid (PCy3)2Ni(13CO2) (150 mg, 2.25 mmol) was
reacted with 18O2 at 278 K. The gas in equilibrium with
the solid phase was analysed using the FTIR technique
and the presence of 13CO2 was clearly shown (band at
2282 cm−1). After 9 h the solid assumed a quite
homogeneous pink–salmon colour and was isolated as
reported above.
2.1.11. Reaction of (PCy3)2Ni(CO4) with styrene in
THF
Complex 2 was dissolved in THF at 270 K and
styrene (molar ratio 2:1 with Ni) was added. The
solution was stirred under dinitrogen and samples with-
drawn at regular interval of time and analysed by
GC–MS. The formation of benzaldehyde, styrene
epoxide, phenylacetaldehyde and methyl phenyl ketone
was confirmed (molar ratio 1:5:1:5) by comparison with
authentic samples.
12 13
Anal. Calc. for C36CH66Ni16O218O2P2: C, 63.57; H,
9.49; Ni, 8.37; P, 8.84. Found: C, 63.48; H, 9.40; Ni,
8.36; P, 8.78%.
2.1.7. Con6ersion of (PCy3)2Ni(13CO4) in CH2Cl2
A solution of (PCy3)2Ni(13CO4) in CD2Cl2 was pre-
pared at 213 K and the 13C and 31P NMR spectra
registered.
3. Discussion
In order to gather information on the mechanism of
conversion of (PCy3)2Ni(CO2) (1) into the peroxocar-
bonate complex (PCy3)2Ni(CO4), we have performed
the reaction of 1 with O2 using the following isotopic
species: 12C16O2, 13C16O2, 12C18O2, 16O2, 18O2.
13C{31P} NMR (CD2Cl2, 200 MHz, 213 K, C3H6O-
d6: ext. ref.): 166.6 ppm (s, CꢁO of peroxocarbonate
moiety).
31P NMR (CH2Cl2, C3H6O-d6 as internal reference,
200 MHz, 213 K, H3PO4 ref.): 43.2 ppm (s).
The solution was heated to 294 K for 20 min and the
colour changed to black from yellow.
Solid NiꢀCO2 complex (1) was reacted with *O2
under vigorous magnetic stirring in a vessel (10 mL)
kept at 278 K. The equilibrium gas was analysed by
FTIR. Starting from differently labelled CO2 and O2,
according to the specific reaction path, different perox-
ocarbonate species were expected identifiable through
the FTIR spectrum. The first important evidence was
that when 1, or its labelled analogues 13CO2 or C18O2,
is reacted with O2 (16O2 or 18O2) the gas phase shows
immediately the appearance of CO2 (or its labelled
analogues), then disappears with time. This brings to
propose reaction 1 as the first step of the solid-state
reaction.
31P NMR (CD2Cl2, 200 MHz, 294 K, H3PO4 ref.):
10.86 ppm (s, weak, free phosphine), 31.21 (s, medium,
carbonate complex), 49.8 (s, weak, free phosphine
oxide).
2.1.8. Stability of (PCy3)2Ni(CO4) under N2 at room
temperature (r.t.)
Solid (PCy3)2Ni(CO4) (0.05 g, 0.07 mmol) was stirred
at 298 K under N2 atmosphere. After 4 h the original
pink solid was changed to the white–greenish carbon-
ate complex (PCy3)(OPCy3)Ni(CO3).
The IR spectrum (KBr, Nujol mull) showed broad
bands at 1620 cm−1 (CO32− moiety) and 1160 cm−1
(phosphine oxide).
(PCy3)2Ni(CO2)solid+O2“(PCy3)2Ni(O2)solid+CO2
1
2
(1)
Complex 2, reported in the literature as a white
powder [10], results to be quite unstable also in the
solid state. The white compound decomposes to a dark
solid. The decomposition is very fast at room tempera-
ture and requires a few minutes at temperature below
273 K, making difficult its elemental analysis. Such
decomposition (rapid oxidation of the phosphine lig-
and) takes place also if 2 is present in traces in mixture
with 3. In this case the whole sample requires careful
handling for IR or other characterisations. The purifi-
cation of 2 by crystallisation at low temperature is
2.1.9. Reaction of (PCy3)2Ni(CO4) with ethylene
Solid (PCy3)2Ni(CO4) (0.05 g, 0.07 mmol) was put
under C2H4 atmosphere and stirred at 270 K. After 4 h
the original pink solid became light greenish. The solid
showed an IR spectrum similar to that of the complex
kept under N2 at r.t. Ethylene bands were masked by
ligand and carbonate bands.
IR (nujol, KBr disk): 1620 cm−1 w(CꢁO), 1160 cm−1
w(PꢁO).

66
M. Aresta et al. / Inorganica Chimica Acta 330 (2002) 63–71
made impossible by the fast decomposition in common
solvents. Complex 2 reacts readily with CO2 affording
3. We assume that starting from 1, the first step of the
reaction is the exchange O2/CO2 that produces 2, the
reactive species precursor of peroxocarbonate (3). It is
worth knowing that if the gas in equilibrium with the
solid sample (reaction 1) is pumped off at regular
intervals of time and replaced by pure O2, the yield of
the peroxocarbonate is much lower as the effect of loss
of CO2, while complex 2 is formed in large amount.
The discussion that follows is, thus, based on the
assumption of the formation of the NiꢀO2 complex
from 1 [10,11].
presence of free CO2 (asymmetric stretching OꢁCꢁO at
2340 cm−1 for reactions 2 and 3; at 2265 cm−1 for
reaction 4; at 2282 cm−1 for reaction 5).
Complete FTIR spectra (from 4000 to 200 cm−1
)
have been registered for all products isolated once the
reaction is gone to completion. The vibrational spec-
trum of co-ordinated PCy3 in all complexes 3 is quite
similar to that of the free molecule, except in the low
frequency region [12], and is not shifted by isotopic 18O
or 13C labelling. It is thus easily recognisable and
described below for the region of interest, below 2000
cm−1, separately from the spectrum of the Ni(CO4)
moieties.
3.1. FTIR spectra of solid Ni(CO4)(PCy3)2 and its 18O
and 13C labelled deri6ati6es
3.2. The 6ibrational spectrum of the (PCy3)2Ni moiety:
bands of the tricyclohexyl phosphine ligands
Solid Ni(CO4)(PCy3)2 (3) and its 18O labelled deriva-
tives (3a–c, Scheme 1) were obtained by the following
coupling reactions:
The deformations l(CH2) of the cyclohexyl groups in
complex 3 are located in the congested region between
1380 and 1100 cm−1 (wagging modes) and are partially
obscured by Nujol absorptions. The CH2 rocking
modes appear at 1050(w), 1004(m), 914(sh) and 894(sh)
cm−1, whereas ring motions are assigned to bands at
(PCy3)2Ni(CO2)+16O2“(PCy3)2Ni(CO4)
(2)
3
(PCy3)2Ni(CO2)+18O2“(PCy3)2Ni(18OOC(O)18O)
887(m), 851(m), 844(sh, m), 820(w) and 789(vw) cm−1
,
3a
(3)
the latter being assigned to the ring breathing. The
vibrational modes of ‘free’ PCy3 are assigned to the
weak bands at 1218, 1160 and 1040 cm−1. In the region
between 800 and 700 cm−1, the stretching modes of
phosphine wa(PC3), w%(PC3) and ws(PC3) are, respectively
observed at 757, 746, 721 and 708 cm−1. In the region
a weak contribution of Nujol at 723 cm−1 is also
present.
(PCy3)2Ni(C18O2)+16O2“(PCy3)2Ni(16O18OC(18O)O)
3b
(4)
(PCy3)2Ni(13CO2)+16O2“(PCy3)2Ni(13CO4)
(5)
3c
FTIR spectrum of the gas in equilibrium with the
solid phase during the reaction, in all cases shows the
At lower wavenumbers, the cyclohexyl ring bending
motions are found. They are the most shifted modes by
co-ordination to the metal. Indeed, fairly intense bands
are found at 564, 546 and 532 for free PCy3, which are
shifted to lower wavenumbers in co-ordinated PCy3, at
519(ms), 515(sh), 510(sh) and 448(m) cm−1. The defor-
mations l(PC3) are expected below 400 cm−1
.
3.3. The 6ibrational spectrum of the Ni(CO2) and
Ni(CO4) moieties
Apart from the Ni(PCy3)2 moiety absorptions, that
are not too sensitive to the conversion of 1 and to the
labelling experiments, there is a set of bands due to the
Ni(CO4) moiety that are shifted by isotopic labelling
experiments. Making a comparative analysis of the
spectrum of compound 3 with those of compounds
obtained by reacting
1
with 18O2 (3a, 3a%),
16O2
(3b, 3b%), and
(PCy3)2Ni(C18O2)
with
(PCy3)2Ni(13CO2) with O2 (3c, Scheme 1), it is possible
to make the total assignment of the bands due to the
Ni(CO4) moiety.
The strong absorption at 1685 cm−1 (Fig. 1) found
in the spectrum of the Ni(12C16O4) species shifted to
Scheme 1.

M. Aresta et al. / Inorganica Chimica Acta 330 (2002) 63–71
67
Fig. 1. FTIR Spectrum of (PCy3)2Ni(12C16O4) (continuous line) and (PCy3)2Ni(16O18OC(18O)16O) (dotted line).
1655 cm−1 by C18O2 labelling, is assigned to the car-
bonyl stretching w(CꢁO).
3.4. Structure of (PCy3)2Ni(O2CO2): a normal mode
analysis
The medium bands at 1258 and 1250 cm−1 have an
average isotopic effect of 20 cm−1 by C18O2 labelling
and are therefore associated to one of the CO stretching
modes. This assignment is also supported by 13CO2 and
18O2 experiments. The stretching vibration w(OO) is
expected in the 900–1000 cm−1 region: we observe a
medium absorption at 995(m, sh) cm−1, as a shoulder
of the medium strong band of PCy3 located at 1005
cm−1, that we attribute to the OꢀO stretching. In fact,
it is correctly shifted to 975 cm−1 by 18O2 labelling and
to 965 cm−1 with C18O2, whereas no shift is observed
for the 13CO2 labelled complex. Isotopic effects at lower
wavenumbers are not easy to detect, except for the
band at 610 cm−1 that is, respectively shifted to 587
and 600 cm−1 by 18O2 and C18O2 labelling and can be
assigned to a nickelꢀoxygen stretching mode. The band
at 658 cm−1 in the molecule made with 18O2 may arise
from the shift of another CO or NiO stretching mode,
hidden in the non-labelled molecule by the strong phos-
phine absorptions. No other bands are observed
below 500 cm−1, except for the molecule issued from
reaction 4, which exhibits weak bands at 434, 393 and
382 cm−1. So, full assignments of the bands in this
region will be made with the help of normal mode
analysis.
As shown by Scheme 1 and reaction 3, two products
can be formed by reacting 1 with 18O2 depending on the
reaction occurs through the opening of the OꢀO bond
(molecule 3a) or one of the NiꢀO bonds (molecule 3a%).
As well, reaction 4 affords molecule 3b if there is an
opening of the OꢀO bond, or molecule 3b% if one of the
NiꢀO bonds is broken.
So, in order to elucidate the mechanism of the reac-
tion of O2 with the carbon dioxide compound and
assign all the observed frequencies, we have performed
a
normal co-ordinate analysis for complex
(PCy3)2Ni(CO2) (1) and all its derivatives obtainable
from isotopic labelling with 18O2 (molecules 3a and 3a%),
and C18O2 (molecules 3b and 3b%) together with the
13CO2 labelled compound (molecule 3c).
There is no structural data for complex 3, but 31P
experiments (see below) have shown that we can rea-
sonably assume a pseudo tetrahedral environment of
the nickel atom. The structural parameters used (see
Scheme 1 and Table 1) were those of the parent com-
plex (PCy3)2Ni(CO2) [13] for the Ni(PCy3)2 unit and of
the peroxocarbonato rhodium complex RhCl(CO4)-
(PEt2Ph)3 [2] for the peroxocarbonato moiety Ni(CO4).
In the same way, the initial force field was taken from
values calculated by Fouassier and co-workers

68
M. Aresta et al. / Inorganica Chimica Acta 330 (2002) 63–71
Table 1
total number of non-zero interaction force constant
being 7 for 36 calculated frequencies (15 observed). In
the refinement procedure, the adjustment of the isotopic
shifts had priority to that of frequencies themselves,
because the former is more sensitive to the geometry of
the molecule.
Table 1 summarises the final results for the valence
force field of complex 2.
Tables 2 and 3 report the assignment of the vibra-
Structure a and valence force field b for pseudo-tetrahedral
(PCy3)2Ni(CO4)
a
Structure
Force constants
Initial
Final
values b
values b
d(NiO2)=1.822
d(NiO3)=1.965
d(O2O4)=1.470
d(C5O3)=1.325
d(C5O4)=1.301
d(C5O6)=1.220
d(NiP)=2.23
w(NiO2)
w(NiO3)
w(O2O4)
w(C5O3)
w(C5O4)
w(C5O6)
w(NiP)
1.95
1.95
3.9
3.21
4.83
9.41
1.6
2.1
2.2
4.2
2.92
4.65
9.55
2.2
tional modes.
Calculated frequencies w(CꢁO), w(CO) and n(OO) in
Table 2 shows that molecules 3b and 3b% are easily
recognisable when considering the isotopic shifts ob-
tained on the w(OO) mode. Indeed, molecule 3b%, ob-
tained by opening of one NiꢀO bond in reaction 4, has
almost no frequency shift on the OꢀO stretching,
whereas molecule 3b, where one oxygen atom of the
peroxo bond is isotopically labelled, presents a shift of
27 cm−1. Therefore, comparison of observed and calcu-
lated values unambiguously shows that reaction 4 leads
to molecule 3b.
d(PC)=1.83
ws(PC3)
w%s(PC3)
wa(PC3)
l(O2NiO3)
l(NiO2O4)
l(O2O4C5)
l(O4C5O3)
l(C5O3Ni)
2.83
2.47
2.47
0.25
0.35
0.35
0.46
0.45
2.97
2.60
2.53
0.25
0.35
0.35
0.46
0.85
h(O2NiO3)=78
h(NiO2O4)=120.3
h(O4C5O3)=118.1
h(C5O3Ni)=115.8
h(O3C5O6)=124.9
h(O4C5O6)=117.0
l(C5O6)
k(C5O6)
l(PNiP)
l(O2NiP)
l(O3NiP)
0.79
0.66
0.34
0.30
0.30
0.95
0.66
0.34
0.30
0.30
This result is confirmed by the isotopic shift obtained
on the stretching mode w(CO4) located around 1250
cm−1 in the non-labelled compound.
h(PNiP)=123
h(O2NiP)=79.5
h(O3NiP)=79.5
h(CPC)=120
Furthermore, considering reaction 3, the opening of
one NiꢀO bond affords the maximum shift on the
w(OO) frequency (54 cm−1, molecule 3a%), whereas
opening of the OꢀO bond leads to a ‘half-labelled’
16Oꢀ18O peroxo bond (molecule 3a) with a frequency
shift of 26 cm−1. Again comparison of experimental
and calculated values permits us to conclude that reac-
tion 3 proceeds through the opening of the OꢀO bond
and affords molecule 3a. This result is also supported
by the stretching mode w(CO4), for which the isotopic
shift must be more important (16 cm−1) in molecule 3a%
than in molecule 3a (1 cm−1). Indeed, the experimental
value is 1 cm−1. When (PCy3)2Ni(C18O2) and 16O2 are
used, the experimental value (15 cm−1) is not very
informative for the choice between 3b (24 cm−1) and
3b% (9 cm−1).
ls(PC3)
l%s(PC3)
la(PC3)
r//(PC3)
r(PC3)
~(PNi)
~(NiO2)
~(NiO3)
~(O2O4)
~(C5O4), ~
(C5O3)
f(NiP, NiP)
f(NiO3, C5O3)
f[(C5O4, l(C5O6)]
f[(C5O3, l(C5O6)]
f(C5O4, C5O3)
f[NiP, ls(PC3)]
f(NiO2, NiO3)
0.57
0.64
0.64
0.46
0.57
0.64
0.64
0.46
0.46
0.0
0.0
0.0
0.0
0.0
0.46
0.0005
0.0005
0.0005
0.0005
0.0
0.0
0.0
−0.8
−0.8
0.57
0.1
0.4
−0.8
−0.8
0.57
−0.15
0.1
−0.15
0.2
If w(CꢁO), w(CO4) and w(OO) modes are relatively
pure, located at usual wavenumbers (1685, 1254 and
995 cm−1, respectively) and easily recognisable, normal
co-ordinate analysis is necessary to assign the other
vibrations. Table 3 summarises the results for molecules
3, 3a, 3b and 3c (observed and calculated values and
assignment). The good agreement between observed
and calculated wavenumbers and isotopic shift (in-
cluded 12/13C shifts) allows us to propose a relevant
force field for this compound. It is generally assumed
that interaction force constant must be lower than
one-tenth of the main corresponding force constants.
The values calculated can then be considered as
accurate.
a
,
Bond lengths in A, angles in °.
Force constants are given in mdyn A for bonds, mdyn A rad
b
−1
−2
,
,
for angles and mdyn rad−1 for bond angle interactions.
for the parent nickel compound [13,14], together with
our previous results obtained for the peroxocarbonato
rhodium complex [2]. The program used to calculate
the generalised valence force field was a modified ver-
sion of that used by Schachtschneider [15].
Ni(PCy3)2(CO4) is a Cs fragment with 14 atoms: 36
vibration modes are then expected (23A%+13A¦), in-
cluding seven torsions. Because of the limited vibra-
tional data in the low-frequency region, the fitting was
limited to 300 cm−1, but all modes were calculated, the
If the final force constant values calculated for both
NiO bonds are almost identical (2.1 or 2.2 mdyn A−1),

M. Aresta et al. / Inorganica Chimica Acta 330 (2002) 63–71
69
Table 2
motion k(CO) is found at a higher frequency (780
cm−1) than in CO2 complexes [12], where this mode is
observed around 550 cm−1, and higher than the corre-
sponding in-plane bending l(OCO). The latter is, in
fact, highly mixed with several other modes (w(CO),
w(NiO), w(NiP) and bending motions of the cycle. This
explains why small isotopic shifts are observed for NiP
stretching modes in 18O labelled species.
Calculated IR wavenumbers and isotopic shifts (in cm−1
OOC(O)O fragment and its 18O and 13C labelled derivatives
) for
MODE
3
3a
−2
3a%
−3
3b
3b%
3c
w(CO)
w(CO4)
w(OO)
w(CO3)
1681
1248
994
−28
−24
−27
−8
−26
−9
−45
−32
0
−1
−26
−31
−16
−54
0
−1
−39
689
−2
Therefore, reminding that CO2 is evolved during the
reaction, our FTIR study lets us to conclude that the
formation of the peroxocarbonato metallacycle from
the reaction Ni(CO2)+O2 proceeds via deco-ordination
of CO2, fixation of O2 in a side-on way on the nickel
fragment, then opening of the OꢀO bond of the Ni(O2)
compounds to allow insertion of CO2. If the deco-ordi-
nation of CO2 is not taken into account, the formation
of the mixed Oꢀ*O peroxo group is hardly explained.
See molecules 3, 3a, 3a%, 3b, 3b% and 3c in Scheme 1.
this is not the case for the CO bonds, for which the
value obtained for the bond relevant to the CꢁO group
is 40% higher than that for the CꢀOꢀM (4.65 vs. 2.92
mdyn A−1). Values calculated for the OO force con-
stant is slightly higher in the nickel compound (4.2
mdyn A−1) than in the rhodium complex (3.9 mdyn
A
−1). Seven interaction force constants were necessary
to fit the calculation of normal modes. Among them,
interaction between NiP stretching and PC3 symmetric
deformation, CO stretching and in-plane OCO bending
are usual values. Interactions between both NiO (0.2
mdyn rad−1) and CO (0.57 mdyn rad−1) stretching
modes are due to the cyclic structure of the compound.
In fact, the high frequency w(CO4) mode at 1258,
1250 cm−1 can be considered as the asymmetric
stretching motion of the O3O4C moiety, whereas the
symmetric counterpart (not observed in molecule 3, but
located at 658 cm−1 in molecule 3a) is lowered because
of its mixing with the NiO3 stretching mode. As already
observed in similar compounds [2], the out-of-plane
3.5. Characterisation and reacti6ity of peroxo-
carbonateꢀnickel complexes in the solid state and in
solution
Complex 3 is stable in the solid state at temperature
below 273 K. If the temperature is raised to 293 K, a
slow decomposition takes place with formation of
phosphine oxide as confirmed by the IR spectrum of
the sample in Nujol mull (band at 1160 cm−1). At the
same time the band of the peroxocarbonate located at
1685 cm−1 disappears and new bands centered around
1620 cm−1 appear due to the formation of the Ni-car-
bonate moiety, L%LNiꢀCO3.
Table 3
Assignment of IR observed frequencies (cm−1) of (PCy3)2Ni(CO4) and its 18O and 13C labelled derivatives with calculated values for molecules
3, 3a, 3b and 3c
Wavenumbers (cm−1
)
Observed 3 Calculated 3 Assignment
Observed 3a Calculated 3a Observed 3b Calculated 3b Observed 3c Calculated 3c
1685(vs)
1258(m)
1250(sh)
995(sh, s)
780(m)
761(m)
755(s)
1681
1248
w(CꢁO)
w(CO4)
1679
1249
1680
1247
1655
1234
1653
1234
1640
1230
1218
995
1636
1218
994
785
760
758
748
747
689
710
708
610
555
437
404
387
w(O2O4)
w(CꢁO)
w%s(PC3)
w%s(PC3)
wa(PC3)
wa(PC3)
w(CO3)+w(NiO3)
ws(PC3)
ws(PC3)
w(NiO3)+w(CO3)
w(NiO2)
975
779
967
782
965
774
966
779
994
760
n.o. a
745(s)
732(s)
n.o. a
658
661
n.o.
683
n.o.
687
709(sh, w)
610(m)
n.o.
n.o.
n.o.
n.o.
587
n.o.
434
393
382
588
538
435
400
381
600
n.o.
n.o.
n.o.
n.o.
604
553
434
400
379
610
n.o.
n.o.
n.o.
n.o.
609
555
436
403
385
w(NiP)
w(NiP)
l(CꢁO)
n.o., not observed.
a Obscured by w(PCy3).

70
M. Aresta et al. / Inorganica Chimica Acta 330 (2002) 63–71
The (PCy3)2Ni(CO4) 13C NMR in CH2Cl2 (acetone-
carbonate and phosphine oxide (see Section 1). Neither
the gas phase in equilibrium with the solid, nor the
solid itself, contains any form of oxidised ethylene, as
demonstrated also by the GC–MS. Ethylene, thus,
interacts with the nickel centre and favours the release
of the phosphine that is oxidised, while the olefin itself
is not oxidised under the reaction conditions, that are
very mild.
Dienes are much more effective agents for the release
of the phosphine. In fact, allene is able to produce the
same effect as ethylene, but in a much shorter time. At
room temperature the phosphine oxide appears within
2 min, at 270 K within 10 min.
If styrene is used in solution, oxidation products are
found, like styrene oxide, methyl–phenyl ketone,
phenylacetaldehyde, benzaldehyde. The latter derives
from a two-oxygen transfer to the olefin, while the
former products are formed via a one-oxygen transfer
to the olefin. These findings are similar to those we
have reported for the Rh–peroxocarbonates–styrene
system and confirm the role of the peroxocarbonate
species in the oxidation of olefins. Transition metal
peroxocarbonate complexes may behave, thus, as mimic
systems of mono-oxygenases [17]. One oxygen atom of
the dioxygen molecule is transferred to the olefin, the
second to CO2, which is converted into carbonate, a
form that does not prevent the metal to enter a catalytic
cycle. In fact, we have demonstrated that transition
metal systems can promote the deoxygenation of car-
bonate moiety to CO2 [18], with oxygen transfer to an
external oxophile, regenerating, thus, the reduced form
of the metal that is able to co-ordinate again carbon
dioxide and dioxygen and start a new cycle [5,18].
d6 as internal reference) shows an absorption at 166.6
ppm at 213 K. The parent compound 1 shows (in
toluene-d8, 298 K) a resonance at 156 ppm [15]. To the
best of our knowledge, this is the first report about a
13C spectrum of two relative metalꢀCO2 and
ꢀperoxocarbonate complexes. The 31P spectrum of 3
shows a single peak at 43.2 ppm at 213 K. Lowering
the temperature to 188 K the spectrum does not
change. From these data it is possible to infer a tetrahe-
dral structure for the Ni(II)ꢀperoxocarbonate in solu-
tion as a square-planar structure would give rise to two
31P signals of the non-equivalent P-trans to the CꢀO or
CꢀOO bonds, respectively, as it occurs in 1. In fact, the
parent complex 1 shows a single 31P signal at 36.2 ppm
in solution (toluene-d8, 298 K), due to its fluxional
behaviour [16]. Conversely, two 31P NMR resonances
are found for 1 both in the solid state (48.9 and 20.9
ppm) and in solution at 188 K (51.7 and 20.2 ppm),
showing that at this temperature the complex has a
rigid conformation [16].
These data show that the CO2-complex and its perox-
ocarbonate derivative have different structures in solu-
tion, 1 being a distorted square-planar, fluxional
molecule, while 3 is tetrahedral. Interestingly, as theo-
retical calculation [12] and experimental data [2,16]
have shown, the tetrahedral structure is not likely to be
the second limit form of 1 in solution. In fact, the
fluxional mode concerns the deformation (elongation)
of the NiꢀC bond [2].
In CD2Cl2 or other solvents, the oxygen-transfer
from the peroxocarbonate moiety is much faster, as the
solvent assists the dissociation of the ligand and favours
the external attack to the Ni-bonded O-atom of the
peroxo-group to afford the phosphine oxide. Again
Ni-carbonate is formed. This is demonstrated by the
fact that when a solution of 3 in CD2Cl2 is heated to
294 K, the 31P NMR spectrum changes dramatically in
a few minutes, and peaks at 10.86 (s, weak, free phos-
phine), 31.21 (s, medium, co-ordinated carbonate) and
49.8 ppm (s, weak, free phosphine oxide) appear (see
Section 2).
Acknowledgements
The Italian authors thank the Ministry of Research,
MURST
Project
PIN
9803026360-98
and
MM03027791, and the University of Bari for financial
support. All authors acknowledge the COST financial
support, Project COST D 09-012-1998.
These data are identical with those found in the 31P
spectrum of 3, after the solid sample was kept at room
temperature for 4 h and the NMR spectrum registered
at 213 K.
That the oxygen transfer from 3 is an inter-molecular
more than an intra-molecular process, is demonstrated
not only by the higher stability of solid 3, with respect
to its solution, at room temperature, but also by the
fact that if another gaseous ligand is added to solid 3,
the phosphine is released and easily oxidised. In fact,
when solid 3 is exposed to one atmosphere of ethylene
at 270 K, the original pink solid converts into a green-
ish compound which IR spectrum shows the signals of
References
[1] (a) P.J. Hayward, D.M. Blake, C.J. Nyman, G. Wilkinson,
Chem. Commun. (1969) 987;
(b) P.J. Hayward, D.M. Blake, G. Wilkinson, C.J. Nyman, J.
Am. Chem. Soc. 92 (1970) 5873;
(c) L. Malatesta, R. Ugo, J. Chem. Soc. (1963) 2080;
(d) M. Aresta, E. Quaranta, A. Ciccarese, C1 Mol. Chem. 1
(1985) 267.
[2] M. Aresta, I. Tommasi, E. Quaranta, C. Fragale, J. Mascetti, M.
Tranquille, F. Galan, M. Fouassier, Inorg. Chem. 35 (1996)
4254.
[3] J.C. Dobrowolski, M.H. Jamroz, J.K. Kazimirski, K. Bajdor,
M.A. Borowiak, R. Larsson, J. Mol. Struct. (1999) 483.

M. Aresta et al. / Inorganica Chimica Acta 330 (2002) 63–71
71
[4] M. Borowiak, M.G.H. Jamro´z, J.C. Dobrowolski, K. Bajdor,
J.K. Kazimirski, J. Mascetti, E. Quaranta, I. Tommasi, M.
Aresta, J. Mol. Catalysis, Sect. A 165 (2001) 45.
[5] M. Aresta, I. Tommasi, A. Dibenedetto, Energy Fuels 15 (2001)
269.
[6] M. Aresta, R. Larsson, M. Borowiak, J. Mascetti, COST-D3/
0007/94 Project, Activity Report, 1997–98, p. 79.
[7] M. Aresta, C.F. Nobile, J. Chem. Soc., Dalton Trans. (1977)
708.
[8] M. Aresta, E. Quaranta, I. Tommasi, 215th ACS National
Meeting, Marh 29–April 2, Dallas, TX, 1998, coll 086.
[9] M. Aresta, C.F. Nobile, V.G. Albano, E. Forni, M. Manassero,
J. Chem. Soc., Chem. Commun. (1975) 636.
(PCy3)2Ni(O2). This signal is not found in the spectrum of the
solid reaction mixture after approximately 8 h of reactions.
[12] C. Jegat, M. Fouassier, M. Tranquille, J. Mascetti, I. Tommasi,
M. Aresta, F. Ingold, A. Dedieu, Inorg. Chem. 32 (1993) 1279.
[13] C. Jegat, M. Fouassier, J. Mascetti, Inorg. Chem. 30 (1991)
1521.
[14] C. Jegat, M. Fouassier, M. Tranquille, J. Mascetti, Inorg. Chem.
30 (1991) 1529.
[15] J.H. Schachtschneider, R.G. Snyder, Spectrochim. Acta 19
(1963) 117.
[16] M. Aresta, R. Gobetto, E. Quaranta, I. Tommasi, Inorg. Chem.
31 (1992) 4286.
[17] (a) M. Aresta, E. Quaranta, A. Ciccarese, J. Mol. Catal. 41
(1987) 355;
(b) M. Aresta, C. Fragale, E. Quaranta, I. Tommasi, J. Chem.
Soc., Chem. Commun. (1992) 315 and references therein.
[18] M. Aresta, A. Dibenedetto, I. Tommasi, I. Eur. J. Inorg. Chem.
(2001) 1801.
[10] G. Wilke, H. Schott, P. Heimbach, Angew. Chem. Int. Ed. Engl.
6 (1967) 92.
[11] The 31P NMR spectrum at 203 K in CD2Cl2 of a solution of 1
with gaseous O2 shows (in addition of the signals of 1 and 3) the
appearance of an absorption at 53.2 ppm due to complex