Monomeric alkoxide and alkylcarbonate complexes of nickel and palladium stabilized with the iPrPCP pincer ligand: A model for the catalytic carboxylation of alcohols to alkyl carbonates

Article abstract of DOI:10.1039/c8dt04919j

Monomeric alkoxo complexes of the type [(^iPrPCP)M-OR] (M = Ni or Pd; R = Me, Et, CH₂CH₂OH; ^iPrPCP = 2,6-bis(diisopropylphosphino)phenyl) react rapidly with CO₂ to afford the corresponding alkylcarbonates [(^iPrPCP)M-OCOOR]. We have investigated the reactions of these compounds as models for key steps of catalytic synthesis of organic carbonates from alcohols and CO₂. The MOCO-OR linkage is kinetically labile, and readily exchanges the OR group with water or other alcohols (R′OH), to afford equilibrium mixtures containing ROH and [(^iPrPCP)M-OCOOH] (bicarbonate) or [(^iPrPCP)M-OCOOR′], respectively. However, [(^iPrPCP)M-OCOOR] complexes are thermally stable and remain indefinitely stable in solution when these are kept in sealed vessels. The constants for the exchange equilibria have been interpreted, showing that CO₂ insertion into M-O bonds is thermodynamically more favorable for M-OR than for M-OH. Alkylcarbonate complexes [(^iPrPCP)M-OCOOR] fail to undergo nucleophilic attack by ROH to yield organic carbonates ROCOOR, either intermolecularly (using neat ROH solvent) or in intramolecular fashion (e.g., [(^iPrPCP)M-OCOOCH₂CH₂OH]). In contrast, [(^iPrPCP)M-OCOOMe] complexes react with a variety of electrophilic methylating reagents (MeX) to afford dimethylcarbonate and [(^iPrPCP)M-X]. The reaction rates increase in the order X = OTs < IMe ? OTf and Ni < Pd. These findings suggest that a suitable catalyst design should combine basic and electrophilic alcohol activation sites in order to perform alkyl carbonate syntheses via direct alcohol carboxylation.

Full text of DOI:10.1039/c8dt04919j

Dalton
Transactions
View Article Online
View Journal
PAPER
Monomeric alkoxide and alkylcarbonate
complexes of nickel and palladium stabilized with
the iPrPCP pincer ligand: a model for the catalytic
carboxylation of alcohols to alkyl carbonates†
Cite this: DOI: 10.1039/c8dt04919j
Luis M. Martínez-Prieto,
Pilar Palma
and Juan Cámpora
*
Monomeric alkoxo complexes of the type [(iPrPCP)M-OR] (M = Ni or Pd; R = Me, Et, CH2CH2OH; iPrPCP =
2,6-bis(diisopropylphosphino)phenyl) react rapidly with CO2 to afford the corresponding alkylcarbonates
[(iPrPCP)M-OCOOR]. We have investigated the reactions of these compounds as models for key steps of
catalytic synthesis of organic carbonates from alcohols and CO2. The MOCO-OR linkage is kinetically
labile, and readily exchanges the OR group with water or other alcohols (R’OH), to afford equilibrium mix-
tures containing ROH and [(iPrPCP)M-OCOOH] (bicarbonate) or [(iPrPCP)M-OCOOR’], respectively.
However, [(iPrPCP)M-OCOOR] complexes are thermally stable and remain indefinitely stable in solution
when these are kept in sealed vessels. The constants for the exchange equilibria have been interpreted,
showing that CO2 insertion into M-O bonds is thermodynamically more favorable for M-OR than for
M-OH. Alkylcarbonate complexes [(iPrPCP)M-OCOOR] fail to undergo nucleophilic attack by ROH to yield
organic carbonates ROCOOR, either intermolecularly (using neat ROH solvent) or in intramolecular
fashion (e.g., [(iPrPCP)M-OCOOCH2CH2OH]). In contrast, [(iPrPCP)M-OCOOMe] complexes react with a
variety of electrophilic methylating reagents (MeX) to afford dimethylcarbonate and [(iPrPCP)M-X]. The
reaction rates increase in the order X = OTs < IMe ≪ OTf and Ni < Pd. These findings suggest that a suit-
able catalyst design should combine basic and electrophilic alcohol activation sites in order to perform
alkyl carbonate syntheses via direct alcohol carboxylation.
Received 14th December 2018,
Accepted 16th December 2018
DOI: 10.1039/c8dt04919j
rsc.li/dalton
alcohols with CO2 to yield alkyl carbonate and water is only
weakly exothermic, and endergonic under the ambient con-
Introduction
There is currently a strong interest in new methods for the pro- ditions (e.g., for methanol ΔH° ≈ −4 kcal mol−1 and ΔG° ≈
duction of organic carbonates from carbon dioxide, avoiding +6 kcal mol−1 at 1 atm and 298 K) largely due to the unfavor-
the use of phosgene or its derivatives as starting materials.1 able entropy contribution of gaseous CO2.4,6 The equilibrium
The best developed of such processes involves the reaction of constant rapidly decreases with temperature in such a way that
epoxides with carbon dioxide.2 This is one of the most promis- heating to enhance the reaction rate severely limits the
ing routes for the incorporation of CO2 in downstream chemi- maximum theoretical yield, unless the entropy effect is com-
cals and polymers, but epoxides are highly reactive and toxic pensated by rising pressure. However, calculations based on
intermediates, and they are obtained mostly from non-renew- standard thermodynamic values have shown that, even at mod-
able feedstocks.3 Direct carboxylation of alcohols with CO2 erately high temperatures, the synthesis of dimethyl carbonate
(Scheme 1, top) is an attractive alternative that leads to organic from CO2 and methanol only becomes exergonic under
carbonates directly from alcohols, and water as the only by- enormous pressures.6b Therefore, apart from obvious strat-
product.4 This is a clean and atom-efficient process that fully egies, such as using high CO2 pressure (ideally, supercritical
complies with the main requisites of green and sustainable CO2 as solvent) or water removal from the reaction mixture
chemistry.5 Unfortunately, direct combination of aliphatic (either by physical or chemical methods), the use of efficient
catalysts would be essential to allow carbonate formation at
the lowest possible temperatures to shift the equilibrium to
Instituto de Investigaciones Químicas. CSIC-Universidad de Sevilla,
the right. Various catalysts have been investigated, including
C/Américo Vespucio, 49, 41092 Sevilla, Spain. E-mail: campora@iiq.csic.es
both heterogeneous7 and homogeneous.8–12 Among the latter,
†Electronic supplementary information (ESI) available: Further computational
soluble metal alkoxides,10–12 in particular diorganotin(IV)
details (energies, molecular drawings and atomic coordinates for optimized
dimethoxides (R′2Sn(OMe)2)11 and niobium(V) methoxide,12
structures). See DOI: 10.1039/c8dt04919j
This journal is © The Royal Society of Chemistry 2019
Dalton Trans.

View Article Online
Paper
Dalton Transactions
Scheme 1
are among the most efficient catalysts for the synthesis of
Our group has reported the syntheses of some of the first
dimethyl carbonate from methanol. The mechanisms associ- examples of hydroxide and methoxide complexes of Ni and Pd
ated to such catalysts were studied in detail by the groups of stabilized with iPrPCP pincer ligands (iPrPCP = 2,6-bis(diiso-
Ballivet-Tkatchenko,11,13 Aresta6a,12b,14 and others.4,11 In both propylphosphino)phenyl).19a,b We described the reactions of
cases, the carboxylation process begins with the CO2 insertion monomeric hydroxo complexes [M(iPrPCP)OH] to afford mono-
into M–O bonds to afford species containing a methyl- nuclear bicarbonate derivatives [M(iPrPCP)OCOOH], and their
carbonate (or hemicarbonate) linkage, [M]–OCOOMe. partial decarboxylation to afford thermally stable binuclear
However, the processes leading to the release of dimethyl car- carbonates [{M(iPrPCP)}2(μ2:κ2-O,O′-CO3)].19d Next, we studied
bonate and regeneration of the reactive alkoxo linkage are less β-hydrogen elimination reactions in the Ni and Pd methoxides
well understood. Experimental and computational research on [M(iPrPCP)OMe], which initially lead to the corresponding
the niobium system suggest that the latter step involves methyl hydrides [M(iPrPCP)H], and, subsequently, to different M(0)
transfer from methanol to the coordinated methoxycarbonyl species, depending on M.20 While the Pd methoxide decom-
ligand. As shown in Scheme 1, the acidic metal centre activates poses readily in alcohol solvents, the nickel counterpart is con-
two molecules of methanol, which cooperate to propitiate the siderably more stable, particularly when it is dissolved in
electrophilic methylation of the negatively polarized methyl- methanol. Recently, we investigated the hydrolytic stability of a
carbonate oxygen.12a
series of nickel and palladium alkoxides showing that Ni com-
Mechanistic studies on the mentioned tin and niobium alk- plexes are more prone to be hydrolyzed than Pd ones.19e
oxide catalysts are complicated because multiple species of Building on these results, we now report our investigation of
different nuclearity coexist in their systems. In addition, the such Ni and Pd monomeric alkoxides as models for funda-
efficiency of such systems is decreased by their low tolerance to mental reactions of metal alkoxides with CO2, the thermo-
water,6a,11b,d with which they react irreversibly to afford in- dynamic relationships between alkylcarbonate and bicarbon-
soluble materials devoid of catalytic activity. Well-defined ate species, and some other processes that could potentially
organometallic alkoxide complexes of transition metals provide lead to the formation of organic carbonates.
more accessible systems, better suited for detailed studies of
the basic processes involved in catalytic alcohol carboxylation.
For example, Bergman15 and Darensbourg16 used 18e, coordina-
tively saturated carbonylrhenium(I) and carbonylmanganese(I)
methoxide complexes to ascertain the fundamental mechanistic
features of CO2 insertion into M–O bonds. These studies
Results and discussion
Reactions of Ni and Pd alkoxide complexes with CO2 and
PhNCO
showed that CO2 does not bind to the metal prior to insert, but As mentioned in the introduction, we have shown that hydroxo
attacks directly on the alkoxide oxygen atom, followed by a fast pincer complexes [(iPrPCP)M-OH] (M = Ni, 1a or Pd, 1b) react
ligand rearrangement to the alkylcarbonate product.
with CO2, quantitatively affording the terminal bicarbonate
In recent years, pincer ligands have become a widely used complexes [(iPrPCP)M-OCOOH] (M = Ni, 2a; Pd, 2b).19d Their
platform for the study of the fundamental reactivity of tran- tendency to undergo partial decarboxylation to the corres-
sition metals.17 Strong chelating scaffolds such as PCP ligands ponding binuclear carbonates [{(iPrPCP)M}2(μ2:κ2-O,O′-CO3)]
can efficiently stabilize the otherwise labile hydroxo- or alkoxo (3a or 3b, respectively) complicated their isolation. However,
species, focusing their reactivity on the M–O bond. Thus, they taking advantage of their low solubility in hexane, both bicar-
provide a unique opportunity to explore the details of CO2 bonates were crystallized and their X-ray diffraction structures
insertion and other elemental processes relevant to alcohol determined. A similar behaviour has been recently reported by
carboxylation.18,19 In addition, the stability of the metal-pincer Wendt in other Ni-pincer systems.18f,g As shown in Scheme 2,
fragment is expected to radically improve tolerance to water, alkoxide derivatives [(iPrPCP)M-OR] (M = Ni or Pd; 4a/b and
avoiding the coalescence of hydroxo species into insoluble 5–8) exhibit very similar reactivity: when their solutions are
solids.
exposed to CO2 they also react rapid and quantitatively
Dalton Trans.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Dalton Transactions
Paper
Scheme 2
affording the corresponding alkylcarbonates 9a, 9b and 10–13.
No colour change is observed for the colourless Pd derivative
4b, but the reddish-orange solutions of the Ni alkoxides
rapidly change to yellow. In contrast with the analogous bicar-
bonates, alkylcarbonate complexes do not precipitate when the
reaction is carried out in hexane, in which they are very
soluble. However, like the bicarbonates, the alkylcarbonates
undergo partial conversion to the corresponding carbonates,
3a or 3b, when their solutions are partially concentrated under
vacuum. Due to their lower solubility, these binuclear carbon-
ates crystallize preferentially (vide infra). So far, this behaviour
has prevented us from isolating pure samples of 9–13, but
NMR characterization was readily accomplished using samples
generated in situ by bubbling a small amount of dry CO2
through solutions of the purified alkoxide precursors in C6D6.
These reactions are very clean; the only other species detected
were negligible amounts of the corresponding bicarbonate, 2a
or 2b, arising from water traces. In every case investigated, CO2
insertion into the M–O bonds was fast and quantitative, as
Fig. 1 1H NMR (300 MHz, C6D6, 25 °C) spectrum of compound 13. The
expansion shows the glycoxide CH2–CH2 region, and a simulation of the
sub-spectrum using the coupling constants indicated in the inset. The
*marks a small amount of water.
evinced by the clean replacement of the 31P{1H} signal of alk- the metal-bound (ipso) carbon atom of the pincer ligands of
oxides by that of the corresponding alkylcarbonate, usually alkylcarbonates (M = Ni, 152–153 ppm; Pd, 154.5 ppm) are
shifted some 3 ppm downfield from the starting complexes.
almost in the same position for alkylcarbonates and bicarbon-
The characteristic low-field 13C –OC(O)OR carbonyl reso- ates, indicating that the influence of the R group on the
nance of alkylcarbonate ligands appears within a narrow σ-donor capacity of the –OCOOR (either alkyl or hydrogen)
region between 157 and 159 ppm. This is shifted by ca. 5 ppm ligand is very small, as expected for partially ionic M–O
upfield from those of the corresponding bicarbonates bonds.19e
(∼162 ppm). Moreover, CO2 insertion causes moderate
In addition to CO2 insertions, we briefly studied the reac-
changes to the 1H and 13C signals of the OR moiety. The 1H tion of the nickel alkoxide 4a with phenylisocyanate
spectrum of 13 (Fig. 1) provides a remarkable example. This (Scheme 3). 31P{1H} NMR monitoring showed that this reaction
shows two multiplets at δ 4.05 and 3.69, assigned to the CH2 also proceeds rapidly to afford a single insertion product, 14,
groups of the glycoxide fragment. The shape of these multi- characterized by a new resonance at δ 52.6. A second, low-
plets indicates that the OCH2–CH2O bond does not rotate intensity 31P signal was also observed at δ 56.5. This was attrib-
freely, presumably due to the formation of a rigid structure by uted to a secondary product arising from the reaction of
intramolecular H-bonding, as previously noted for the parent PhNCO with the small amount of hydroxide 1a that is usually
glycoxide, 8.19e Yet, in contrast with the broad CH2 signals of present in solutions of 4a. The identity of this species (14′) was
8, those of 13 are sharp and well-defined signals, suggesting confirmed by conducting the reaction of 1a with PhNCO separ-
1
that the H bond interaction is stronger in the latter. As ately. Two features of the H NMR spectrum of 14 suggest that
shown in Fig. 1, the shape of the signals can be simulated by this product is the N-bound (κ1-N) product arising from CvN
assuming a rigid AA′XX′ spin system characteristic of a stable insertion, rather than its O-bound isomer (κ1-O). First, the
cyclic system. Notwithstanding, the signals corresponding to ortho phenyl protons experience a strong deshielding (δ 8.79)
This journal is © The Royal Society of Chemistry 2019
Dalton Trans.

View Article Online
Paper
Dalton Transactions
any other organic products were formed in amounts large
1
enough to allow their identification in the final H spectrum
of the sample. The only identifiable signals in the final 31P
{1H} spectrum were those of the tetranuclear carbonyl
[Ni4(CO)5(iPrPCHP)2] and free iPrPCHP ligand, which had been
recognized in our previous studies as the signature of the irre-
versible decomposition of methoxide 4a. This process involves
a complex sequence of steps that ultimately cause full dehydro-
genation of the OMe ligand down to CO and the collapse of
the [(iPrPCP)Ni] unit.20b From this we deduce that the thermal
instability of 13 relates to the properties of the alkoxide (result-
ing from CO2 de-insertion), rather than to the sought organic
carbonate elimination.
Scheme 3
The above-discussed experiments lead us to conclude that
alkylcarbonate complexes are thermally stable in solution, at
least at the ambient temperature. Their facile transformation
into the corresponding binuclear carbonates does only take
place when their solutions are allowed to exchange volatile
components (e.g., under reduced pressure). In order to
monitor these transformations under controlled conditions,
we heated NMR samples of the alkylcarbonate complexes in
toluene-d8 at 100 °C, kept in a closed thermostatic vessel
purged with a slow N2 stream, designed to remove CO2 or
other gases while minimizing exposure to air or moisture (see
Experimental section). The samples were periodically removed
to measure their NMR spectra. Under these conditions, nickel
methylcarbonate 9a was quantitatively transformed into 3a
within 15 h, but the organic products were also lost. The
heavier n-butylcarbonate derivative 11 would give more chance
to detect less volatile organic products. When 11 was subjected
to the same procedure, the transformation took 48 h to com-
that points to their location near to the axis of the square
planar Ni centre. Second, the N-phenyl and the pincer isopro-
pyl groups give broad resonances. This is consistent with the
restricted rotation of the bulky κ1-N carbamate ligand with
regard to the Ni-pincer fragment, but not with its relatively
unhindered κ1-O isomer. These signals become sharp, well
resolved lines when the spectrum is recorded at 55 °C. The
same selectivity has been reported for the reactions of isocya-
nates with different late transition-metal alkoxide com-
plexes.15,21 In contrast with the alkylcarbonate complexes
arising from CO2 insertion, both 14 and 14′ are stable and can
be isolated as pure microcrystalline solids without decompo-
sition. Hence, PhNCO insertion into the Ni–O bonds of 4a and
1a can be regarded irreversible processes, at least under the
usual experimental conditions.
Kinetic vs. thermodynamic stability of alkylcarbonate
complexes
1
plete. The only organic product detected in the H spectra was
As discussed before, Ni and Pd alkylcarbonate complexes a small amount of n-butanol, that disappeared once 11 was
[(iPrPCP)M-OCOOR] are prone to evolve into the corresponding fully consumed. In a separate test, we checked that, despite its
binuclear carbonate 3a or 3b, but ascertaining the final fate of high boiling point (118 °C), n-butanol gradually leaves a
the alkoxo (OR) groups was not trivial, because the transform- toluene-d8 solution under the conditions of the experiment.
ation is usually partial, and they end up in volatile organic pro- Other potential products, like di-n-butylcarbonate or di-n-butyl-
ducts that are not easily detected. By analogy with the de- ether, were not detected. Since these are even less volatile than
carboxylation of inorganic bicarbonate salts to the corres- n-butanol (b. p. 207 and 141 °C, respectively) it can be con-
ponding carbonates, one could speculate that the byproducts cluded that the R group is lost exclusively as ROH. To test this
of the decomposition of alkylcarbonates might be organic car- proposal, we conducted one further experiment, in which a
bonates (RO)2CO or ethers R2O, the counterparts of carbonic solution of 0.025 mmol of 9a in ca. 0.7 mL of C6D6 was evapor-
acid and water, respectively. In order to clarify this point, we ated to dryness, collecting the volatiles in a small trap cooled
heated C6D6 solutions of the nickel alkylcarbonates 9a with liquid nitrogen. NMR analyses showed that the solid
(R = Me), 12 (R = iPr), and 13 (R = CH2CH2OH) at 100 °C in residue was a 5 : 1 mixture of 9a and 3a, while the C6D6 col-
NMR sample tubes sealed with gas-tight PTFE valves, record- lected in the trap only contained methanol. As shown in
1
ing their 31P{1H} or H NMR spectra at regular intervals for up Scheme 4, transformation of alkylcarbonate complexes into
to 12 h. No changes were observed in the first two samples, the corresponding carbonates and free alcohol requires water.
indicating that these compounds are thermally very stable pro- Most likely, this is taken from moisture traces present in the
vided that loss of volatile components is avoided. In contrast,
glycoxide 13 is proved somewhat less stable, as its spectrum
decayed and fully disappeared after heating for four days. This
appeared potentially interesting, because the glycoxide moiety
might be favoring elimination of cyclic ethylene carbonate via
intramolecular nucleophilic cyclization of the 2-hydroxyethyl-
carbonate fragment. However, neither ethylene carbonate nor
Scheme 4
Dalton Trans.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Dalton Transactions
Paper
solvents or glassware. However, the advance of the reaction At the room temperature, the signals exhibit some broadening
requires removal of the alcohol from the system. due to the intermolecular exchange. As expected, these effects
Hydrolysis of alkylcarbonate complexes to afford carbonate become more evident as the temperature is increased, but the
compounds has been seldom mentioned in the literature,22,23 original spectrum is reproduced when the sample is cooled
although it might be fairly common reaction for late transi- again to 25 °C. Addition of a large excess of methanol (10-fold)
tion metal derivatives. In 1992, Bergman and Simpson showed shifts the equilibrium to the right side, removing the reso-
that rhenium alkylcarbonate complexes react with two equiva- nance of the carbonate complex.
lents of water at 45 °C to afford a binuclear carbonate deriva-
From a mechanistic point of view, the reaction of the Ni or
tive.15 However, such experimental conditions suggest that Pd carbonates with methanol can be seen as a protolytic clea-
the latter process was not as facile as the hydrolysis of the Ni vage of one of the M–O bonds of the strongly basic carbonate
and Pd alkylcarbonate derivatives 9–13. Furthermore, in this complex. Initially, this should generate a 1 : 1 mixture of the
group 10 system, the hydrolysis is reversible and alkylcarbo- corresponding methoxide 4 and bicarbonate 2, that sub-
nates are fully restored upon addition of CO2 and alcohol to sequently equilibrates with the hydroxide/methylcarbonate
the corresponding binuclear carbonate. In fact, addition of pair 1/9. The whole process is so fast that all products appear to
extra CO2 is not strictly required. Thus, Ni or Pd carbonates 3 be formed simultaneously. Actually, the equilibrium constant
in C6D6 solution react instantly with stoichiometric amounts Keq favors the 1/9 pair (i.e., Keq ≫ 1). This circumstance may
of MeOH to afford equilibrium mixtures containing 9 and give the false impression that methanol breaks one of the
hydroxide 1, together with minor amounts of the corres- carbonate-metal fragments, to afford methylcarbonate 9 and
ponding bicarbonate (2) and methoxide (4), as shown in hydroxide 1 (see spectrum ii in Fig. 2). This could suggest that,
Scheme 5. The different species in the equilibria were identi- under more strict conditions, the same process could be forced
1
fied by comparison of their H and 31P{1H} NMR signals with to proceed in the remaining M–O bond to afford dimethyl car-
those of authentic samples, and the equilibrium condition bonate and hydroxide, 1. Nevertheless, despite our many
was confirmed by studying the perturbation of the system by efforts, such a reaction was never observed (see below).
changes in the temperature and the 3/MeOH ratio. Fig. 2 illus-
Rapid equilibration of the species placed within the braces
trates these experiments for the nickel derivative 3a. As long at the right side of Scheme 5 implies that CO2 “jumps” freely
as the amount of methanol added is close to stoichiometry from one molecule to another, exchanging the insertion site
(i.e. [M] : MeOH ≈ 1 : 1), the equilibrium mixture contains between M–OH and M–OMe. Most likely, the mechanism of
measurable amounts of each of the equilibrium components. CO2 transfer implies reversible elimination and re-insertion of
this molecule. In a previous article, we showed that, in fact,
CO2 insertion can be reverted for bicarbonates, 2, to afford the
corresponding hydroxides, 1.19d However, despite the lability
of alkylcarbonates (9–13), we have never observed directly the
reversal of CO2 insertion into metal–alkoxide bond. That is to
say, we could not detect alkoxides [(iPrPCP)M-OR] being
formed by CO2 extrusion from the alkylcarbonate [(iPrPCP)
M-OCOOR] complexes. In the literature, the reversibility of CO2
insertion into metal–alkoxide bonds has been supported with
isotope labelling experiments, using 13C-labeled CO2.15,16 In
the same vein, we performed an experiment bubbling dry
13CO2 through a solution of 9a in C6D6. A rapid enhancement
of the Ni–OCOOMe carbonate resonance took place in the 13C
NMR spectrum. This demonstrates the intermolecular CO2
exchange, that apparently takes place as shown in Scheme 6.
However, a control experiment revealed that this test could be
deceptive. According to the proposed mechanism, binuclear
carbonate 3a should not be able to exchange CO2, because this
compound is reluctant to eliminate CO2 even under forcing
conditions19d (such a process would lead to a high-energy, un-
precedented μ-oxo Ni(II) species). Yet, 3a exchanges 13CO2, at a
comparable rate to that observed in the case of 9a. We believe
that this exchange is actually catalysed by small amounts of
nickel bicarbonate complex 2a, in equilibrium with 3a. The
impurity 2a appears unavoidable, despite our efforts to
thoroughly dry the 13CO2 by storage over P2O5 or concentrated
H2SO4 for a long period of time (up to 2 months). The same
mechanism could be responsible for the 13CO2 exchange in
Scheme 5
Fig. 2 Effect of the addition of methanol on the 31P{1H} spectrum of 3a
at different temperatures. Starting from the bottom: (i) Spectrum 3a;
(ii) addition of MeOH (2 eq.); (iii–vi) the same spectrum recorded at 45,
65 and 85 °C, and again at room temperature; vii (top), excess MeOH
(10 equiv.). Key to signals: a: carbonate (3a); b: hydroxide (1a); c: methyl-
carbonate (9a) + bicarbonate (2a); d: methoxide (4a).
This journal is © The Royal Society of Chemistry 2019
Dalton Trans.

View Article Online
Paper
Dalton Transactions
Scheme 6
the case of 9a. We finally gained evidence for the reversibility of NMR-based titration of the hydroxide complexes (1a or 1b)
the CO2 insertion into the Ni–OMe bond by performing a with the corresponding alcohol.19e The same method proved
different kind of exchange experiment, in which 9a was reacted useful to determine the equilibrium constants for eqn (1) (K1),
with PhNCO to afford carbamate complex 14. In this case, the taking advantage of the ready availability of both Ni and Pd
irreversibility of isocyanate insertion, either into Ni–OMe or Ni– bicarbonates 2 as pure, crystalline materials. Combining eqn
OH bonds, rules out any potential route catalysed by the bicar- (1) with the reversal of eqn (3), one obtains the equilibrium
bonate impurity (Scheme 6). Monitoring the reaction of 9a with shown in Scheme 7, whose equilibrium constant Keq equals
PhNCO in a gas-tight NMR tube showed that, although initially the ratio K1/K3. This process is interesting because it provides
fast, it becomes noticeably slower as the reaction advances, a measure of the relative tendency of CO2 to insert into M–OR
nearly stopping at 40% conversion. This is expected if a CO2 or M–OH bonds. Note that this is the same partial equilibrium
elimination/insertion equilibrium precedes the irreversible shown within brackets in Scheme 5, which, in a real experi-
PhNCO insertion, as the released CO2 accumulates in the ment, cannot be isolated from the complication posed by the
sample tube. Similar conclusions have been reached by other formation of carbonate species (3).
authors when comparing the insertion reactions of CO2 into M–
A potential problem for the bicarbonate titration experi-
ments is that the 31P resonances of the bicarbonate and alkyl-
OR bonds with those of CS2 or other heterocumules.15,21c
As a consequence of the kinetic lability of CO2 insertion, carbonate derivatives do appear very close, and in some situ-
bicarbonates and alkylcarbonates are readily exchangeable. We ations, like the carbonate–methanol exchange (see Fig. 2),
checked that both bicarbonates 2a and 2b do in fact react with these may coalesce in a single peak (signal labeled “c”).
methanol, affording equilibrium mixtures containing the However, in contrast with these spectra, the processes
methylcarbonate and water. This kind of process is in fact very described in eqn (1) and (2) involve only two 31P sites. As a con-
general, as both bicarbonates 2 do react similarly with sequence, in these experiments the 31P{1H} signals appear
different alcohols, as shown in eqn (1). Not only bicarbonates narrow enough to be integrated individually. Thus, we under-
but also alkylcarbonates exchange readily with different alco- took K1 measurements for the exchanges of 2a and 2b with
hols, as illustrated by the exchange of 9a with ethanol (eqn methanol, ethanol and 2-methoxyethanol (the latter was
(2)). These reactions can be compared with classic esterifica- selected as a simplified surrogate of the glycoxide, in order to
tions and transesterifications, but their different nature is avoid strong intramolecular hydrogen bonds). Only the equili-
revealed by their very different rates. Carboxylic acids and brium corresponding to the reaction of 2a with ethanol (K1 Ni,
esters usually undergo these reactions much more slowly, even EtOH) posed serious problems for the integration of the 31P
in the presence of catalysts, while the “inorganic” versions {1H} signals, but in this case the value of the equilibrium con-
shown in eqn (1) and (2) are essentially instantaneous at the stant was estimated indirectly, as K1 (Ni, EtOH) = K2 × K1 (Ni,
room temperature. Equilibrium concentrations are attained MeOH).
shortly after mixing the reagents.
Table 1 collects the values of K1 and Keq, determined as dis-
cussed above, and Table 2 compares Gibbs free energies, ΔG°,
K1
ð
iPrPCPÞM–OCOOH þ ROH Ð ðiPrPCPÞM–OCOOR þ H2O
M ¼ Ni 2a or Pd; 2b
ð1Þ
ð
iPrPCPÞNi–OCOOMe þ EtOH
9a
K2¼0:5ð2Þ
)ꢀꢀꢀꢀꢀꢀ ðiPrPCPÞNi–OCOOEt þ MeOH
ꢀꢀꢀꢀꢀꢀ*
10a
ð2Þ
In a previous study, we determined the equilibrium con-
stants for the process described in eqn (3) (K3) using 31P{1H}
Scheme 7
Dalton Trans.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Dalton Transactions
Paper
Table 1 Bicarbonate/alcohol equilibrium constants, K1 and Keq
the M–OCOOR linkage, therefore the strength of the M–O
bonds in alkylcarbonates and bicarbonates is alike, and K1
values are closer to unit. Since the values of K3 are much
smaller than K1, the equilibrium constant Keq (= K1/K3) are
quite large. Moreover, since Keq are dominated by K3, these
show similar trends but in opposite directions. Because the
Ni(II) atom has a smaller size than Pd(II), electrostatic effects
are more marked for the former, resulting in comparatively
larger Keq values in the nickel system.
MeOCH2CH2OH
MeOH
EtOH
K1, Ni
0.23(4)
2.6(5)
1.7(4)
1.3(3)a
K1, Pd
0.20(3)
0.5(1)
K3,b Ni
K3,b Pd
Keq,c Ni
Keq,c Pd
5.7(2) × 10−3
5.0(10) × 10-2
40(7)
1.6(3) × 10−2
4.5(6) × 10−2
160(40)
38(9)
1.4(2) × 10−3
2.6(8) × 10−2
930(30)
4(1)
19(7)
a Calculated as K1 (Ni, MeOH) × K2. b Taken from ref. 19e. c Keq = K2/K3.
Even though the experimental values of Keq have relatively
large uncertainties, our results plainly indicate that the
thermodynamics of CO2 insertion favors the formation of
alkylcarbonate (M–OCOOR) over bicarbonates (M–OCOOH).
This is an intuitive concept, as it merely confirms that CO2
insertion takes place more readily into the weaker alkoxide M–
OR bond than in the stronger M–OH bond of hydroxides.
Furthermore, since hydroxides, alkoxides and their CO2 inser-
tion products are labile and rapidly exchanging in solution, it
can be concluded that under typical catalytic conditions, i.e.,
alcohol as the solvent and high CO2 pressure, the alkylcarbo-
nate M–OCOOR will always be the prevailing species. Since the
value of Keq is dominated by the inverse of K3, the preference
for the alkylcarbonate complex increases in the order Pd < Ni
and MeOCH2CH2OH < MeOH < EtOH. Furthermore, it is
Table 2 Experimentala and DFTb energy balances for eqn (3) (kcal mol−1
)
MeOCH2CH2OH
MeOH
EtOH
ΔG° exp., Ni
−2(1)
−1.4
−3.0(9)
−4.5
−4.0(9)
−4.8
ΔG° DFT, Ni
ΔE (SCF + ZPE), Ni
ΔG° exp., Pd
−2.2
−3.6
−3.9
−0.82(9)
+0.7
+0.1
−2.1(9)
−2.4
−1.5
−1.7(8)
−1.5
−0.5
ΔG° DFT, Pd
ΔE (SCF + ZPE), Pd
a ΔG° exp. = −RT·Ln(Keq). b PBE (6-31+G*, CPCM)//(6-311++G(3df, 2p),
CPCM).
corresponding to the Keq with theoretical values computed expected that the tendency to form the alkylcarbonate species
applying DFT methods validated in our previous work for eqn will be stronger for less acidic alcohols, like iPrOH.
(2).19e There is reasonable agreement between the experi-
Formation of organic carbonates
mental and theoretical set of data, and the trends are well
reproduced by the DFT calculation, reinforcing our confidence
in the general validity of the set of Keq data. In addition,
Table 2 displays temperature-independent energy balances
ΔE(SCF + ZPE), that can be related to the enthalpy term. These
are not very different from the experimental ΔG° values, indi-
cating that the contribution of thermal corrections is small.
Doubtless, this circumstance contributes to the good agree-
ment between experimental and computational data, as the
thermal corrections for DFT in solution phase are only
approximate.
As can be seen in Table 1, K1 values are about two orders of
magnitude larger than those of K3 (the alcohol exchange),
implying that bicarbonate–alkylcarbonate equilibria (1) are
much closer to thermoneutral than hydroxide–alkoxide (3).
This observation can be readily justified in the light of our pre-
vious work.18e In brief, the trends in the K3 and K1 depend on
the different intensity in which the R group influences the
thermodynamic stability of the partially ionic Mδ+–Oδ− bonds.
When comparing alkoxides and hydroxides, the ability of
polarizable R groups to disperse the excess of negative charge
that resides on the oxygen atom of alkoxides, decreases the
coulombic attractive force that makes up for a significant part
of the M–OR bond strength. The consequence is that the small
and negatively charged hydroxide anion binds more strongly
to the cationic metal fragment than an alkoxo OR ligand. This
results in a significant destabilization of the alkoxides relative
to hydroxides, hence in small values of K3. In contrast, the R
has a comparatively minor influence on the electrostatics of
We have investigated two types of processes that may poten-
tially cause the displacement of alkylcarbonate ligands as
organic alkylcarbonates: nucleophilic attack by the alcohol,
and electrophilic alkylation (Scheme 8). We show in this
section that, even though pincer alkylcarbonates are unreactive
in the first of these two pathways, they do react with electro-
philes to afford organic alkyl carbonates.
As expected, when hydroxides 1a and 1b are dissolved in
methanol-d4 and exposed to 4 bar of dry CO2 in thick-walled
pressure NMR tubes, they are fully converted into the corres-
ponding methylcarbonate complexes, 9a or 9b. In order to
generate dimethylcarbonate, these products should be able to
react further with methanol, according to eqn (4) (Scheme 8).
This reaction would regenerate the starting hydroxides, initiat-
ing a catalytic cycle. Therefore, in an attempt to force such pro-
Scheme 8
This journal is © The Royal Society of Chemistry 2019
Dalton Trans.

View Article Online
Paper
Dalton Transactions
cesses, the NMR samples were heated to 80 °C, and their 31P
{1H} and 1H NMR were recorded at regular intervals. The
colour and the spectra of the samples of the nickel complex
remained essentially unchanged over prolonged heating
(24 h), indicating that 9a is stable under the experimental con-
ditions, but those prepared from the Pd derivative become
gradually dark, at the same time that 31P{1H} signals for 9b
complex faded away. The different stabilities of the Ni and Pd
Scheme 9
methylcarbonates parallels the previously investigated behav- duced (neither dimethyl carbonate, nor the carbonate complex
iour of the corresponding methoxides 4 in methanol,20b 3a, see Scheme 9). Under the conditions applied, the only
suggesting that 4b may be involved in the decomposition of process observed was the thermal decomposition of the meth-
9b. GC-analyses of the samples did not reveal the presence of oxide 4a, that takes place as reported before.20b
dimethyl carbonate in neither of these experiments (M = Ni
As mentioned above, Lewis-acidic catalysts can activate
or Pd). The possibility that dimethyl carbonate could be methanol to behave as an electrophilic methylating agent
formed, but immediately degraded back to methanol and (Scheme 1). Such a mechanism is unlikely in the case of the
CO2 was definitively ruled out by means of an isotope label- mildly Lewis acidic, 16-electron Ni(II) or Pd(II) alkylcarbonate
ing experiment: a solution containing dimetylcarbonate and complexes. Notwithstanding, once we established the inability
hydroxide 1a in methanol-d4 was heated at 80 °C under 4 bar of nickel alkylcarbonate complexes to undergo nucleophilic
of dry 13CO2. If any dimethyl carbonate had been formed attack with alcohols, we decided to explore the opposite
under these conditions, the label would have been incor- approach and investigated the reactions of 9a and 9b with a
porated, and a significant enhancement of the MeOCOOMe variety of electrophilic alkylating reagents: methyl iodide,
resonance would have been observed in the 13C NMR spec- methyl tosylate and methyl triflate (see eqn (5) in Scheme 8).
trum of the mixture. However, the intensity of this signal The experiments were carried out at 60 °C in gas-tight NMR
remained unaltered after heating for >2 days. In contrast, the tubes. Solutions of Ni and Pd methylcarbonate complexes 9
intensity of the 1H CH3 signal of dimethyl carbonate was were generated in situ bubbling a small excess of CO2 through
observed to decay over time, which we attribute to the 0.3 M solutions of the corresponding alkoxides, 4a or 4b in
exchange of the methoxy groups of the deuteromethanol C6D6. The progress of such reactions was monitored using
1
used as solvent.
both 31P{1H} and H NMR. Table 3 collects rate and selectivity
The negative results of the precedent experiments were not data for these reactions. As observed through the 31P channel,
surprising, because the endergonic character of the reaction of the reactions appear straightforward, the signal of the starting
CO2 with methanol severely limits the feasibility of the cata- material being cleanly replaced by that of the corresponding
lytic cycle under the mild pressures accessible in an NMR complexes [(iPrPCP)M–X] (M = Ni, Pd; X = I (15a/b), OTf (16a/b)
tube. To improve the chances of detecting the formation of or OTs (17a/b)), identified by comparison with authentic
dimethyl carbonate, we carried out reactions at 20 atm of CO2 samples of such compounds prepared independently (see
at 80 and 100 °C in steel reactors, using neat methanol as a Experimental section). The simplicity of these 31P{1H} spectra
solvent. The final mixtures were analyzed using 31P{1H} and provide a convenient method to monitor the progress of the
GC-MS. In addition, we attempted the carboxylation of metha- reactions. In contrast, the 1H channel showed that the
nol or ethyleneglycol using sc-CO2 at 170 °C. In either case, the expected organic product, dimethylcarbonate (singlet at δ
Pd catalyst (generated from 1b) decomposed under the reac- 3.72 ppm), is usually formed along with some methanol (over-
tion conditions, while the nickel catalyst generated from 1a led lapping signals at δ 3.04 and 3.05 for Me and OH, respectively).
to yellow solutions containing mostly the corresponding alkyl- No other organic products were detected. The ratio of dimethyl
carbonate species, 9a or 13, but no organic carbonates were carbonate to methanol varies strongly, depending on the metal
formed.
complex and the reagent used, and in general it does not cor-
The lack of activity of the nickel complexes in the presence relate to the amount of alkylating reagent used. The latter
of methanol and CO2, under various conditions, indicates that observation rules out the possibility that methanol could arise
the alkylcarbonate ligand (i.e., M–OCOOR) is unable to experi- from traces of acid (HX) in the alkylating reagents (MeX). A
ence nucleophilic attack by the alcohol, even if this might more likely explanation is that MeOH arises from the competi-
happen intramolecularly (when R = CH2CH2OH). In an tive methylation of the ubiquitous bicarbonates 2, as methyl-
attempt to further explore the reactivity of alkylcarbonates carbonic acid is unstable and readily dissociates into the free
towards nuclophilic attack, we confronted 9a with the methox- alcohol and CO2.24 To explain the relatively large amounts of
ide complex 4a, a much stronger nucleophile than methanol methanol formed in some of these reactions, we assume that
itself. With this aim, we reacted a solution of 4a in C6D6 with a bicarbonates react with the methylating reagent more rapidly
volume of CO2 corresponding to ca. 0.5 equiv., to generate a than methylcarbonates. This proposal is supported by the
sample containing approximately equimolar amounts of 4a early disappearance of the small 31P signal of the bicarbonate
and 9a. The sample was sealed in a gas-tight PTFE valve NMR impurity, within the initial minutes of each experiment. The
tube and heated at 80 °C. No carbonate products were pro- competition between alkylation and hydrolysis equilibrium
Dalton Trans.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Dalton Transactions
Paper
Table 3 Yields and rates for the reactions of 9a and 9b with MeX (60 °C, C6D6)
k
ð
iPrPCPÞM‐OCOOMe þ MeX
ꢀꢀꢀ!
ð
iPrPCPÞM‐X þ MeOCOOMe ðþMeOH þ CO2Þ
ðtraces of H2OÞ
MeOHa (%)
Me2CO3 + 0.5·MeOH, %
T50
Rate (k)c × 106
b
M
X
MeX/[M]OCOOMe
Me2CO3 a (%)
d
Ni
Ni
Ni
Ni
Ni
Ni
Ni
Pd
Pd
Pd
Pd
Pd
I
I
I
I
1 : 1
10 : 1
15 : 1
20 : 1
10 : 1
20 : 1
1 : 1
10 : 1
20 : 1
10 : 1
20 : 1
1 : 1
36
50
74
45
68
83
86
68
88
65
56
81
52
40
19
54
10
7
22
30
20
9
62
70
83
72
73
86
97
83
98
69
57
95
>1000
240
50
—
8e
40e
68e
5
29
OTs
OTs
OTf
I
2100
410
29
f
f
—
—
90
11
121
1070
8
I
OTs
OTs
OTf
1500
110
3
28
109
f
f
—
—
a 1H NMR yields, % with regard to the converted organometallic complex. b Time for 50% consumption of the starting material, min. c First order
apparent rate constants unless otherwise stated. d Slow reaction. e Apparent 1/2-order rate constants, (mol L−1 1/2
)
s−1. f Fast reaction.
leads to more efficient water capture when the former reaction
is slow, as shown in Scheme 10. Note that a single molecule of
4a/b leads to the net production of two equivalents of
methanol.
The palladium methylcarbonate 4b is appreciably more
reactive towards electrophiles than that of nickel, 4a, but reac-
tion rates show the same trend for both complexes, increasing
in the order MeOTs < MeI ≪ MeOTf. Under the conditions
applied, reactions with MeOTf are too fast to allow direct NMR
monitoring. In the other two cases, we anticipated that the
decay of the 31P resonance of the starting material would
exhibit first-order kinetic dependencies both on the metal
complexes and the electrophile. Accordingly, these would
exhibit simple pseudo-first order kinetics if the electrophile/
complex ratio is large enough (10 : 1 or higher), and the appar-
ent rate constant would increase proportionally to the initial
electrophile concentration. However, rather more complex
kinetic behaviour was observed.
Although most experiments gave reasonably linear first-
order kinetic plots, those for the reactions of the nickel
complex 9a with MeI showed evident deviations (see Fig. 3).
Such kinetic behaviours are best described with a kinetic order
of 0.5 on 9a. In contrast, the first order plot for the palladium
complex, 9b shows only minor deviations. Apparent rate con-
stants (pseudo-first or -half order) are listed in Table 3,
together with T50 (time required to convert 50% of the starting
Fig. 3 Kinetic plots for the reactions of methylcarbonate complexes 9a
and 9b with MeI at 60 °C (C6D6). The plots mark the disappearance of
31
■
the P signal of the starting complex. [M] : [IMe] ratio = 1 : 10 ( ), 1 : 15
●
▲
(
), 1 : 20 ( ). A: First order plots for M = Ni (9a); B: same data as A, in a
Scheme 10
0.5-order plot; C: first order plot for M = Pd (9b).
This journal is © The Royal Society of Chemistry 2019
Dalton Trans.

View Article Online
Paper
Dalton Transactions
complexes 9). As expected, reaction rates increase with the confirmed our previous impressions, in the sense that radical
initial concentration of the electrophile but, surprisingly, the mechanisms are not playing any relevant role in these electro-
increases are in all cases much more pronounced than the philic ligand alkylations.
expected for a simple first order dependency. We believe that
The first order kinetic dependency on RX observed in most
this is probably a bulk solvent effect rather than a genuine of these reactions is uninformative, as it is equally compatible
kinetic dependency, as for 10–20 fold excess of MeX, with direct alkylation of the –OCOOMe ligand (mechanism i),
this reagent represents a significant fraction (5–10%) of the or oxidative addition followed by reductive elimination
total volume, which has important effects on the polarity (mechanism ii). However, the fractional 12-kinetic order
and solvating properties of the medium in relation to neat observed in the reactions of 9a with MeI is peculiar, and
C6D6.25
points to some less usual pathway. This feature is strongly
Among the conceivable mechanisms for the reactions of reminiscent for mechanisms in the upper part of Scheme 11.
organometallic compounds with electrophiles, three of them Ionization of the methylcarbonate/bicarbonate ligand gener-
appear more likely for the reaction of 9a/b with MeX: (i) direct ates a coordinatively unsaturated cationic metal species,
electrophilic attack on the coordinated methylcarbonate [(iPrPCP)M]+. Coordination of the electrophile (step marked
ligand; (ii) oxidative alkylation of the electrophile to the M(II) “1”) activates the X–Me bond towards nucleophilic attack by
centre, followed by Me⋯OC(O)Me reductive coupling at the the counteranion, [ROCO2]− to yield [(iPrPCP)MX] and the
resulting M(IV) intermediate, and (iii) free radical mechanisms organic carbonate (step “2”). In favour of such a mechanism, it
including an initial outer sphere electron transfer step.26 The could be argued that some soft metal cations (e.g. Ag+ or Hg2+)
latter possibility is more likely for Ni complexes. However, the are known to promote substitution reactions on alkyl halides
comparatively much higher reactivity of MeOTf than IMe R–X.19c,27,30 Note that the relative rates of steps 1 and 2 control
seems to point to a mechanism of type (i) and (ii) for both Ni the kinetic order of the overall process. When step 1 is rate-
and Pd.27 Seeking some additional proof in favour or against determining (RDS) the overall reaction rate depends on the
1
2
mechanism (iii), we carried out reactions of 9a and 9b with concentration of the cationic intermediate, hence the -order
1-iodohexen-5-ene, a typical “radical clock” test for free rad- kinetic dependency on 9. This could be the situation when the
icals.26,28 Such reactions yield the expected iodocomplexes 15, electrophile is IMe, a very poor ligand. On the other hand,
but the organic outcome was less clean than with methylating with a more coordinating electrophile like methyl tosylate,
electrophiles. For palladium (9b), the main organic product step 2 would become RDS. In this case, overall first-order
1
identified in the H spectrum was the cross-coupled carbonate dependency on the starting complex (9) would be observed. In
Me–OCO(CH2)4CHvCH2,29 identified on the basis of its general, the RDS role of steps 1 or 2 might not be clear-cut,
characteristic signals, a low field triplet (δ 4.06) and singlet (δ resulting in the observed deviations from ideal rate laws. It is
3.48) for the OC(O)O–CH2–C5H9 and CH3–OCO groups, worth recalling that a simple first-order dependency would be
respectively. For Ni (9a), the 1H spectrum showed an ill- also compatible with other mechanisms, such as the oxidative
defined mixture of organic products and the amount of the addition step, also drawn in Scheme 11 (bottom). This would
mixed carbonate was small. However, in none of these reac- be more feasible for Pd26,31 than for Ni.32 The RDS in this
tions we detected diagnostic doublet signals between 2.5 and mechanism should be the oxidative addition step (kOA), also
3.5 ppm, expected for the exo-methylene group of any cyclo- leading to an ionic pair. Therefore, the influence of the
pentyl products of the type X–CH2-cyclo-C5H9 (X = MeOC(O)O–, medium polarity on both oxidative addition and the dissocia-
HO–, I– etc.), that would be produced if the mechanism tive routes. Finally, methanol production is readily accommo-
involved free methyl-5-hexen-1-yl radical as an intermediate.28 dated in this mechanistic scenario assuming that 9 equili-
This result can be taken as a negative “radical clock” test, and brates with 2, as described above.
Scheme 11
Dalton Trans.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Dalton Transactions
Paper
Conclusions
Experimental section
In this study, we have investigated the interaction of CO2 with All operations were carried out under oxygen free nitrogen
monomeric nickel and palladium alkoxide complexes stabil- atmosphere using conventional Schlenk techniques or a nitro-
ized with the iPrPCP pincer ligand, and their potential as a gen-filled glove box. Solvents were rigorously dried and
model to improve our understanding of the fundamental steps degassed before use. Microanalyses were performed by the
of catalytic alcohol carboxylation.
Analytical Service of the Instituto de Investigaciones Químicas.
We have shown that insertion of CO2 into the M–O bonds NMR spectra were recorded on Bruker DPX-300, DRX-400 and
of Ni and Pd alkoxides is essentially instantaneous and quanti- DRX-500. Chemical shifts (δ) are in ppm. Solvents were used as
1
tative. The resulting alkylcarbonate complexes [(iPrPCP)M– internal standards for the reference of H and 13C spectra, but
OCOOR] are kinetically labile, readily exchanging their –OR chemical shifts are referenced with respect to TMS. 31P spectra
functionalities with water, to afford equilibrium mixtures con- are reported with respect to external H3PO4. Assignment of
taining the corresponding alcohol R-OH and bicarbonates signals were assisted by combined one-dimensional
[(iPrPCP)M–OCOOH]. In the absence of CO2, the bicarbonates (gated-13C) and two-dimensional 2D homonuclear 1H–1H
are prone to undergo partial decarboxylation into the corres- COSY and phase-sensitive NOESY, and heteronuclear 13C–1H
ponding binuclear carbonates. These reactions are fully revers- HSQC and HMBC heterocorrelations. IR spectra were recorded
ible, and the alkylcarbonate complexes are immediately regen- on a Bruker Tensor 27 FTIR spectrophotometer. Abbreviations
erated if the required ROH and CO2 are added to the system. for multiplicities are as usual, and the letters b and v denote
In fact, equilibrium constant measurements demonstrate that “broad” and “virtual”, respectively (e.g.: bm, broad multiplet,
the alkylcarbonate species [(iPrPCP)M–OCOOR] are very stable dvt = doublet of virtual triplets). Apparent constants for virtual
in solution, and thermodynamically favoured over the remain- couplings are marked with asterisk (*). Complexes [iPrPCP]M-X
ing species in equilibrium, namely, bicarbonates, carbonates, (M = Ni, X = Me,19c I (15a),19c OH (1a),19a OMe (4a),19b OEt
hydroxides or alkoxides. Provided that enough ROH and CO2 (5),19e OnBu (6),19e OiPr (7),19e OCH2CH2OH (8)19e OTf (16a);19c
are present in the equilibrium, alkylcarbonates are always M = Pd, X = Me,19c I (15b),19c OH (1b),19a OMe (4b),20 OTf
strongly prevalent, usually the only components detectable in (16b)19c) were prepared as previously described. Spectral simu-
the system. Therefore, if the M–OCOOR linkage could be lations were carried out with gNMR.33 GC-MS analyses were
cleaved by alcohol to release hydroxide [(iPrPCP)M–OH] and carried out in a ThermoQuest TRACE-GC Gas Chromatograph
dialkylcarbonate, RO–CO–OR, carboxylation would occur cata- equipped with a TRB-1 capillary column and an Automass
lytically, and reagents and products would attain their thermo- MULTI quadruple mass detector in electron-impact mode.
dynamic equilibrium concentrations. Unfortunately, our work
Synthesis of alkylcarbonates complexes [(iPrPCP)M–OCOOR]
[M = Ni, R = Me (9a); Et, (10); nBu, (11); iPr, (12); CH2CH2OH,
has demonstrated conclusively that simple alcohols like MeOH
are not competent to perform the nucleophilic displacement
of the alkylcarbonate unit. Similar conclusions were reached
(13); M = Pd, R = Me (9b)]
also for a diol, ethyleneglycol, even though elimination of the These complexes were formed by reacting the corresponding
cyclic ethylene carbonate would happen intramolecularly. alkoxides with CO2. The alkylcarbonate products could not be
Although Pd alkylcarbonates [(iPrPCP)Pd–OCOOR] decompose isolated in pure state due to their high solubility in common
irreversibly when heated in the neat alcohols under high CO2 organic solvents and their facile decarboxylation during the
pressures, their nickel counterparts survive even the under the evaporation process under vacuum, therefore they were charac-
harshest conditions, e.g., MeOH/sc-CO2 showing that thermal terized in solution by NMR spectroscopy.
instability is not the problem in this stystem. Conversely,
The general procedure is illustrated with the preparation of
methylcarbonate complexes [(iPrPCP)M–OCOOMe] behave as compound 9a: this was prepared in situ by bubbling CO2 into
mild nucleophiles towards electrophilic methylating reagents, a solution of 4a in C6D6. The reaction is immediate, and it
Me–X, to yield dimethylcarbonate and the corresponding “in- causes a colour change, from the characteristic orange of 4a to
organic” species, [(iPrPCP)M–X]. Our study has shown that a yellow hue. The transformation is extremely clean, and the
these reactions do involve true electrophilic attack, either at the only byproduct detected is the bicarbonate 2a, due to the pres-
metal centre or directly on the ligand, and not electron-transfer ence of a small amount of the hydroxide 1a.
mechanisms. Thus, successful closure of the carboxylation cata-
Spectroscopic and analytical data for 9a/b, 10–13
lytic cycle requires a catalyst that performs the simultaneous
1
3
nucleophilic activation of CO2, achieved at a basic alkoxo frag- 9a: H NMR (400.1 MHz, C6D6, 25 °C) δ 1.00 (dvt, 12H, JHH
≈
3
ment, and electrophilic activation of the alcohol to become an *JHP ≈ 6.9 Hz, CHMeMe); 1.40 (dvt, 12H, JHH ≈ *JHP ≈ 7.3 Hz,
alkyl transfer agent. The latter feature demands a strongly acidic CHMeMe); 2.01 (m, 4H, CHMeMe); 2.66 (vt, 4H, *JHP = 3.7 Hz,
3
active site, not available in our model system. These conclusions CH2 (PCP)); 3.72 (s, 3H, OOCH3); 6.79 (d, 2H, JHH = 7.3 Hz,
3
are in line with the fact that the best catalysts so far are metal CarHm); 6.98 (t, 1H, JHH = 7.3 Hz, CarHp). 13C{1H} NMR
alkoxides with a Lewis-acidic metal centre, and highlight the (100.6 MHz, C6D6, 25 °C): 18.0 (s, CHMeMe); 18.9 (s,
need for advanced catalyst design combining nearby basic and CHMeMe); 23.7 (vt, *JCP = 9.9 Hz, CHMeMe); 31.1 (vt, *JCP
=
acid reactive sites in close proximity.
13.2 Hz, CH2 (PCP)); 53.0 (s, OCH3); 122.4 (vt, *JCP = 8.7 Hz,
This journal is © The Royal Society of Chemistry 2019
Dalton Trans.

View Article Online
Paper
Dalton Transactions
2
CarHm); 125.4 (s, CarHp); 152.2 (t, JCP = 17.4 Hz, Car–i); 153.0 (s, CHMeMe); 18.8 (s, CHMeMe); 23.8 (vt, *JCP = 9.7 Hz,
(vt, *JCP = 13.1 Hz, Car–o); 158.1 (s, OCOO). 31P{1H} NMR CHMeMe); 30.9 (vt, *JCP = 12.8 Hz, CH2 (PCP)); 64.3 (s,
(162.0 MHz, C6D6, 25 °C): 56.5.
CH2OH); 68.4 (s, OCH2); 122.5 (vt, *JCP = 8.6 Hz, CarHm); 125.6
2
9b: 1H NMR (400.1 MHz, C6D6, 25 °C) δ 0.92 (dvt, 12H, 3JHH (s, CarHp); 151.4 (s, Car–i); 153.1 (t, JCP = 12.7 Hz, Car–o); 159.2
≈ *JHP ≈ 7.2 Hz, CHMeMe); 1.30 (dvt, 12H, JHH ≈ *JHP ≈ 7.8 (s, OCOO). 31P{1H} NMR (121.5 MHz, C6D6, 25 °C): 56.8.
3
Hz, CHMeMe); 2.07 (m, 4H, CHMeMe); 2.74 (vt, 4H, *JHP = 3.7
3
Synthesis of carbamate complexes 14 [(iPrPCP)Ni–NPhCOOMe]
and 14′ [(iPrPCP)Ni–OCONHPh]
Hz, CH2 (PCP)); 3.81 (s, 3H, OOCH3); 6.90 (d, 2H, JHH = 7.4
3
Hz, CarHm); 7.00 (t, 1H, JHH = 7.4 Hz, CarHp). 13C{1H} NMR
(100.6 MHz, C6D6, 25 °C): 17.9 (s, CHMeMe); 18.8 (s, 14: To a solution of 128 mg (0.3 mmol) of complex 4a in
CHMeMe); 24.4 (vt, *JCP = 10.7 Hz, CHMeMe); 32.0 (vt, *JCP 15 mL of hexane was added 36.1 μl (0.33 mmol) of phenyl iso-
=
11.6 Hz, CH2 (PCP)); 53.18 (s, OCH3); 122.9 (vt, *JCP = 10.4 Hz, cyanate. The colour of the solution changed from orange to
CarHm); 125.1 (s, CarHp); 151.4 (vt, *JCP = 10.7 Hz, Car–o); 154.5 pale yellow. The solvent was then evaporated under reduced
(s, Car–i); 159.5 (s, OCOO). 31P{1H} NMR (162.0 MHz, C6D6, pressure, the residue was extracted with hexane, and the solu-
25 °C): 59.7.
tion was filtered and concentrated. The product 14 crystallized
10: 1H NMR (300.1 MHz, C6D6, 25 °C) δ 1.02 (dvt, 12H, 3JHH (orange crystals) when the hexane solution was concentrated
≈ *JHP ≈ 6.8 Hz, CHMeMe); 1.25 (t, 3H, 3JHH = 7.0 Hz, CH2Me); to 1 or 2 mL and cooled to −30 °C. Yield: 71%. 1H NMR
1.42 (dvt, 12H, JHH ≈ *JHP ≈ 7.2 Hz, CHMeMe); 2.04 (m, 4H, (300 MHz, C6D6, 25 °C) 0.99 (m, 24H, CHMeMe); 1.84 (m, 4H,
3
CHMeMe); 2.67 (vt, 4H, *JHP = 4.1 Hz, CH2 (PCP)); 4.23 (c, 2H, CHMeMe); 2.86 (vt, 4H, *JHP = 3.5 Hz, CH2 (PCP)); 3.72 (s, 3H,
3
3JHH = 7.0 Hz, OCH2); 6.80 (d, 2H, JHH = 7.4 Hz, CarHm); 6.98 OMe); 6.82 (d, 2H, 3JHH = 7.5 Hz, CarHm); 6.87 (t, 1H, 3JHH = 7.3
3
3
3
(t, 1H, JHH = 7.4 Hz, CarHp). 13C{1H} NMR (75.5 MHz, C6D6, Hz, CPhHp); 6.96 (t, 1H, JHH = 7.5 Hz, CarHp); 7.28 (t, 2H, JHH
25 °C): 15.8 (s, CH2Me); 18.0 (s, CHMeMe); 18.8 (s, CHMeMe); = 7.9 Hz, CPhHm); 8.79 (d, 2H, JHH = 8.0 Hz, CPhHo). 13C{1H}
3
23.8 (vt, *JCP = 9.8 Hz, CHMeMe); 31.1 (vt, *JCP = 13.2 Hz, CH2 NMR (75.5 MHz, C6D6, 25 °C): 17.6 (s, CHMeMe); 19.1 (s,
(PCP)); 61.2 (s, OCH2); 122.4 (vt, *JCP = 8.6 Hz, CarHm); 125.3 (s, CHMeMe); 24.3 (vt, *JCP = 9.0 Hz, CHMeMe); 33.1 (vt, *JCP
CarHp); 152.3 (t, JCP = 17.5 Hz, Car–i); 153.0 (vt, *JCP = 13.0 Hz, 12.8 Hz, CH2 (PCP)); 50.2 (s, OMe); 119.7 (s, CPhHp); 121.8 (vt,
Car–o); 157.6 (s, OCOO). 31P{1H} NMR (121.5 MHz, C6D6, *JCP = 7.8 Hz, CarHm); 124.2 (s, CPhHo); 125.6 (s, CarHp); 127.8
=
2
25 °C): 56.5.
(s, CPhHm); 151.3 (vt, *JCP = 11.6 Hz, Car–o); 152.5 (s, CPh,i);
2
11: 1H NMR (500 MHz, C6D6, 25 °C) δ 1.02 (dvt, 12H, 3JHH
≈
155.8 (t, JCP = 17.0 Hz, Car–i); 158.4 (s, CO). 31P{1H} NMR
3
*JHP ≈ 6.8 Hz, CHMeMe); 1.31 (d, 6H, JHH = 6.0 Hz, CH(Me)2; (121.5 MHz, C6D6, 25 °C): 52.6. IR (Nujol mull): ν(CvO)
1.41 (dvt, 12H, JHH ≈ *JHP ≈ 7.2 Hz, CHMeMe); 2.05 (m, 4H, 1648 cm−1, ν(C–O) 1082 cm−1, ν(C–N) 1331 cm−1. Anal. Calcd
3
CHMeMe); 2.67 (ma, CH2 (PCP)); 5.05 (hept, 1H, OCH); 6.80 (d, for C28H4NNiO3P2: C, 61.56; H, 7.93; N, 2.56. Found: C, 61.58;
3
3
2H, JHH = 7.5 Hz, CarHm); 6.98 (t, 1H, JHH = 7.4 Hz, CarHp). H, 7.99; N, 2.52.
13C{1H} NMR (125.7 MHz, C6D6, 25 °C): 18.0 (s, CHMeMe);
14′: This complex was characterized on the basis of its NMR
18.9 (s, CHMeMe); 23.1 (s, CH(Me)2); 23.8 (vt, *JCP = 9.9 Hz, spectra, without the isolation of the pure product. In an NMR
CHMeMe); 31.1 (vt, *JCP = 13.3 Hz, CH2 (PCP)); 67.3 (s, OCH); tube, 10.3 mg (0.025 mmol) of hydroxide 1a was dissolved in
122.4 (vt, *JCP = 8.6 Hz, CarHm); 125.3 (s, CarHp); 152.2 (t, 2JCP
=
C6D6 and their 1H and 31P{1H} NMR spectra were registered.
17.0 Hz, Car–i); 153.1 (vt, *JCP = 13.2 Hz, Car–o); 157.3 (s, OCOO). After the addition of 2.2 μl (0.027 mmol) of phenyl isocyanate
31P{1H} NMR (202.4 MHz, C6D6, 25 °C): 56.4.
into the NMR tube, the quantitative formation of a new
12: 1H NMR (300 MHz, C6D6, 25 °C) δ 0.88 (t, 3H, 3JHH = 7.4 product was observed, for which we propose the structure
3
Hz, CH2Me); 1.02 (dvt, 12H, JHH ≈ *JHP ≈ 7.0 Hz, CHMeMe);
(
iPrPCP)Ni–OCONHPh (14′). 1H NMR (500 MHz, C6D6, 25 °C)
3
1.38 (m, 2H, CH2Me); 1.38 (dvt, 12H, JHH ≈ *JHP ≈ 7.1 Hz, 1.01 (dvt, 12H, 3JHH ≈ *JHP ≈ 6.9 Hz, CHMeMe); 1.41 (dvt, 12H,
CHMeMe); 1.62 (m, 2H, CH2CH2Me); 2.03 (m, 4H, CHMeMe); 3JHH ≈ *JHP ≈ 7.3 Hz, CHMeMe); 1.99 (m, 4H, CHMeMe); 2.69
3
3
2.67 (vt, 4H, *JHP = 4.2 Hz, CH2 (PCP)); 4.12 (c, 2H, JHH = 6.6 (vt, 4H, *JHP = 3.8 Hz, CH2 (PCP)); 6.82 (d, 2H, JHH = 7.4 Hz,
Hz, OCH2); 6.74 (d, 2H, 3JHH = 7.4 Hz, CarHm); 6.90 (t, 1H, 3JHH CarHm); 6.85 (t, 1H, JHH = 7.6 Hz, CPhHp); 7.00 (t, 1H, JHH
=
3
3
= 7.5 Hz, CarHp). 13C{1H} NMR (75.5 MHz, C6D6, 25 °C): 14.1 7.4 Hz, CarHp); 7.24 (t, 2H, JHH = 7.9 Hz, CPhHm); 7.75 (d, 2H,
(s, CH2Me); 18.0 (s, CHMeMe); 18.8 (s, CHMeMe); 19.7 (s, 3JHH = 8.5 Hz, CPhHo). 13C{1H} NMR (125.7 MHz, C6D6, 25 °C):
CH2Me); 23.8 (vt, *JCP = 9.5 Hz, CHMeMe); 31.1 (vt, *JCP = 13.1 17.7 (s, CHMeMe); 18.5 (s, CHMeMe); 23.6 (vt, *JCP = 9.8 Hz,
3
Hz, CH2 (PCP)); 32.4(s, CH2CH2); 65.5 (s, OCH2); 122.4 (vt, CHMeMe); 30.9 (vt, *JCP = 13.1 Hz, CH2 (PCP)); 117.0 (s,
2
*JCP = 8.1 Hz, CarHm); 125.3 (s, CarHp); 152.3 (t, JCP = 17.3 Hz,
CPhHp); 119.0 (s, CPhHo); 122.1 (vt, *JCP = 8.5 Hz, CarHm); 124.9
Car–i); 153.0 (vt, *JCP = 13.0 Hz, Car–o); 157.7 (s, OCOO). 31P{1H} (s, CarHp); 128.5 (s, CPh,m); 143.2 (s, CPh,i); 152.5 (vt, *JCP = 13.0
NMR (121.5 MHz, C6D6, 25 °C): 56.6.
Hz, Car–o); 152.6 (t, 2JCP = 17.7 Hz, Car–i); 158.3 (s, CO).
1
3
13: H NMR (300 MHz, C6D6, 25 °C) 1.00 (dvt, 12H, JHH
*JHP ≈ 7.0 Hz, CHMeMe); 1.40 (dvt, 12H, JHH ≈ *JHP ≈ 7.1 Hz,
≈
3
Independent syntheses of tosylate complexes 17a/b
CHMeMe); 1.97 (m, 4H, CHMeMe); 2.64 (vt, 4H, *JHP = 4.1 Hz, These compounds were prepared from the methyl complexes
CH2 (PCP)); 3.59 (s, 1H, OH); 3.81 (m, 2H, CH2OH); 4.17 (m, [(iPrPCP)M(Me)] following a similar procedure to that described
2H, OCH2); 6.77 (d, 2H, 3JHH = 7.3 Hz, CarHm); 6.97 (t, 1H, 3JHH before for the analogous triflate derivatives 16a/b.19c Below is
= 7.4 Hz, CarHp). 13C{1H} NMR (75.5 MHz, C6D6, 25 °C): 18.0 described the synthesis for compound 17a: to a solution of
Dalton Trans.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Dalton Transactions
Paper
114 mg (0.21 mmol) of [(iPrPCP)Ni(Me)] in 15 mL of THF components were collected in a small trap cooled with liquid
cooled at −30 °C was added 40 mg of p-toluenesufonic acid nitrogen. The solid residue was analyzed by 31P{1H} NMR, that
1
(HOTs) in 5 ml THF. The mixture was allowed to warm to the showed a 1 : 5 mixture of 9a and 3a, while the H NMR spec-
room temperature, and was taken to dryness under vacuum. The trum and GC analysis of the condensate recovered from the
solid residue was recrystallized from Et2O at – 20 °C. Yield: trap revealed only methanol in C6D6.
1
3
64%. H NMR (400.1 MHz, C6D6, 25 °C) 0.91 (dvt, 12H, JHH
≈
3
Reactions of carbonates (3a/b) with methanol
*JHP ≈ 6.9 Hz, CHMeMe); 1.53 (dvt, 12H, JHH ≈ *JHP ≈ 7.9 Hz,
CHMeMe); 1.96 (s, 3H, CH3); 2.27 (m, 4H, CHMeMe); 2.62 (vt, The reactions of carbonate complexes (3a/b) with MeOH were
3
4H, *JHP = 3.6 Hz, CH2 (PCP)); 6.74 (d, 2H, JHH = 7.4 Hz, investigated by NMR spectroscopy using NMR tubes fitted with
3
3
C
TolHm); 6.89 (d, 2H, JHH = 7.8 Hz, CarHm); 6.95 (t, 1H, JHH
=
gas-tight Teflon valves. The samples (solutions of the corres-
7.4 Hz, CPhHp); 8.03 (d, 2H, JHH = 8.1 Hz, CTolHo). 13C{1H} ponding carbonate complexes in C6D6 with a specific amount
NMR (100.6 MHz, C6D6, 25 °C): 18.3 (s, CHMeMe); 19.5 (s, of MeOH) were prepared in a glove-box.
CHMeMe); 21.1 (s, CH3); 24.0 (vt, *JCP = 10.1 Hz, CHMeMe);
3
Titration of alkylcarbonate (9a/b) complexes with alcohols
30.0 (vt, *JCP = 13.5 Hz, CH2 (PCP)); 122.7 (vt, *JCP = 8.5 Hz,
CarHm); 125.8 (s, CarHp); 126.6 (s, CTolHm); 128.7 (s, CTolHo); The reversible reactions of bicarbonate complexes (2a/b) with
2
139.6 (s, CTol,p); 143.5 (s, CTol,i); 147.7 (t, JCP = 17.0 Hz, Car–i); alcohols were performed in regular, septum-capped NMR
153.1 (vt, *JCP = 12.3 Hz, Car–o). 31P{1H} NMR (202.4 MHz, tubes, using C6D6 as solvent. Alcohols used in this protocol
C6D6, 25 °C): 58.8. IR (Nujol mull): ν(S–O) 1159 cm−1
,
were dried by distillation from the corresponding sodium alk-
1113 cm−1, 1002 cm−1. Anal. Calcd for C27H42NiO3P2S: C, oxides, generated in situ by dissolving pieces of sodium in the
57.16; H, 7.46; S, 5.65. Found: C, 57.39; H, 7.17; S, 5.56. alcohol prior to the distillation. The following experimental
17b: This was obtained similarly to its Ni analogue in 70% protocol was applied: in the glovebox, an amount close to
1
3
yield. H NMR (500 MHz, C6D6, 25 °C) 0.87 (dvt, 12H, JHH
≈
0.025 mmol (2a: 11.4 mg; 2b 11.9 mg) of the corresponding
3
*JHP ≈ 7.0 Hz, CHMeMe); 1.41 (dvt, 12H, JHH ≈ *JHP ≈ 7.9 Hz, bicarbonate was accurately weighed in an analytical balance.
CHMeMe); 1.98 (s, 3H, CH3); 2.29 (m, 4H, CHMeMe); 2.70 (bs, The sample was dissolved in 0.6 mL of C6D6 and transferred to
4H, CH2 (PCP)); 6.88 (d, 2H, JHH = 7.3 Hz, CTolHm); 6.92 (d, an NMR tube capped with a rubber septum. A second septum-
3
3
3
2H, JHH = 7.7 Hz, CarHm); 6.99 (t, 1H, JHH = 7.3 Hz, CPhHp); capped NMR tube was used to store the required alcohol. The
3
8.17 (d, 2H, JHH = 7.8 Hz, CTolHo). 13C{1H} NMR (125.7 MHz, tubes were taken from the box, and 31P{1H} and 1H NMR
C6D6, 25 °C): 17.7 (s, CHMeMe); 18.9 (s, CHMeMe); 20.7 spectra of the complex sample solution were recorded on the
(s, CH3); 24.2 (vt, *JCP = 10.9 Hz, CHMeMe); 30.9 (vt, *JCP = 11.7 300 MHz NMR spectrometer. With gastight syringes of suitable
Hz, CH2 (PCP)); 123.0 (vt, *JCP = 10.5 Hz, CarHm); 125.2 (s, volume, increasing amounts of the alcohol were taken from
CarHp); 126.6 (s, CTolHm); 128.4 (s, CTolHo); 139.1 (s, CTol,p); the reservoir tube and successively added to the sample of the
143.7 (s, CTol,i); 151.0 (vt, *JCP = 10.1 Hz, Car–o); 151.6 (s, Car–i). complex (e.g., 0.5, 1.5, 3.5, 6.5 μL, etc.). After each addition, the
31P{1H} NMR (162.0 MHz, C6D6, 25 °C): 61.7. IR (Nujol mull): sample was gently shaken, returned to the NMR probe, and
ν(S–O) 1152 cm−1, 1112 cm−1, 1001 cm−1. Anal. Calcd for the 31P{1H} spectra were recorded. Control experiments were
C27H42O3P2PdS: C, 52.73; H, 6.88; S, 5.21. Found: C, 52.91; H, performed to check that the equilibration of the species was
6.99; S, 4.99.
essentially instantaneous. The equilibrium constant K1 was
calculated as [MOCOOR]/[MOCOOH] × [H2O]/[ROH] for each
addition. The alkylcarbonate to bicarbonate ratio was directly
Transformation of alkylcarbonate 4a into carbonate 3a
0.025 mmol of complexes 4a or 6 were dissolved in 0.6 ml of obtained from the 31P integrals, and the ratio of water to
toluene-d8 and transferred to an NMR tube capped with a alcohol was deduced from the total amount of alcohol added
rubber septum. The septum was pierced with a thin hypoder- using stoichiometry relationships.
mic needle to enable slow gas exchange. A Schlenk tube,
To investigate the reaction between complex 9a and EtOH
capped with a unpierced Sub-a-Seal™ septum, was flushed the same protocol was applied, but in this case the alkylcarbo-
with N2 and placed in a thermostatic oil bath at 100 °C. Once nate sample was prepared in situ bubbling CO2 into a solution
its temperature was stabilized, the Schlenk tube was opened of complex 4a.
and the NMR sample was introduced inside in N2 counterflow.
The system was closed and the Suba-a-Seal septum was
Attempted carboxylation of alcohols using 1a or 1b as catalysts
pierced with a needle. The N2 flow was then adjusted to 2 cc The following experiments were designed to determine the
min−1 with an accurate flow regulator. The system was care- potential formation of organic carbonates.
fully opened at regular intervals, and the sample was removed
from the tube to record its NMR spectra (1H and 31P{1H}) as of the Ni or Pd hydroxides 1a or 1b, respectively (ca.
described in the text. 0.025 mmol), were dissolved in dry methanol (0.6 mL) and
(a) Pressure NMR tube experiments: In the glove box, samples
To confirm that free alcohol is the only organic product of transferred to thick-walled glass NMR tubes provided with
the alkylcarbonate decarboxylation, a solution of compound screw PTFE valves and Swagelock-type fittings for 1/16″ steel
9a in C6D6, generated from 12 mg (0.025 mmol) of complex 4a, tube. These were taken from the glove box, attached to a
was evaporated under reduced pressure. The volatile pressure line and flushed three times with CO2. In the last
This journal is © The Royal Society of Chemistry 2019
Dalton Trans.

View Article Online
Paper
Dalton Transactions
cycle, the pressure was set to 4 bar. The 31P{1H} NMR spectra Computational details
confirmed that the conversion of hydroxides 1a and 1b into
All calculations were performed with the Spartan’16 software
package.35 The approach used was the same described
before.19e The structures of all complexes involved in the CO2
exchange reaction shown in Scheme 7 were modelled without
simplifications, using the experimental X-ray diffraction data
as the initial guess for bicarbonates 2a and 2b, and adding the
corresponding R groups for alkylcarbonates. Models for the
hydroxide and alkoxide complexes were taken as previously
reported.18e Geometry optimizations and vibrational analyses
were carried out with the PBE functional and the 6-31G* basis
set for nickel complexes, and LACVP* for the Pd series. The
latter comprises the same 6-31G* functions for light elements
and the LANL2DZ pseudopotential for Pd. Solvent effects
(benzene) were included in the optimization procedure with
the CPCM implicit model, setting the Bondi-type PCM radii
for Ni and Pd as 2.28 and 2.48 Å, respectively. Electronic (SCF)
energies were refined with a single point calculation at the
PBE/6-311++G(3df,2p)/CPCM level. At this level of the theory,
Spartan uses the all-electron def2-TZVP basis set to describe
the Pd atom. Single-point calculations were accelerated using
the “dual” option, that specifies that SCF convergence is
achieved at the 6-311G* level, and then corrected perturbatively
for the effect of additional diffuse and polarization functions.
The final ΔG° and ΔE values for each molecular model were
obtained adding the thermal and ZPE corrections from the 6-
31G* calculation, respectively (thermal correction = ΔG° −
E(SCF)).
the methylcarbonates 9a and 9b, respectively, was complete.
In two series of experiments, the samples were heated at 80 °C
or 100 °C for 24 h, after which time the samples were analyzed
by 31P{1H} NMR and GC-MS. A similar experiment was con-
ducted with 1a using 4 bar of a 3 : 1 CO2/13CO2 (25% 13C) in
methanol-d4, and 0.01 ml of dimethyl carbonate were added.
13C{1H} spectra taken above and after the experiment showed
no enhancement of the carbonyl signal of dimethyl carbonate.
(b) Medium pressure experiments: A 50 mL steel reactor was
charged with 1a or 1b (0.1 mmol) in 15 ml of methanol. The
reactor was purged with CO2 and pressurized with the same
gas at 20 bar. Then it was heated at 100 °C for 24 h, after
which time it was cooled down, and the pressure was carefully
vented.
(c) sc-CO2 experiments: The reactor, once degassed and
purged with CO2 was loaded with a solution containing
0.2 mmol of the complex (1a or 1b) and 0.25 ml (2.35 mmol)
of toluene as internal standard. Next, 15 mL of liquid CO2
were pumped in, and the reactor was heated to 100 °C for 2 h.
The pressure rose up to 170 bar. The system was then cooled
to −80 °C and carefully vented.
The solutions obtained in all experiments were analyzed
using GC-MS techniques. These analyses ruled out net pro-
duction of organic carbonates. In all experiments with Pd, a
dark suspension was invariably formed and the 31P{1H}
spectra showed that the pincer framework does not survive the
conditions used.
Reactions of alkylcarbonate (9a/b) complexes with alkylating
agents (MeI, MeOTf, MeOTs)
Conflicts of interest
These reactions were studied by 31P{1H} NMR spectroscopy.
The identity of the products was established by comparison
with 31P data for authentic samples. NMR samples of the alkyl-
carbonates (9a/b) were prepared dissolving 0.025 mmol of the
corresponding methoxide (4a/b) in C6D6 inside the globe-box.
Then, the samples were transferred to NMR tubes, and CO2
was bubbled into the solution. The prescribed amount of alky-
lating agent (1, 10, 15 or 20 eqs) was added to each solution
with the help of a micro syringe, then solvent was added to
adjust a fixed volume (0.6 ml) and the tubes were sealed. In
the cases the reactions were fast enough, the reactions were
carried out in the NMR probe, pre-heated at 60 °C and moni-
tored continuously. For slower reactions, the samples were
placed in a thermostatic oil bath, and the spectra were
recorded each 6–12 h.
There are no conflicts to declare.
Acknowledgements
The Spanish Ministerio de Economía
y Competitividad
(MINECO) and the FEDER funds of the European Union
(CTQ2015-68978-P) supported this work. We warmly thank
Prof. David J. Cole-Hamilton (St. Andrews University, UK) for
the use of reactor facilities to investigate the reactions of
hydroxides 1a and 1b with methanol in sc-CO2.
Notes and references
Reactions of 9a,b with 6-iodo-1-hexene
1 (a) M. Aresta and A. Dibenedetto, Dalton Trans., 2007, 2975;
(b) D. J. Darensbourg, Inorg. Chem., 2010, 49, 10765;
(c) S. Huang, B. Yan, S. Wang and X. Ma, Chem. Soc. Rev.,
2015, 44, 3079.
2 (a) M. Cokoja, M. E. Wilhelm, M. H. Anthofer,
W. A. Herrmann and F. E. Kühn, ChemSusChem, 2015, 8,
2436; (b) M. R. Kember, A. Buchard and C. K. Williams,
Chem. Commun., 2011, 47, 141.
These reactions were investigated as described above except
that the alkylating reagent was 6-iodo-1-hexene, and the
9 : alkyl iodide was 1 : 2. The samples were heated in an oil
bath at 100 °C, and they were taken at regular intervals to
perform NMR analyses until all the starting material was con-
sumed. 6-Iodo-1-hexene was prepared from 5-hexen-1-ol follow-
ing a literature synthetic method.34
Dalton Trans.
This journal is © The Royal Society of Chemistry 2019

View Article Online
Dalton Transactions
Paper
3 G. Sienel, R. Rieth and K. T. Rowbottom, in Ullmann’s
Encyclopedia of Industrial Chemistry, Wiley-VCH, 2000.
4 M. Aresta, A. Dibenedetto, A. Angelini and I. Papái, Top.
Catal., 2015, 58, 2.
Berlin Heidelberg, 2013; (e) The Privileged Pincer-Metal
Platform: Coordination Chemistry & Applications, ed. G. van
Koten and R. A. Gossage, Springer, Berlin, Heidelberg,
2016.
5 Handbook of Green Chemistry and Technology, ed. J. Clark 18 For some relevant examples, see: (a) D. Huang,
and D. Mcquarrie, Blackwell Publishing, Oxford, 2002.
6 (a) M. Aresta, A. Dibenedetto and C. Pastore, Inorg. Chem.,
2003, 42, 3256; (b) Q. Cai, B. Lu, L. Guo and Y. Shan, Catal.
Commun., 2009, 10, 605.
7 (a) A. Dibenedetto, M. Aresta, A. Angelini, J. Ethiraj and
B. M. Aresta, Chem. – Eur. J., 2012, 18, 10324; (b) M. Aresta,
A. Dibenedetto, C. Pastore, A. Angelini, B. Aresta and
I. Pápai, J. Catal., 2010, 269, 44; (c) M. Aresta,
A. Dibenedetto, C. Pastore, C. Cuocci, B. M. Aresta and
E. De Giglio, Catal. Today, 2008, 137, 125; (d) Y. Yoshida,
Y. Arai, S. Kado, K. Kunimori and K. Tomishige, Catal.
Today, 2006, 115, 95; (e) K. Tomisighe, Y. Ikeda,
T. Sakaihori and K. Fujimoto, J. Catal., 2000, 192, 355.
O. Makhlynets, L. L. Tan, S. C. Lee, E. V. Rybak-Akimova
and R. H. Holm, Proc. Natl. Acad. Sci. U. S. A., 2010, 108,
1222; (b) D. Huang, O. Makhlynets, L. L. Tan, S. C. Lee,
E. V. Rybak-Akimova and R. H. Holm, Inorg. Chem.,
2011, 50, 10070; (c) H. W. Suh, T. J. Smeier, N. Hazari,
R. A. Kemp and M. K. Tankase, Organometallics, 2012,
31, 8225; (d) Y. E. Kim, S. Oh, S. Kim, O. Kim, J. Kim,
S. W. Han and Y. Lee, J. Am. Chem. Soc., 2015, 137,
4280; (e) S. Oh, S. Kim, D. Lee, J. Gwak and Y. Lee,
Inorg. Chem., 2016, 55, 12863; (f) H. Mousa, J. Bendix
and O. F. Wendt, Organometallics, 2018, 37, 2581;
(g) K. J. Jonasson, A. H. Mousa and O. F. Wendt,
Polyhedron, 2018, 143, 132.
8 (a) Q. Cai, H. Jin, B. Lu, H. Tangbo and Y. Shan, Catal. 19 (a) J. Cámpora, P. Palma, D. Del Río and E. Álvarez,
Lett., 2005, 103, 225; (b) S. Fujita, B. M. Bhanage,
Y. Ikushima and M. Arai, Green Chem., 2001, 3, 87.
Organometallics, 2004, 23, 1652; (b) J. Cámpora, P. Palma,
D. Del Río, M. M. Conejo and E. Álvarez,
Organometallics, 2004, 23, 5653; (c) L. M. Martínez-Prieto,
C. Melero, D. Del Río, P. Palma, J. Cámpora and
9 (a) F. D. Bobbink, W. Guruszka, M. Hulla, S. Das and
P. J. Dyson, Chem. Commun., 2016, 52, 10787; (b) M. Aresta,
A. Dibenedetto, E. Fracchiolla, P. Giannocaro, C. Pastore,
I. Pápai and G. Schubert, J. Org. Chem., 2005, 70, 6177.
10 (a) K. Kohno, J.-C. Choi, Y. Oshima, H. Yasuda and
T. Sakakura, ChemSusChem, 2008, 1, 186; (b) M. Ruf,
F. A. Schell, R. Walz and H. Vahrenkamp, Chem. Ber./Recl.,
1997, 130, 101.
E.
Álvarez,
Organometallics,
2012,
31,
1425;
(d) L. M. Martínez-Prieto, C. Real, E. Ávila, E. Álvarez,
P. Palma and J. Cámpora, Eur. J. Inorg. Chem., 2013,
5555; (e) L. M. Martínez-Prieto, P. Palma, E. Álvarez and
J. Cámpora, Inorg. Chem., 2017, 56, 13086.
20 (a) C. Melero, L. M. Martínez-Prieto, P. Palma, D. Del Río,
E. Álvarez and J. Cámpora, Chem. Commun., 2010, 40, 8851;
(b) L. M. Martínez-Prieto, E. Ávila, P. Palma, E. Álvarez and
J. Cámpora, Chem. – Eur. J., 2015, 21, 9833.
11 (a) M. Poor Kalhor, H. Chermette, S. Chambrey and
D. Ballivet-Tkatchenko, Phys. Chem. Chem. Phys., 2011, 13,
2401; (b) G. Laurenczy, M. Picquet and L. Plasseraud,
J. Organomet. Chem., 2011, 696, 1904; (c) T. Sakakura, 21 (a) J. Cámpora, I. Matas, P. Palma, E. Álvarez, C. Graiff and
J.-C. Choi, Y. Saito and T. Sako, Polyhedron, 2000, 19, 573;
(d) D. Ballivet-Tkatchenko, O. Douteau and S. Stutzmann,
Organometallics, 2000, 19, 4563; (e) G. Laurenczy,
A. F. Dalebrook, M. Picquet and L. Plasseraud,
J. Organomet. Chem., 2015, 796, 53.
A. Tiripicchio, Organometallics, 2007, 26, 3840; (b) J. Ruiz,
M. T. Martínez, F. Florenciano, V. Rodríguez, G. López,
P. A. Chalonnier and P. B. Hitchcock, Inorg. Chem., 2003,
42, 3650; (c) E. Hevia, J. Pérez, L. Riera, V. Riera, I. del Río,
S. García-Granda and D. Miguel, Chem. – Eur. J., 2002, 8,
4510; (d) D. S. Glueck, L. J. Newman-Winslow and
R. G. Bergman, Organometallics, 1991, 10, 1462.
12 (a) M. Aresta, A. Dibenedetto, C. Pastore, I. Pápai and
G. Schubert, Top. Catal., 2006, 40, 71; (b) A. Dibenedetto,
C. Pastore and M. Aresta, Catal. Today, 2006, 115, 88.
13 M. Pool Kahlor, H. Chermette and D. Ballivet-Tkatchenko,
Polyhedron, 2012, 32, 73.
14 M. Aresta, A. Dibenedetto, L. Gianfrate and C. Pastore,
J. Mol. Catal. A: Chem., 2003, 204, 245.
15 R. D. Simpson and R. G. Bergman, Organometallics, 1992,
11, 4306.
22 Depending, of course, on the existence of stable carbonate
complexes. See for example ref. 10b.
23 Although not explicitly reported, on the basis of published
data it is very likely that the hydrolysis of a bismuth(III)
methylcarbonate stabilized by a dianionic pincer ligand
would afford the corresponding binuclear carbonate. See:
S. F. Yin, J. Maruyama, T. Yamashita and S. Shimada,
Angew. Chem., Int. Ed., 2008, 47, 6590.
16 D. J. Darensbourg, W.-Z. Lee, A. L. Phelps and E. Guidry,
Organometallics, 2003, 22, 5585.
17 (a) Pincer Compounds, Synthesis and Applications, ed. D.
24 A. Dibenedetto, M. Aresta, P. Giannocaro, C. Pastore,
I. Pápai and G. Schubert, Eur. J. Inorg. Chem., 2006, 908.
Morales-Morales,
Elsevier,
Amsterdam,
2018; 25 M. B. Smith and J. March, in March’s Advanced Organic
(b) S. Murugesan and K. Kirchner, Dalton Trans., 2018, 45,
416; (c) Pincer and Pincer-Type Complexes: Applications in
Chemistry. Reactions, Mechanism, and Structure, John Wiley
& Sons, Hoboken, New Jersey, USA, 2007, ch. 9.
Organic Synthesis and Catalysis, ed. K. J. Szabo and 26 X. Hu, Chem. Sci., 2011, 2, 1867.
O. F. Wendt, VCH, 2014; (d) Organometallic Pincer 27 See ref. 24, pp. 504–505.
Complexes, ed. G. van Koten and D. Milstein, Springer, 28 D. Griller and K. U. Ingold, Acc. Chem. Res., 1980, 13, 317.
This journal is © The Royal Society of Chemistry 2019
Dalton Trans.

View Article Online
Paper
Dalton Transactions
29 B. M. Trost, M. R. Machacek and Z. T. Ball, Org. Lett., 2003, 33 Budzelaar, P. H. M. gNMR, NMR simultation software,
5, 1895.
available in the web at https://home.cc.umanitoba.ca/~bud-
zelaa/gNMR/gNMR.html.
34 D. M. Hodgson, A. H. Labande, F. Y. T. M. Pierard and
M. A. E. Castro, J. Org. Chem., 2003, 68, 6153.
30 T. Laube, A. Weidenhaupt and R. Hunziker, J. Am. Chem.
Soc., 1991, 113, 2561.
31 H. Zhang and A. Lei, Dalton Trans., 2011, 40, 8745.
32 N. M. Camasso and M. S. Sanford, Science, 2015, 347, 35 Spartan’16, Wavefunction, Inc., Irvine, CA, USA, Version
1218.
2.0.8., 2017.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2019