Unprecedented Carbonato Intermediates in Cyclic Carbonate Synthesis Catalysed by Bimetallic Aluminium(Salen) Complexes

Article abstract of DOI:10.1002/cssc.201501664

The mechanism by which [Al(salen)]₂O complexes catalyse the synthesis of cyclic carbonates from epoxides and carbon dioxide in the absence of a halide cocatalyst has been investigated. Density functional theory (DFT) studies, mass spectrometry and ¹H NMR, ¹³C NMR and infrared spectroscopies provide evidence for the formation of an unprecedented carbonato bridged bimetallic aluminium complex which is shown to be a key intermediate for the halide-free synthesis of cyclic carbonates from epoxides and carbon dioxide. Deuterated and enantiomerically-pure epoxides were used to study the reaction pathway. Based on the experimental and theoretical results, a catalytic cycle is proposed.

Full text of DOI:10.1002/cssc.201501664

DOI: 10.1002/cssc.201501664
Communications
Unprecedented Carbonato Intermediates in Cyclic
Carbonate Synthesis Catalysed by Bimetallic
Aluminium(Salen) Complexes
JosØ A. Castro-Osma,[a] Michael North,*[a] Willem K. Offermans,*[b] Walter Leitner,[c] and
Thomas E. Müller[b]
The mechanism by which [Al(salen)]2O complexes catalyse the
synthesis of cyclic carbonates from epoxides and carbon diox-
ide in the absence of a halide cocatalyst has been investigated.
Density functional theory (DFT) studies, mass spectrometry
1
and H NMR, 13C NMR and infrared spectroscopies provide evi-
dence for the formation of an unprecedented carbonato
bridged bimetallic aluminium complex which is shown to be
a key intermediate for the halide-free synthesis of cyclic carbo-
nates from epoxides and carbon dioxide. Deuterated and
enantiomerically-pure epoxides were used to study the reac-
tion pathway. Based on the experimental and theoretical re-
sults, a catalytic cycle is proposed.
Utilization of carbon dioxide as a sustainable carbon source for
the synthesis of organic chemicals has emerged as a major
challenge.[1] One commercial process that uses carbon dioxide
as a feedstock is the synthesis of cyclic carbonates 2 from ep-
oxides 1. Cyclic carbonates have a number of applications in-
cluding as electrolytes in lithium-ion batteries.[2] Current indus-
Scheme 1. Synthesis of cyclic carbonates from epoxides and CO2.
trial processes for the synthesis of 2 utilize inefficient quaterna-
ry ammonium or phosphonium salt catalysts and require the
use of high temperatures and pressures.[3] Aluminium(salen)
complexes 3, in conjunction with a halide co-catalyst are
amongst the most active catalyst systems for this reaction at
or close to room temperature and one bar pressure.[4] The
halide acts as a nucleophile to ring-open the epoxide. We re-
cently reported, however, that complex 3 can act as a catalyst
for the synthesis of 2 at 50–1008C and 10–50 bar without the
need for a halide co-catalyst (Scheme 1).[5] This unexpected
finding is incompatible with the established mechanisms for
cyclic carbonate synthesis[6] and prompted a study of the reac-
tion mechanism using density functional theory (DFT) calcula-
tions along with spectroscopic and isotope labelling experi-
ments. The resulting unprecedented reaction pathways are re-
ported herein.
DFT calculations were initially carried out using [Al(acen)]2O
4 as a model to accelerate the calculations. Accuracy and suita-
bility of the applied DFT methodology were determined by
benchmark calculations and comparison with literature data as
detailed in the Supporting Information.[7] Complexes 3 and 4
are both catalysts for the reaction shown in Scheme 1.[8] First
calculations focused on the interaction between complex 4
and epoxides 1a–c, carbon dioxide, and product 2a. These cal-
culations indicated that complex 4 would form an adduct with
carbon dioxide, but not with epoxides or cyclic carbonates.
One possible interaction between complex 4 and carbon di-
oxide is via an oxygen atom of carbon dioxide and an alumini-
um atom of 4. However, the DFT calculations did not provide
evidence for such an interaction.[7] Another possible interaction
is that between the bridging oxygen of complex 4 and the
carbon of carbon dioxide. DFT calculations confirmed the pres-
ence of a stable adduct of this type (adduct state in Figure 1).
[a] Dr. J. A. Castro-Osma, Prof. Dr. M. North
Green Chemistry Centre of Excellence, Department of Chemistry
The University of York
York, YO10 5DD (UK)
E-mail: michael.north@york.ac.uk
[b] Dr. W. K. Offermans, Dr. T. E. Müller
CAT Catalytic Center
RWTH Aachen University
Worringerweg 2
D-52074 Aachen (Germany)
E-mail: Willy.Offermans@CatalyticCenter.RWTH-Aachen.de
[c] Prof. Dr. W. Leitner
Lehrstuhl für Technische Chemie und Petrolchemie
ITMC, RWTH Aachen University
Worringerweg 1
D-52074 Aachen (Germany)
A
The molar standard entropy change (DrS ) upon adduct forma-
Supporting Information for this article can be found under
http://dx.doi.org/10.1002/cssc.201501664.
tion from complex 4 and carbon dioxide was À81 JmolÀ1 KÀ1,
ChemSusChem 2016, 9, 791 – 794
791
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
Figure 2. Possible bonding arrangements of the Al-OC(O)O-Al-unit in
[Al(acen)]2CO3.
carbon dioxide, were found to be À15, À13, and À6 kJmolÀ1
A
for A–C, respectively. Thus, the differences in DrG values are
°
A
small. The Gibbs energies of activation (Dr G ) were calculated
to be 24 and 19 kJmolÀ1 for the transformation of A to B and
from B to C, respectively. Since the isomeric states A-C are
close in free energy and since they can easily interconvert,
they provide extra stability to the carbonato bridged complex.
To confirm that the modelling on acen complexes was rele-
vant to salen complexes, the three states shown in Figure 1
were recalculated as the corresponding salen complexes de-
rived from complex 3a.[7] Essentially identical structures and
energetics for the transition state and the carbonato bridged
complex 5a were obtained. The transition state again had
A
Figure 1. Models and free energies at standard conditions (DG ) of the cal-
culated adduct, transition, and product states of CO2 insertion into the Al-m-
O-Al unit of 4. Colour coding: hydrogen white, carbon grey, oxygen red, ni-
trogen blue and aluminium pink.
which is consistent with formation of an associative adduct.
Adduct formation was exothermic with a molar standard en-
À1
°
À1
A
A
thalpy change (DrH ) of À38 kJmol . This led to a molar stan-
a free activation energy (Dr G ) of 58 kJmol and formation
À1
A
A
dard free energy change (DrG ) of À13 kJmol . So the adduct
of the carbonato complex had free energy (DrG ), enthalpy
A
A
formation is exergonic at standard conditions. Significantly
therefore, DFT calculations indicate that [Al(acen)]2O is a better
Lewis base than Lewis acid with respect to carbon dioxide.
The reactivity of the adduct was further investigated by DFT
calculations, which indicated that carbon dioxide can insert
into one of the aluminium m-oxygen bonds to form a carbonato
bridged bimetallic aluminium complex through the transition
state shown in Figure 1. The energetics of this incorporation at
standard conditions and with respect to the adduct state were
calculated as À2 kJmolÀ1, À13 kJmolÀ1, and À38 JmolÀ1 KÀ1,
(DrH ), and entropy changes (DrS ) of 6 kJmolÀ1, À7 kJmolÀ1
and À45 JmolÀ1 KÀ1 at standard conditions relative to the initial
carbon dioxide adduct.
There is no precedent for carbon dioxide insertion into the
bridging metal-oxygen-metal bond of an oxo-bridged bis-met-
al(salen) complex or into an oxo-bridged aluminium complex.[9]
However, the insertion of carbon dioxide into metal-oxygen
bonds in general is well-known.[10] The insertion of carbon di-
oxide into complex 3c (Scheme 2) was experimentally con-
firmed when complex 3c as a solid was treated with carbon di-
oxide at 50–1008C and 50 bar pressure. The resulting
carbonato complex 5 was found to be metastable, slowly loos-
ing carbon dioxide to reform complex 3c. Nevertheless, 5 was
characterised by 1H NMR, infrared and mass spectrometry.[7]
The 13C labelled complex was also prepared in situ by passing
13C labelled carbon dioxide into a CDCl3 solution of complex
A
A
for free energy (DrG ), enthalpy (DrH ), and entropy change
A
(DrS ), respectively. Hence, there is a thermodynamic driving
force for the formation of [Al(acen)]2CO3 from complex 4 and
CO2. The transition state for the incorporation was located and
°
A
characterised. The free activation energy (Dr G ) was only
58 kJmolÀ1 with respect to the adduct state. Hence, the model-
ling revealed that the incorporation is feasible and
facile at standard conditions. Since it was shown that
the free energy changes were nearly pressure and
temperature independent,[7] we can assume that the
free energy changes also apply to CO2 incorporations
under the experimental conditions specified in
Scheme 1.
The carbonato bridge formed through CO2 inser-
tion can be bound to the aluminium atoms in four
different ways (A–D), as indicated in Figure 2. The
m2k1:k2-CO3 complex, A, reflects the product state, de-
picted in Figure 1. This complex was calculated to
isomerize to complexes B and C.[7] For the m2k1:k1-
CO3 complex, D, a stable minimum could not be lo-
cated as the compound transformed to C upon ge-
A
ometry optimization. The free energy changes (DrG ),
relative to parent complex 4 and non-interacting Scheme 2. Formation of carbonato bridged bis-aluminum(salen) complex 5c.
ChemSusChem 2016, 9, 791 – 794
www.chemsuschem.org
792
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
Two control experiments were also carried out. Complex 3c
and tetrabutylammonium bromide were used as catalysts for
the reaction of substrates 6 at room temperature and one bar
pressure, and 100% retention of the epoxide stereochemistry
was observed.[7] Cis-6 was also treated with one equivalent of
carbonato complex 5 at 508C under 1 bar nitrogen in the ab-
sence of carbon dioxide. Exclusive formation of trans-7 was ob-
served, indicating that this reaction proceeded with complete
inversion of epoxide stereochemistry.
The stereochemical consequences at the more hindered
carbon atom of the epoxide was investigated using (R)-styrene
oxide [(R)-1c] and (S)-glycidyl phenyl ether [(S)-1d] at 508C
and 10 bar carbon dioxide pressure. In the presence of tetrabu-
tylammonium bromide, cyclic carbonate synthesis from either
enantiomerically pure epoxide proceeded with 100% retention
of the epoxide stereochemistry as determined by chiral
HPLC.[7] However, in the absence of tetrabutylammonium bro-
mide, (R)-1c gave a 97:3 ratio of (R)- and (S)-2c and (S)-1d
gave a 99.5:0.5 ratio of (S)- and (R)-2d indicating that some
cleavage of the more substituted carbon-oxygen bond of the
epoxide occurred.
Figure 3. 13C NMR spectrum of complex 3c in CDCl3 in the presence of 13CO2
(top) in comparison to parent 3c in CDCl3 (bottom).
3c at À788C, confirming that the insertion is feasible and
facile, as predicted by DFT calculations. The resulting 13C NMR
spectrum (Figure 3) showed a resonance at 165.60 ppm, char-
acteristic of a carbonato bridged complex,[11] and a shift in the
other resonances relative to 3c. 13C NMR exchange spectrosco-
py (EXSY) experiments showed that complex 5 was in equilibri-
um with 3c and magnetisation transfer was observed between
the carbonato group of complex 5 (165.60 ppm) and dissolved
13CO2 (124.83 ppm).[7]
A catalytic cycle for cyclic carbonate synthesis catalysed by
complexes 3 in the absence of tetrabutylammonium bromide
consistent with the above results is shown in Scheme 4.
Carbon dioxide first inserts into the aluminium-oxygen bond
of complex 3 to give carbonato complex 5. Subsequent com-
plexation of the epoxide gives adduct 8 which undergoes in-
tramolecular epoxide ring-opening at the less hindered carbon
atom of the epoxide by the carbonate with inversion of config-
uration to give intermediate 9. From complex 9 two pathways
lead to the cyclic carbonate. In path 1, the alkoxide of complex
9 cyclises onto the carbonyl of the carbonato group forming
the cyclic carbonate and regenerating complex 3. This path-
way involves a single inversion of configuration at the epoxide
and requires no additional carbon dioxide beyond formation
of 5. Hence it accounts for the inversion of epoxide configura-
tion observed when complex 5 is used to induce the reaction
in the absence of carbon dioxide.
To investigate the relevance of carbonato complex 5 to the
catalytic cycle, it was dissolved together with tetrabutylammo-
nium bromide in epoxide 1b under nitrogen at room tempera-
1
ture. H NMR and GC analysis[7] confirmed the formation of car-
bonate 2b, showing that complex 5 is catalytically active and
that the carbonato bridge acts as a source of carbon dioxide.
The reaction mechanism in the absence of tetrabutylammo-
nium bromide was then studied using monodeuterated cis-
and trans-decylene oxide[12] 6 as substrates at 508C and 10 bar
carbon dioxide pressure to investigate the stereochemistry of
the reaction (Scheme 3). The results indicate that the reaction
proceeds mainly with retention of epoxide stereochemistry,
giving a 3:1 ratio of retention to inversion products. Formation
of both cis- and trans-cyclic carbonate 7 in the absence of tet-
rabutylammonium bromide indicates that two different reac-
tion pathways are operative.
Pathway 2 involves a second carbon dioxide insertion into
the aluminium alkoxide bond[13] of complex 9 to form bis-
carbonato complex 10. The carbonato groups in complex 10
will be good nucleophiles and good leaving groups, allowing
formation of the cyclic carbonate to occur by a second intra-
molecular substitution reaction with a second inver-
sion of stereochemistry, giving overall retention of
epoxide stereochemistry. Formation of complex 10
also provides an explanation of the partial racemiza-
tion by SN1 type cleavage of the carbonato group at-
tached to the more substituted carbon atom.
Additional DFT calculations were carried out to
support the routes shown in Scheme 4.[7] Thus, carbo-
nato bridged complexes were found to coordinate
epoxides as required for the formation of species 8.
The six-coordinate aluminium in complex 8 was
found to exist in a cis-b configuration,[14] which ena-
bles the intramolecular rearrangement to form com-
plex 9.
Scheme 3. Cyclic carbonate synthesis from deuterated epoxides 6.
ChemSusChem 2016, 9, 791 – 794
www.chemsuschem.org
793
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
Chem. 2015, 17, 5114–5130; d) Q. Liu, L. Wu, R. Jackstell, M.
Beller, Nat. Commun. 2015, 6, 5933 (DOI: 10.1038/
ncomms6933); e) M. Poliakoff, W. Leitner, E. S. Streng, Fara-
day Discuss. 2015, 183, 9–17; f) New and Future Develop-
ments in Catalysis: Activation of Carbon Dioxide (Ed.: S. L.
Suib), Elsevier, Amsterdam, 2013, pp. 1–644; g) M. Aresta,
A. Dibenedetto, A. Angelini, Chem. Rev. 2014, 114, 1709–
1742; h) Carbon Dioxide Utilisation: Closing the Carbon Cycle
(Eds.: P. Styring, A. Quadrelli, K. Armstrong), Elsevier, Amster-
dam, 2015, pp. 1–311.
[2] B. Schäffner, F. Schäffner, S. P. Verevkin, Chem. Rev. 2010,
110, 4554–4581.
[3] a) W. J. Peppel, Ind. Eng. Chem. 1958, 50, 767–770; b) S. Fu-
kuoka, M. Kawamura, K. Komiya, M. Tojo, H. Hachiya, K. Ha-
segawa, M. Aminaka, H. Okamoto, I. Fukawa, S. Konno,
Green Chem. 2003, 5, 497–507; c) W. E. Evans, M. L. Hessm,
M. Matusz, E. Van Kruchten, M. G. Andre, (Shell International
Research), WO2009140318, 2009; d) J. G. J. Beckers, E. Van
der Heide, G. M. M. Van Kessel, J. P. Lange, (Shell Internation-
al Research), WO2005051939, 2005; e) T. M. Nisbet, G. G. Va-
porciyan, (Shell International Research), WO2008128956,
2008.
[4] a) M. North, R. Pasquale, C. Young, Green Chem. 2010, 12,
1514–1539; b) C. Martín, G. Fiorani, A. W. Kleij, ACS Catal.
2015, 5, 1353–1370; c) A. C. Kathalikkattil, R. Babu, J.
Tharun, R. Roshan, D.-W. Park, Catal. Surv. Asia 2015, 19,
223–235; d) B. Yu, L.-N. He, ChemSusChem 2015, 8, 52–62;
e) M. Cokoja, M. E. Wilhelm, M. H. Anthofer, W. A. Herrmann,
F. E. Kühn, ChemSusChem 2015, 8, 2436–2454; f) B.-H. Xu, J.-
Q. Wang, J. Sun, Y. Huang, J.-P. Zhang, X.-P. Zhang, S.-J.
Zhang, Green Chem. 2015, 17, 108–122.
Scheme 4. Proposed mechanism for the synthesis of cyclic carbonates from epoxides
and CO2 catalysed by complexes 3 in the absence of Bu4NBr. The salen ligand attached
to each aluminium ion is not shown for clarity.
[5] J. A. Castro-Osma, M. North, X. Wu, Chem. Eur. J. 2014, 20,
15005–15008.
[6] a) J. W. Comerford, I. D. V. Ingram, M. North, X. Wu, Green Chem. 2015,
17, 1966–1987; b) S. Elmas, M. A. Subhani, M. Harrer, W. Leitner, J. Sun-
dermeyer, T. E. Müller, Catal. Sci. Technol. 2014, 4, 1652–1657.
[7] See Supporting Information.
[8] M. North, C. Young, Catal. Sci. Technol. 2011, 1, 93–99.
[9] An aluminium carbonate complex has been reported by reaction of
a monometallic iminopyridine aluminium hydroxide with carbon diox-
ide: T. W. Myers, L. A. Berben, Chem. Commun. 2013, 49, 4175–4177.
[10] a) X. Yin, J. R. Moss, Coord. Chem. Rev. 1999, 181, 27–59; b) W. Leitner,
Coord. Chem. Rev. 1996, 153, 257–284; c) S. Elmas, M. A. Subhani, H.
Vogt, W. Leitner, T. E. Müller, Green Chem. 2013, 15, 1356–1360.
[11] H. S. Kim, J. J. Kim, B. G. Lee, O. S. Jung, H. G. Jang, S. O. Kang, Angew.
Chem. Int. Ed. 2000, 39, 4096–4098; Angew. Chem. 2000, 112, 4262–
4264.
In conclusion, the combination of theoretical and experi-
mental studies have allowed key features of the mechanism
of cyclic carbonate synthesis catalysed by bimetallic
aluminium(salen) complexes 3 in the absence of a cocatalyst
to be elucidated. The reaction proceeds by an unprecedented
carbonato bridged bimetallic aluminium(salen) complex which
has been characterized and shown to be catalytically active.
Acknowledgements
The research leading to these results has received funding from
the European Union Seventh Framework Programme FP7-NMP-
2012 under grant agreement number 309497.
[12] J. Martínez, J. A. Castro-Osma, A. Earlam, C. Alonso-Moreno, A. Otero, A.
Lara-Sµnchez, M. North, A. Rodríguez-DiØguez, Chem. Eur. J. 2015, 21,
9850–9862.
[13] For CO2 insertion into aluminum alkoxide bonds see: T. Aida, S. Inoue,
J. Am. Chem. Soc. 1983, 105, 1304–1309.
[14] K. Matsumoto, B. Saito, T. Katsuki, Chem. Commun. 2007, 3619–3627.
Keywords: aluminum catalyst · carbon dioxide fixation · cyclic
carbonate · density functional calculations · reaction pathway
[1] a) A. Otto, T. Grube, S. Schiebahn, D. Stolten, Energy Environ. Sci. 2015,
8, 3283–3297; b) G. Fiorani, W. Guo, A. W. Kleij, Green Chem. 2015, 17,
1375–1389; c) A. J. Martín, G. O. Larrazµbal, J. PØrez-Ramírez, Green
Received: December 18, 2015
Published online on March 31, 2016
ChemSusChem 2016, 9, 791 – 794
www.chemsuschem.org
794
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim