Citation: Brüggemann, D.; Machat,
M.R.; Schomäcker, R.; Heshmat, M.
Catalytic Ring-Opening
Polymerisation of Cyclic Ethylene
Carbonate: Importance of Elementary
Steps for Determining Polymer
Properties Revealed via DFT-MTD
Simulations Validated Using Kinetic
Measurements. Polymers 2024,16, 136.
https://doi.org/10.3390/
polym16010136
Academic Editor: Beom Soo Kim
Received: 19 November 2023
Revised: 21 December 2023
Accepted: 24 December 2023
Published: 31 December 2023
Copyright: © 2023 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
polymers
Article
Catalytic Ring-Opening Polymerisation of Cyclic Ethylene
Carbonate: Importance of Elementary Steps for Determining
Polymer Properties Revealed via DFT-MTD Simulations
Validated Using Kinetic Measurements
Daniel Brüggemann 1,2 , Martin R. Machat 2,3 , Reinhard Schomäcker 1and Mojgan Heshmat 3,*
1Institut für Chemie—Technische Chemie, Technische Universität Berlin, Straße des 17. Juni 124,
2Covestro Deutschland AG, Kaiser-Wilhelm-Alle 60, D-51373 Leverkusen, Germany
3Institute of Technical and Macromolecular Chemistry, CAT Catalytic Center, RWTH Aachen Universität,
Worringerweg 2, D-52074 Aachen, Germany
*Correspondence: mojgan.heshmat@wur.nl; Tel.: +31-31317481375
Abstract: The production of CO
2
-containing polymers is still very demanding in terms of controlling
the synthesis of products with pre-defined CO
2
content and molecular weight. An elegant way of
synthesising these polymers is via CO
2
-containing building blocks, such as cyclic ethylene carbonate
(cEC), via catalytic ring-opening polymerisation. However, to date, the mechanism of this reaction
and control parameters have not been elucidated. In this work, using DFT-metadynamics simula-
tions for exploiting the potential of the polymerisation process, we aim to shed more light on the
mechanisms of the interaction between catalysts (in particular, the catalysts K
3
VO
4
, K
3
PO
4
, and
Na
2
SnO
3
) and the cEC monomer in the propagation step of the polymeric chain and the occurring
CO
2
release. Confirming the simulation results via subsequent kinetics measurements indicates that,
depending on the catalyst’s characteristics, it can be attached reversibly to the polymeric chain during
polymerisation, resulting in a defined lifetime of the activated polymer chain. The second anionic
oxygen of the catalyst can promote the catalyst’s transfer to another electrophilic cEC monomer,
terminating the growth of the first chain and initiating the propagation of the new polymer chain.
This transfer reaction is an essential step in controlling the molecular weight of the products.
Keywords: DFT-metadynamics; kinetics; ring-opening polymerisation; cyclic ethylene carbonate;
biodegradable polymers; aliphatic cyclic carbonates
1. Introduction
Increasing concern about climate change, connected to the use of (diminishing) fossil
resources, necessitates the development of new, sustainable technologies from renewable
sources for the generation of chemicals, energy, and materials [
1
–
5
]. To this end, cyclic
ethylene carbonate (cEC in short), due to ecological and economic reasons, and its ready
availability from ethylene oxide and carbon dioxide, seems to be a very attractive monomer
and reagent for polyether carbonate production [
6
,
7
]. The obtained aliphatic polycarbonates
are of interest due to their new and valuable applications in biodegradable and biocompat-
ible materials. Hence, the presence of carbonate functional groups in the polymer chain
facilitates the depolymerisation of aliphatic polycarbonates. Interestingly, in ring-opening
polymerisation (ROP) of cEC, the polymers formed may have lower densities than the
monomers (volume expansion may accompany polymerisation), which can be beneficial
for industrial applications [
8
–
11
]. The catalysed ROP of cEC is a low-pressure technology,
representing a technical advantage compared to the direct copolymerisation of ethylene
oxide (EO) and CO
2
[
12
–
19
]. Both thermodynamic and kinetic factors are crucial for the
ROP of cEC. It is worth noting that the maximum temperature for the homopolymerisation
Polymers 2024,16, 136. https://doi.org/10.3390/polym16010136 https://www.mdpi.com/journal/polymers
Polymers 2024,16, 136 2 of 24
of CEC is below 25
◦
C, but polymerisations have also occurred at over 100
◦
C [
17
]. At
high polymerisation temperatures, the resultant polymers’ repeat units are a mixture of
monomeric units (carbonate units) and the corresponding oxide units, meaning that CO
2
release occurs during the polymerisation. The release of the CO
2
makes the overall reaction
∆
S positive, so this polymerisation becomes thermodynamically possible only at high
temperatures. In cEC polymerisation, the purpose is not to completely avoid CO
2
release,
but keeping CO
2
to some extent in the final polymer product is advantageous [
17
]. To over-
come the kinetic stability of cEC and polymerise the cEC, modern catalysts and catalytic
processes have been developed in the last few decades to achieve the ROP quickly. Notably,
an efficient catalyst must conduct the reaction in such a way that CO
2
release is minimised
during the polymerisation of cEC. A series of recently investigated catalysts approved for
cEC polymerisation is of particular industrial interest, i.e., they are structurally simple,
commercially available, and efficient catalysts. The relatively high conversion values and
CO
2
content percentages that were experimentally obtained and approved by the Covestro
company [
4
,
5
,
20
] in the ROP of cEC confirm that these catalysts are exciting candidates
with the potential for obtaining a more fundamental understanding with kinetic studies
and computational methods. Among the examined series of 30 different catalysts, we
selected K
3
VO
4
, K
3
PO
4
, and Na
2
SnO
3
as representative examples for the current study.
The cyclic carbonate conversion reactions cover carbonate formation and ether linkages in
the final product, along with some CO2release (Scheme 1).
Polymers 2024, 16, x FOR PEER REVIEW 2 of 27
technology, representing a technical advantage compared to the direct copolymerisation
of ethylene oxide (EO) and CO
2
[12–19]. Both thermodynamic and kinetic factors are cru-
cial for the ROP of cEC. It is worth noting that the maximum temperature for the homo-
polymerisation of CEC is below 25 °C, but polymerisations have also occurred at over 100
°C [17]. At high polymerisation temperatures, the resultant polymers’ repeat units are a
mixture of monomeric units (carbonate units) and the corresponding oxide units, meaning
that CO
2
release occurs during the polymerisation. The release of the CO
2
makes the over-
all reaction ΔS positive, so this polymerisation becomes thermodynamically possible only
at high temperatures. In cEC polymerisation, the purpose is not to completely avoid CO
2
release, but keeping CO
2
to some extent in the final polymer product is advantageous [17].
To overcome the kinetic stability of cEC and polymerise the cEC, modern catalysts and
catalytic processes have been developed in the last few decades to achieve the ROP
quickly. Notably, an efficient catalyst must conduct the reaction in such a way that CO
2
release is minimised during the polymerisation of cEC. A series of recently investigated
catalysts approved for cEC polymerisation is of particular industrial interest, i.e., they are
structurally simple, commercially available, and efficient catalysts. The relatively high
conversion values and CO
2
content percentages that were experimentally obtained and
approved by the Covestro company [4,5,20] in the ROP of cEC confirm that these catalysts
are exciting candidates with the potential for obtaining a more fundamental understand-
ing with kinetic studies and computational methods. Among the examined series of 30
different catalysts, we selected K
3
VO
4
, K
3
PO
4,
and Na
2
SnO
3
as representative examples for
the current study. The cyclic carbonate conversion reactions cover carbonate formation
and ether linkages in the final product, along with some CO
2
release (Scheme 1).
Scheme 1. Catalytic ROP of cEC and exemplary molecular structure of the respective polyether
carbonate polyol formed.
For the selected catalysts, the cations are either equal (K
3
PO
4
and K
3
VO
4
) or, if they
are different, no significant effect from the cationic counterpart is observed in simulations
and experiments (Na
2
SnO
3
vs. K
2
SnO
3
). Hence, the main difference in catalytic perfor-
mance for the selected catalysts is caused by the anionic part, and we have considered the
anionic pathway (Scheme 2) for the current study. Anionic ROP (Scheme 2) begins with a
nucleophilic attack by an anionic initiator, such as the anionic counterpart of the catalyst,
or can be generated, e.g., via proton abstraction from starter alcohol. The nucleophile can
either attack the carbonyl carbon or the methylene carbon of the cyclic carbonate [21–23].
Scheme 1. Catalytic ROP of cEC and exemplary molecular structure of the respective polyether
carbonate polyol formed.
For the selected catalysts, the cations are either equal (K
3
PO
4
and K
3
VO
4
) or, if they
are different, no significant effect from the cationic counterpart is observed in simulations
and experiments (Na
2
SnO
3
vs. K
2
SnO
3
). Hence, the main difference in catalytic perfor-
mance for the selected catalysts is caused by the anionic part, and we have considered the
anionic pathway (Scheme 2) for the current study. Anionic ROP (Scheme 2) begins with a
nucleophilic attack by an anionic initiator, such as the anionic counterpart of the catalyst,
or can be generated, e.g., via proton abstraction from starter alcohol. The nucleophile can
either attack the carbonyl carbon or the methylene carbon of the cyclic carbonate [21–23].
Polymers 2024, 16, x FOR PEER REVIEW 2 of 27
technology, representing a technical advantage compared to the direct copolymerisation
of ethylene oxide (EO) and CO
2
[12–19]. Both thermodynamic and kinetic factors are cru-
cial for the ROP of cEC. It is worth noting that the maximum temperature for the homo-
polymerisation of CEC is below 25 °C, but polymerisations have also occurred at over 100
°C [17]. At high polymerisation temperatures, the resultant polymers’ repeat units are a
mixture of monomeric units (carbonate units) and the corresponding oxide units, meaning
that CO
2
release occurs during the polymerisation. The release of the CO
2
makes the over-
all reaction ΔS positive, so this polymerisation becomes thermodynamically possible only
at high temperatures. In cEC polymerisation, the purpose is not to completely avoid CO
2
release, but keeping CO
2
to some extent in the final polymer product is advantageous [17].
To overcome the kinetic stability of cEC and polymerise the cEC, modern catalysts and
catalytic processes have been developed in the last few decades to achieve the ROP
quickly. Notably, an efficient catalyst must conduct the reaction in such a way that CO
2
release is minimised during the polymerisation of cEC. A series of recently investigated
catalysts approved for cEC polymerisation is of particular industrial interest, i.e., they are
structurally simple, commercially available, and efficient catalysts. The relatively high
conversion values and CO
2
content percentages that were experimentally obtained and
approved by the Covestro company [4,5,20] in the ROP of cEC confirm that these catalysts
are exciting candidates with the potential for obtaining a more fundamental understand-
ing with kinetic studies and computational methods. Among the examined series of 30
different catalysts, we selected K
3
VO
4
, K
3
PO
4,
and Na
2
SnO
3
as representative examples for
the current study. The cyclic carbonate conversion reactions cover carbonate formation
and ether linkages in the final product, along with some CO
2
release (Scheme 1).
Scheme 1. Catalytic ROP of cEC and exemplary molecular structure of the respective polyether
carbonate polyol formed.
For the selected catalysts, the cations are either equal (K
3
PO
4
and K
3
VO
4
) or, if they
are different, no significant effect from the cationic counterpart is observed in simulations
and experiments (Na
2
SnO
3
vs. K
2
SnO
3
). Hence, the main difference in catalytic perfor-
mance for the selected catalysts is caused by the anionic part, and we have considered the
anionic pathway (Scheme 2) for the current study. Anionic ROP (Scheme 2) begins with a
nucleophilic attack by an anionic initiator, such as the anionic counterpart of the catalyst,
or can be generated, e.g., via proton abstraction from starter alcohol. The nucleophile can
either attack the carbonyl carbon or the methylene carbon of the cyclic carbonate [21–23].
Scheme 2. The anionic ROP of cyclic carbonates (considered in this work).
Polymers 2024,16, 136 3 of 24
In this work, we aim to understand the details of the interaction between the catalysts
and cEC monomer, employing DFT-MTD simulations and kinetic measurements to shed
more light on the elementary steps for the CO
2
release mechanism and differences between
various considered catalysts. The main questions that we address are the consumption
of the entire starter molecule in the initiation step, which pathways are dominant in
the propagation step throughout the interaction between catalyst and monomer, and the
possible effects on the CO2content and molecular weight of the polymer are.
Recent findings obtained using advanced ab initio molecular dynamics (AIMD) sim-
ulations showed that the possibility of identification of alternative pathways of various
complex reactions could be challenged in a dynamic picture [
24
–
29
]. This motivated us
to investigate the ROP of cEC in the presence of the catalysts, as mentioned above, us-
ing DFT-metadynamics simulations as the computational method of choice. We employ
metadynamics simulations to calculate the free energy surfaces (FESs) for the ROP and
CO
2
release mechanisms at a defined temperature. Simulation of a free energy landscape
rather than a zero Kelvin potential energy surface allows the incorporation of entropic
contributions due to the flexibility of motion of free molecules in the reaction environ-
ment. According to previous studies, the entropic contributions affect the formation of
molecular complexes from small molecules and their respective transition state barriers of
formation [30–33].
Understanding the underlying mechanisms and identifying the critical steps of catal-
ysed ring opening and chain growth of cyclic carbonates and CO
2
release pathways is the
basis for computational chemistry-assisted research, leading to improved catalyst struc-
tures and performance. Using the results and insights obtained in this work, novel catalyst
candidates can be rationally designed for better performance and process optimisation.
To make this possible, comparing the computer-calculated values with experimental
ones is an elegant approach. By determining the kinetics, it is possible to understand
the reaction better. Understanding the kinetics also opens up the control possibilities of
ring-opening polymerisation. The mechanism of the ROP is not fully clarified, so it is still
unclear why the basicity has an influence and why specific catalysts no longer react during
the reaction [
6
]. These questions are, therefore, the central focus of this paper, and with
them, the clarification of the mechanism of ring-opening polymerisation.
These polymers give access to a wide-ranging product portfolio. For example, using
a hydrophobic initiator, functional non-ionic surfactants can be produced that are both
environmentally friendly due to their excellent biodegradability and can be recycled. For
example, these surfactants have already been proven to purify the microplastic contam-
ination of water [
34
]. The production of low-molecular polymers (700–1200 g/mol) for
the production of surfactants was the aim of the funded project on which this work is
based. However, the findings of this work can also be applied directly to higher degrees
of polymerisation.
2. Materials and Methods
2.1. Computational Details
All DFT calculations, including geometry optimisations and ab initio molecular dy-
namics (AIMD) simulations, were performed using the CP2K program [
35
] with the
Gaussian and plane-wave (GPW) method. The valence orbitals were expanded in the
DZVP-MOLOPT Gaussian basis set in combination with Goedecker, Teter, and Hutter
pseudopotentials and were used with a plane wave cutoff energy of 280 Ry. We used the
PBE density functional [
36
] augmented with Grimme D3 dispersion correction [
37
]. The
criterion for self-consistent field convergence was set to 5.0
×
10
−7
. The AIMD simulations
were done in the NVT ensemble, with the temperature controlled by a CSVR (canonical
sampling through velocity rescaling) thermostat, set at various temperatures (413, 423,
433, and 443 K), and a period of 500 fs. The MD time step was 0.5 fs. To investigate the
reaction mechanism at finite temperature, characterise the reaction pathway, and identify
the transition state region between reactant and product states, we performed metady-
Polymers 2024,16, 136 4 of 24
namics simulations using the PLUMED 2.8 plugin in combination with CP2K [
38
–
40
]. The
simulations were initiated from the optimised molecular structures and conducted until
several transition state re-crossing events could be observed. To prevent sampling from
unnecessary regions of the FES, harmonic walls with force constant K = 250 kcal/molÅ
2
were used for CVs. The Gaussian bias potentials were added every 100 steps. Multiple test
simulations were run to set the computational parameters, including the number and type
of CVs, the height and width of Gaussian bias potentials, and quadratic walls for each part
of the reaction mechanism.
We performed our simulations on a molecular cluster schematically shown in Scheme 3.
As depicted in Scheme 3, the molecular cluster consists of six cEC monomers, one alcohol
molecule (F = 1 for simplification), and a catalyst molecule. The typical catalyst molecule
includes the anionic (MO43−or MO32−) and cationic (A+) counterions.
Polymers 2024, 16, x FOR PEER REVIEW 4 of 27
reaction mechanism at finite temperature, characterise the reaction pathway, and identify
the transition state region between reactant and product states, we performed metady-
namics simulations using the PLUMED 2.8 plugin in combination with CP2K [38–40]. The
simulations were initiated from the optimised molecular structures and conducted until
several transition state re-crossing events could be observed. To prevent sampling from
unnecessary regions of the FES, harmonic walls with force constant K = 250 kcal/molÅ
2
were used for CVs. The Gaussian bias potentials were added every 100 steps. Multiple test
simulations were run to set the computational parameters, including the number and type
of CVs, the height and width of Gaussian bias potentials, and quadratic walls for each part
of the reaction mechanism.
We performed our simulations on a molecular cluster schematically shown in
Scheme 3. As depicted in Scheme 3, the molecular cluster consists of six cEC monomers,
one alcohol molecule (F = 1 for simplification), and a catalyst molecule. The typical catalyst
molecule includes the anionic (MO
43−
or MO
32−
) and cationic (A
+
) counterions.
Scheme 3. The considered molecular cluster for simulations; in this model cluster, there are six
cEC molecules, one starter alcohol, and one typical catalyst molecule.
We considered six cEC monomers in our simulations since six cECs can make the first
shell interact with the catalyst with a reasonable computational cost for AIMD calcula-
tions.
In fact, we investigated the interaction between the catalyst anion and cEC mono-
mer, and a significant number of molecules participating in reaction mechanisms is one
of each species. To estimate the interactions between cEC monomers, ROH starter, and
catalyst–anion, we have probed the variation of average H-bond distances between ROH
and six surrounding cEC monomers (ROH…O(cEC)) in the first solvation shell through
an unbiased equilibrium MD simulation in Figure S1A in SI. The ROH-cEC distances are
compared with the ROH-O(anion-CAT). The main interactions involved are those be-
tween ROH and the anion (shorter distances) that result in proton transfer in the initiation
step of polymerisation. The interaction between ROH and O-cEC (according to distances)
is weaker and O-cEC mainly interacts with the surrounding cations. Hence, ROH does not
exist that long in the reaction medium and is deprotonated by the anion of the catalyst.
After the proton abstraction, the main H bond is between the protonated catalyst anion
and the surrounding molecules (cEC monomers and alkoxide), which can cover the pro-
tonated anion sufficiently. The reaction temperature in the experiment is above 150 °C,
and the catalysts are expected to be solubilised entirely under these conditions. Addition-
ally, lab results indicate that the catalyst cannot be separated adequately via filtration after
the reaction. Both facts conclude that considering homogenous reaction conditions is a
valid starting point. The collective variables (CVs) were the relevant distances in each re-
action step. We used VMD 1.9.4 software to visualise the MD trajectories and snapshots
[41].
Scheme 3. The considered molecular cluster for simulations; in this model cluster, there are six cEC
molecules, one starter alcohol, and one typical catalyst molecule.
We considered six cEC monomers in our simulations since six cECs can make the first
shell interact with the catalyst with a reasonable computational cost for AIMD calculations.
In fact, we investigated the interaction between the catalyst anion and cEC monomer,
and a significant number of molecules participating in reaction mechanisms is one of
each species. To estimate the interactions between cEC monomers, ROH starter, and
catalyst–anion
, we have probed the variation of average H-bond distances between ROH
and six surrounding cEC monomers (ROH
· · ·
O(cEC)) in the first solvation shell through
an unbiased equilibrium MD simulation in Figure S1A in SI. The ROH-cEC distances are
compared with the ROH-O(anion-CAT). The main interactions involved are those between
ROH and the anion (shorter distances) that result in proton transfer in the initiation step
of polymerisation. The interaction between ROH and O-cEC (according to distances) is
weaker and O-cEC mainly interacts with the surrounding cations. Hence, ROH does not
exist that long in the reaction medium and is deprotonated by the anion of the catalyst.
After the proton abstraction, the main H bond is between the protonated catalyst anion and
the surrounding molecules (cEC monomers and alkoxide), which can cover the protonated
anion sufficiently. The reaction temperature in the experiment is above 150
◦
C, and the
catalysts are expected to be solubilised entirely under these conditions. Additionally, lab
results indicate that the catalyst cannot be separated adequately via filtration after the
reaction. Both facts conclude that considering homogenous reaction conditions is a valid
starting point. The collective variables (CVs) were the relevant distances in each reaction
step. We used VMD 1.9.4 software to visualise the MD trajectories and snapshots [41].
2.2. Chemicals Used and Suppliers
The following chemicals with the indicated purities were used for the experiments.
Ethylene carbonate (cEC, 99%) was obtained from Alfa Aesar (Haverhill, MA, USA). Ethy-
lene glycol (99%) was obtained from Roth (Karlsruhe, Germany). Methanol (99.9%) was
obtained from VWR (Dresden, Germany). Dichlormethan-D2 (99.5%) was obtained from
Polymers 2024,16, 136 5 of 24
Roth (Karlsruhe, Germany). Tichlormethan-D1 (99.8%) was obtained from Roth (Karlsruhe,
Germany). Potassium phosphate tribasic anhydrous (99.5%) was obtained from VWR
(Dresden, Germany). Sodium orthovanadate (V) (99.9%) was obtained from Alfa Aesar
(Haverhill, MA, USA) and Potassium tin (IV) oxide trihydrate (95%) was obtained from
Alfa Aesar (Haverhill, MA, USA).
2.3. Experimental Method
All reactions proceeded at the same conditions using a 50 mL reactor with a heatable
jacket. The reactor has a GL25 opening for measurement with an in situ IR probe and
two feed valves for operation under inert gas. The reactor was heated with a Haake F6
thermostat. Silicone oil AP200 was chosen as the operating fluid for the thermostat. The
IKA RCT Basic magnetic stirrer was used for mixing. A Mettler Toledo Reakt IR 15 was
used for the in situ measurements. The spectrometer has a Si probe connected to the
spectrometer via AgX 9.5 mm
×
1.5 mm fibre optics. The measured wavenumber range
is from 4000 cm
−1
to 800 cm
−1
with a resolution of 4 cm
−1
. The reactor was brought to
experimental temperature according to Tables S2 and S6 (see Supporting Information), and
the in situ IR probe was supplied with fresh nitrogen (l) for cooling the detector. The blank
value of the IR probe was measured with the empty reactor, in which only the magnetic
stirring bar was present. After measuring the blank spectrum, cyclic ethylene carbonate
(20 g, 1 eq.), which had been preheated in a 60
◦
C water bath, was added to the reactor, the
stirrer was set to 600 RPM, and the in situ measurement was started At least 10 min waiting
time was used to obtain a baseline measurement for cEC. Ethylene glycol (10 mol%, 0.1 eq.)
and catalyst (1 mol%, 0.01 eq.) were weighed beforehand and added quickly to the reactor.
As the reaction progressed, the changes in the characteristic bands were monitored in the
Software IC IR 7 (see Figure 1): 1800–1760 cm
−1
[
42
] for cEC (C-O-C plugin vibrations,
strained system) [
42
], 1750–1700 cm
−1
[
42
] for CO
2
(linear unstrained system) [
42
] and the
ether band at 1260 cm
−1
[
42
], (these are the expected signals for a nascent poly(ethylene
ether carbonate)) [
42
–
45
]. The reaction was allowed to proceed until the band typical for
cEC at 1800–1750 cm−1was diminished entirely.
Polymers 2024, 16, x FOR PEER REVIEW 6 of 27
2.2. Chemicals Used and Suppliers
The following chemicals with the indicated purities were used for the experiments.
Ethylene carbonate (cEC, 99%) was obtained from Alfa Aesar (Haverhill, MA, USA). Eth-
ylene glycol (99%) was obtained from Roth (Karlsruhe, Germany). Methanol (99.9%) was
obtained from VWR (Dresden, Germany). Dichlormethan-D2 (99.5%) was obtained from
Roth (Karlsruhe, Germany). Tichlormethan-D1 (99.8%) was obtained from Roth (Karls-
ruhe, Germany). Potassium phosphate tribasic anhydrous (99.5%) was obtained from
VWR (Dresden, Germany). Sodium orthovanadate (V) (99.9%) was obtained from Alfa
Aesar (Haverhill, MA, USA) and Potassium tin (IV) oxide trihydrate (95%) was obtained
from Alfa Aesar (Haverhill, MA, USA).
2.3. Experimental Method
All reactions proceeded at the same conditions using a 50 mL reactor with a heatable
jacket. The reactor has a GL25 opening for measurement with an in situ IR probe and two
feed valves for operation under inert gas. The reactor was heated with a Haake F6 ther-
mostat. Silicone oil AP200 was chosen as the operating fluid for the thermostat. The IKA
RCT Basic magnetic stirrer was used for mixing. A Mettler Toledo Reakt IR 15 was used
for the in situ measurements. The spectrometer has a Si probe connected to the spectrom-
eter via AgX 9.5 mm × 1.5 mm fibre optics. The measured wavenumber range is from 4000
cm–1 to 800 cm–1 with a resolution of 4 cm–1. The reactor was brought to experimental tem-
perature according to Tables S2 and S6 (see Supporting Information), and the in situ IR
probe was supplied with fresh nitrogen (l) for cooling the detector. The blank value of the
IR probe was measured with the empty reactor, in which only the magnetic stirring bar
was present. After measuring the blank spectrum, cyclic ethylene carbonate (20 g, 1 eq.),
which had been preheated in a 60 °C water bath, was added to the reactor, the stirrer was
set to 600 RPM, and the in situ measurement was started At least 10 min waiting time was
used to obtain a baseline measurement for cEC. Ethylene glycol (10 mol%, 0.1 eq.) and
catalyst (1 mol%, 0.01 eq.) were weighed beforehand and added quickly to the reactor. As
the reaction progressed, the changes in the characteristic bands were monitored in the
Software IC IR 7 (see Figure 1): 1800–1760 cm–1 [42] for cEC (C-O-C plugin vibrations,
strained system) [42], 1750–1700 cm–1 [42] for CO2 (linear unstrained system) [42] and the
ether band at 1260 cm–1 [42], (these are the expected signals for a nascent poly(ethylene
ether carbonate)) [42–46]. The reaction was allowed to proceed until the band typical for
cEC at 1800–1750 cm–1 was diminished entirely.
Figure 1. Example 3D spectrum of the in situ measurement of the reaction at 600RPM 140 °C.
Figure 1. Example 3D spectrum of the in situ measurement of the reaction at 600RPM 140 ◦C.
3. Results and Discussion
Our previous study considered the ring-opening polymerisation of cEC in the presence
of the starter ROH, which begins with the proton abstraction from the starter ROH molecule
by the catalyst anion (step I [
46
]). The generated RO
−
attacks the C=O group of the first
neighbouring cEC. Then, the second generated alkoxide attacks the next cEC (step II). This
mechanism is shown in Figure 2, considering our model cluster. The rate-determining
step for this mechanism is the cEC ring opening (step II). The values of the
∆
G
=
for the
two steps are reported in Table 1.
Polymers 2024,16, 136 6 of 24
Polymers 2024, 16, x FOR PEER REVIEW 7 of 27
3. Results and Discussion
Our previous study considered the ring-opening polymerisation of cEC in the pres-
ence of the starter ROH, which begins with the proton abstraction from the starter ROH
molecule by the catalyst anion (step I [47]). The generated RO
−
attacks the C=O group of
the first neighbouring cEC. Then, the second generated alkoxide attacks the next cEC (step
II). This mechanism is shown in Figure 2, considering our model cluster. The rate-deter-
mining step for this mechanism is the cEC ring opening (step II). The values of the ΔG
≠
for
the two steps are reported in Table 1.
Figure 2. The considered mechanism of the ROP of cEC in a previous study. Step I, the proton ab-
straction by the catalyst anion and nucleophilic attack by the RO
−
to the carbonyl carbon of the cEC.
Step II, the ring opening of the second cEC via the second nucleophilic attack by alkoxide and ring
opening.
Table 1. The calculated ΔG
≠
for steps I and II (depicted in Figure 1).
Catalyst ΔG
≠
[kcal/mol]
I/II
Na
2
SnO
3
5.8/11.2
K
3
VO
4
5.5/7.7
K
3
PO
4
4.5/6.9
As shown in Table 1, the barriers are low in the presence of ROH (initiation). The
barriers, shown in Table 1 for step II, are correlated to the second cEC opening through a
nucleophilic attack by the first cleaved cEC. This step is the rate-determining step since
the proton abstraction (step I) has the lower barrier. However, considering that the con-
centration of the starter molecule is less than the cEC monomer at a particular stage in the
reaction, the starter is consumed, and only cEC and catalyst are present in the reaction
environment. Therefore, the reaction proceeds throughout the catalyst–cEC interaction
(propagation), and investigating this pathway is the primary purpose of the current study.
In this work, we have considered Na
2
SnO
3
, K
3
VO
4,
and K
3
PO
4
as the catalysts for the
cEC ring-opening polymerisation. From a previous study, we found that the M-O (central
atom-oxygen) bond length has a crucial impact on the catalyst basicity, i.e., a longer M-O
bond leads to more basicity (easier proton abstraction) but less nucleophilicity for a nu-
cleophilic attack to the CH
2
of the cEC ring. By the basicity, we refer to the basicity of the
Figure 2. The considered mechanism of the ROP of cEC in a previous study. Step I, the proton
abstraction by the catalyst anion and nucleophilic attack by the RO
−
to the carbonyl carbon of the
cEC. Step II, the ring opening of the second cEC via the second nucleophilic attack by alkoxide and
ring opening.
Table 1. The calculated ∆G=for steps Iand II (depicted in Figure 1).
Catalyst ∆G=[kcal/mol]
I/II
Na2SnO35.8/11.2
K3VO45.5/7.7
K3PO44.5/6.9
As shown in Table 1, the barriers are low in the presence of ROH (initiation). The
barriers, shown in Table 1for step II, are correlated to the second cEC opening through a
nucleophilic attack by the first cleaved cEC. This step is the rate-determining step since
the proton abstraction (step I) has the lower barrier. However, considering that the con-
centration of the starter molecule is less than the cEC monomer at a particular stage in the
reaction, the starter is consumed, and only cEC and catalyst are present in the reaction
environment. Therefore, the reaction proceeds throughout the catalyst–cEC interaction
(propagation), and investigating this pathway is the primary purpose of the current study.
In this work, we have considered Na
2
SnO
3
, K
3
VO
4
, and K
3
PO
4
as the catalysts for
the cEC ring-opening polymerisation. From a previous study, we found that the M-O
(central atom-oxygen) bond length has a crucial impact on the catalyst basicity, i.e., a longer
M-O bond leads to more basicity (easier proton abstraction) but less nucleophilicity for a
nucleophilic attack to the CH
2
of the cEC ring. By the basicity, we refer to the basicity of the
transition metal/main element oxides (K
3
PO
4
, K
3
VO
4
, Na
2
SnO
3
), inherently weaker bases
than KOH and NaOH. The order of decreasing the M-O bond distance of the considered
catalysts is Na
2
SnO
3
> K
3
VO
4
> K
3
PO
4
. These catalysts’ average M-O bond lengths are
1.939, 1.734, and 1.571 (in Å). The average O
· · ·
cations distances are 2.271, 2.617, and 2.604
(in Å) for Na
2
SnO
3
, K
3
VO
4
, and K
3
PO
4
, respectively. The differences in the molecular
structures of the catalysts severely affect their catalytic performance. Hence, a longer
M-O bond distance performs better in keeping a higher CO
2
content in the product. The
planar structure of the SnO
32−
catalyst anion causes complexation of the alkoxide anion
(RO
−
generated from ROH) to the Sn centre for a while during the initiation step. This
complexation delays the reaction in case of SnO
32−
vs. VO
43−
, 9.5 h vs. 5.0 h, respectively,
in experiment, which is in accordance with the higher barrier of the cEC ring opening for
SnO
32−
vs. VO
43−
(Table 2). For a detailed analysis we refer to the [
46
]. In addition, the
Polymers 2024,16, 136 7 of 24
cation’s lability also plays a role in more efficient catalytic performance. Especially for
the smaller anions such as PO
43−
, bigger cations perform better than smaller ones. The
∆
G
=
values of the ring opening decrease from H
+
to K
+
for phosphate anion, and in the
experiment, the CO
2
contribution increases with larger cations. This effect is due to the
longer anion
. . .
ation distance and less-covalent character of the anion-cation interaction,
leading to more accessibility of the catalyst anion in the reaction. For bigger anions
such as SnO
32−
and VO
43−
, switching between Na
+
and K
+
does not affect the catalyst
performance, and due to the stronger ionic character, the
∆
G
=
values and CO
2
-contribution
do not change with changing Na
+
to K
+
. The complete results of the cation influence on
the
∆
G
=
and the experimental CO
2
contribution are reported in Table 2. We note that
for H
3
PO
4
and Li
3
PO
4
catalysts in Table 2only minor conversions (due to the reduced
activity) were observed. However, the conversion was higher than 80 percent for the rest of
the catalysts.
Table 2. The activation free energies (
∆
G
=
) and CO
2
contents for various cations for the PO
43−
, VO
43
,
and SnO32−anions.
Catalyst ∆G=H+Abstraction
[kcal/mol]
∆G=cEC Opening
[kcal/mol]
Exp. CO2Content
[%]
H3PO422.3 50.0 3.0
Li3PO412.8 39.1 5.0
Na3PO47.6 5.2 11.0
K3PO44.2 4.0 15.0
Na2SnO36.2 11.2 26.0
K2SnO36.5 10.8 22.0
Na3VO44.8 4.5 26.0
K3VO45.8 7.7 25.0
Figure 3depicts the considered CO
2
release mechanism triggered by the interaction
between the catalyst anion and the CH2of the cEC ring.
Polymers 2024, 16, x FOR PEER REVIEW 8 of 27
transition metal/main element oxides (K
3
PO
4
, K
3
VO
4
, Na
2
SnO
3
), inherently weaker bases
than KOH and NaOH. The order of decreasing the M-O bond distance of the considered
catalysts is Na
2
SnO
3
> K
3
VO
4
> K
3
PO
4
. These catalysts’ average M-O bond lengths are 1.939,
1.734, and 1.571 (in Å). The average O
…
cations distances are 2.271, 2.617, and 2.604 (in Å)
for Na
2
SnO
3
, K
3
VO
4
, and K
3
PO
4
, respectively. The differences in the molecular structures
of the catalysts severely affect their catalytic performance. Hence, a longer M-O bond dis-
tance performs better in keeping a higher CO
2
content in the product. The planar structure
of the SnO
32−
catalyst anion causes complexation of the alkoxide anion (RO
−
generated
from ROH) to the Sn centre for a while during the initiation step. This complexation delays
the reaction in case of SnO
32−
vs. VO
43−
, 9.5 h vs. 5.0 h, respectively, in experiment, which
is in accordance with the higher barrier of the cEC ring opening for SnO
32−
vs. VO
43−
(Table
2). For a detailed analysis we refer to the reference 47. In addition, the cation’s lability also
plays a role in more efficient catalytic performance. Especially for the smaller anions such
as PO
43−
, bigger cations perform better than smaller ones. The ΔG
≠
values of the ring open-
ing decrease from H
+
to K
+
for phosphate anion, and in the experiment, the CO
2
contribu-
tion increases with larger cations. This effect is due to the longer anion…ation distance
and less-covalent character of the anion-cation interaction, leading to more accessibility of
the catalyst anion in the reaction. For bigger anions such as SnO
32−
and VO
43−
, switching
between Na
+
and K
+
does not affect the catalyst performance, and due to the stronger ionic
character, the ΔG
≠
values and CO
2
-contribution do not change with changing Na
+
to K
+
.
The complete results of the cation influence on the ΔG
≠
and the experimental CO
2
contri-
bution are reported in Table 2. We note that for H
3
PO
4
and Li
3
PO
4
catalysts in Table 2 only
minor conversions (due to the reduced activity) were observed. However, the conversion
was higher than 80 percent for the rest of the catalysts.
Table 2. The activation free energies (ΔG
≠
) and CO
2
contents for various cations for the PO
43−
, VO
43
,
and SnO
32−
anions.
Catalyst ΔG
≠
H
+
Abstraction
[kcal/mol]
ΔG
≠
cEC Opening
[kcal/mol]
Exp. CO
2
Content
[%]
H
3
PO
4
22.3 50.0 3.0
Li
3
PO
4
12.8 39.1 5.0
Na
3
PO
4
7.6 5.2 11.0
K
3
PO
4
4.2 4.0 15.0
Na
2
SnO
3
6.2 11.2 26.0
K
2
SnO
3
6.5 10.8 22.0
Na
3
VO
4
4.8 4.5 26.0
K
3
VO
4
5.8 7.7 25.0
Figure 3 depicts the considered CO
2
release mechanism triggered by the interaction
between the catalyst anion and the CH
2
of the cEC ring.
Figure 3. The nucleophilic attack by the catalyst anion to the CH
2
of the cEC ring results in a terminal
COO
−
and leads to the CO
2
release. CV1 and CV2 are the distances used for MTD simulation.
Figure 3. The nucleophilic attack by the catalyst anion to the CH
2
of the cEC ring results in a terminal
COO−and leads to the CO2release. CV1 and CV2 are the distances used for MTD simulation.
The examination of the CO
2
release path of Figure 3with MTD simulations confirmed
that the higher nucleophilicity of the phosphate anion is the leading cause for easier CO
2
release during the polymerisation of cEC. Considering that, the mechanism shown in
Figure 3, which results in the attachment of the catalyst to the cEC ring, can proceed in
parallel with the pathway shown in Figure 2. This work analyses this pathway and the
elementary steps to conduct this molecular complex (between the catalyst anion and the
cleaved cEC) to further chain growth and polymerisation. The experimental observations
show that higher catalyst concentration increases the CO
2
release. Additionally, at the
beginning of polymerisation, the primary cause of the CO
2
release is due to the cEC
decomposition. Hence, the hidden elementary steps in the pathway shown in Figure 3
can affect the molecular weight of the polymer as well as the CO
2
content and the overall
reaction rate. We investigated this pathway in the current study to better understand the
catalyst–cEC interaction and to address the role of catalyst characteristics.
Polymers 2024,16, 136 8 of 24
3.1. Simulation of the Catalyst–cEC Attachment Path
To run metadynamics simulations on the path shown in Figure 3, we have considered
the distance between the nucleophilic oxygen atom of the catalyst anion and the methylene
carbon in the cEC monomer, i.e., O
−···
C(CH
2
-O), as the first collective variable (CV1) and
the C(H
2
)
···
O(ethereal) inside the cEC as the second collective variable (CV2). The MTD
simulations have been performed at 413, 423, 433, and 443 K. For brevity, we only show the
results of 423 K; for the other temperatures, we include the data in SI (Figures S1–S3). We have
shown the variation of the CV1 versus CV2 in Figure 4for the trajectory of cEC cleavage and
catalyst attachment at 423 K, in which the two reaction coordinates (two distances) are plotted
versus each other (x, y axis for CV1 and CV2, respectively). Investigation of the patterns
created by the CVs in each plot indicates some similarities and differences between the path
of catalyst attachment for different catalysts that can lead us to different elementary steps. We
note that the three catalysts’ computational parameters and temperatures are equal. As shown
in Figure 4, for Na
2
SnO
3
and K
3
VO
4
catalysts, the catalyst attachment path goes through
an intermediate region of CV1
≥
2.5 Å and CV2
≥
2.5 Å. On the other hand, in the case of
K
3
PO
4
, a diagonal transition path (that directly connects the reactant state to the catalyst–cEC
complex) with a more significant density of the transient states, which is invisible for the
SnO
32−
and VO
43−
, appears in the plot. Variation of CV1 versus CV2 in Figure 4identifies
different regions based on the density of the dots (transient structures). These regions are
singled out in Figure 4based on the values of the CVs. As can be seen in Figure 5(left side),
three stable states are observed: the reactant state (CV1
≥
2.5 Å, CV2
≤
1.6 Å), the intermediate
state (CV1
≥
2.5 Å, CV2
≥
2.5 Å), and the state in which the CAT-cEC bond (complex) is
formed (CV1
≤
1.6 Å, CV2
≥
2.5). The regions between these states indicate the transition
state areas, i.e., TS
vert.
, TS
hor.
, and TS
Nu−attack
. On the right side of Figure 5, the schematic
molecular structures corresponding to the highlighted regions are depicted.
Polymers 2024, 16, x FOR PEER REVIEW 10 of 27
Figure 4. Variation of the O
−…
C(CH
2
-O) distance (CV1) versus the C(H
2
)
…
O(ethereal) distance (CV2).
Different regions can be identified based on the density of the dots (transient structures). The solid
red bond shows the CH
2
-O(cEC) that is cleaved via nucleophilic attack by the catalyst anion and the
dashed red line indicates the CH
2
-O(CAT) bond that is formed.
Figure 5. Left: various regions according to the density of the transient states in Figure 4. Right: the
schematic structures that correlate to the highlighted regions on the left side.
In the case of Na
2
SnO
3
and K
3
VO
4
, the mechanism begins with ring cleavage (ethereal
CH
2
-O bond cleavage and intermediate formation) by passing a vertical transition state
region (TS
vert.
). The intermediate either returns to the closed cEC or releases CO
2
. In the
case of K
3
PO
4
, an additional path can be identified: the cEC cleavage and nucleophilic
attack by the catalyst anion happen simultaneously, and a diagonal transition state region
is observed (TS
Nu−attack
). These two distinguishable patterns can be identified in the calcu-
lated FESs according to the relative energies in kcal/mol. Figure 6 shows the calculated
FESs at 423K for the three catalysts. As seen in Figure 6, the stepwise path, which is the
transition through the intermediate, can be observed in the pattern of the FES of Na
2
SnO
3
and K
3
VO
4
. The FES of the K
3
PO
4
identifies a new concerted path in addition to the step-
wise, which corresponds to the diagonal transition in that the two CVs are involved sim-
ultaneously. The numbers at the FESs show the energies in kcal/mol. The more stable in-
termediate, in the case of Na
2
SnO
3
and K
3
VO
4
, can conduct the reaction in both directions;
that is, turning into the catalyst attachment and CO
2
-release or into the backward direction
and generating the cEC, which involves an alkoxide attack and ring opening (Figure 2
Step II). In the case of K
3
PO
4
, a pattern with direct catalyst attachment (diagonal with sim-
ultaneous involvement of CVs) can be observed.
Figure 4. Variation of the O
−· · ·
C(CH
2
-O) distance (CV1) versus the C(H
2
)
· · ·
O(ethereal) distance
(CV2). Different regions can be identified based on the density of the dots (transient structures). The
solid red bond shows the CH
2
-O(cEC) that is cleaved via nucleophilic attack by the catalyst anion
and the dashed red line indicates the CH2-O(CAT) bond that is formed.
In the case of Na
2
SnO
3
and K
3
VO
4
, the mechanism begins with ring cleavage (ethereal
CH
2
-O bond cleavage and intermediate formation) by passing a vertical transition state
region (TS
vert.
). The intermediate either returns to the closed cEC or releases CO
2
. In the case
of K
3
PO
4
, an additional path can be identified: the cEC cleavage and nucleophilic attack by
the catalyst anion happen simultaneously, and a diagonal transition state region is observed
(TS
Nu−attack
). These two distinguishable patterns can be identified in the calculated FESs
according to the relative energies in kcal/mol. Figure 6shows the calculated FESs at 423 K
for the three catalysts. As seen in Figure 6, the stepwise path, which is the transition
Polymers 2024,16, 136 9 of 24
through the intermediate, can be observed in the pattern of the FES of Na
2
SnO
3
and K
3
VO
4
.
The FES of the K
3
PO
4
identifies a new concerted path in addition to the stepwise, which
corresponds to the diagonal transition in that the two CVs are involved simultaneously.
The numbers at the FESs show the energies in kcal/mol. The more stable intermediate,
in the case of Na
2
SnO
3
and K
3
VO
4
, can conduct the reaction in both directions; that is,
turning into the catalyst attachment and CO
2
-release or into the backward direction and
generating the cEC, which involves an alkoxide attack and ring opening (Figure 2Step II).
In the case of K
3
PO
4
, a pattern with direct catalyst attachment (diagonal with simultaneous
involvement of CVs) can be observed.
Polymers 2024, 16, x FOR PEER REVIEW 10 of 27
Figure 4. Variation of the O
−…
C(CH
2
-O) distance (CV1) versus the C(H
2
)
…
O(ethereal) distance (CV2).
Different regions can be identified based on the density of the dots (transient structures). The solid
red bond shows the CH
2
-O(cEC) that is cleaved via nucleophilic attack by the catalyst anion and the
dashed red line indicates the CH
2
-O(CAT) bond that is formed.
Figure 5. Left: various regions according to the density of the transient states in Figure 4. Right: the
schematic structures that correlate to the highlighted regions on the left side.
In the case of Na
2
SnO
3
and K
3
VO
4
, the mechanism begins with ring cleavage (ethereal
CH
2
-O bond cleavage and intermediate formation) by passing a vertical transition state
region (TS
vert.
). The intermediate either returns to the closed cEC or releases CO
2
. In the
case of K
3
PO
4
, an additional path can be identified: the cEC cleavage and nucleophilic
attack by the catalyst anion happen simultaneously, and a diagonal transition state region
is observed (TS
Nu−attack
). These two distinguishable patterns can be identified in the calcu-
lated FESs according to the relative energies in kcal/mol. Figure 6 shows the calculated
FESs at 423K for the three catalysts. As seen in Figure 6, the stepwise path, which is the
transition through the intermediate, can be observed in the pattern of the FES of Na
2
SnO
3
and K
3
VO
4
. The FES of the K
3
PO
4
identifies a new concerted path in addition to the step-
wise, which corresponds to the diagonal transition in that the two CVs are involved sim-
ultaneously. The numbers at the FESs show the energies in kcal/mol. The more stable in-
termediate, in the case of Na
2
SnO
3
and K
3
VO
4
, can conduct the reaction in both directions;
that is, turning into the catalyst attachment and CO
2
-release or into the backward direction
and generating the cEC, which involves an alkoxide attack and ring opening (Figure 2
Step II). In the case of K
3
PO
4
, a pattern with direct catalyst attachment (diagonal with sim-
ultaneous involvement of CVs) can be observed.
Figure 5. Left: various regions according to the density of the transient states in Figure 4.Right: the
schematic structures that correlate to the highlighted regions on the left side.
Polymers 2024, 16, x FOR PEER REVIEW 11 of 27
Figure 6. The FESs of the CO
2
release mechanism and the energetics of the regions correspond to
the areas depicted in Figure 5 for the three catalysts K
3
VO
4
, Na
2
SnO
3,
and K
3
PO
4
at 423 K. The energy
values (numbers) are in kcal/mol. Two different patterns can be identified in the case of K
3
VO
4
and
Na
2
SnO
3
(stepwise via an intermediate) versus K
3
PO
4
(concerted).
In summary, the different patterns observed for stannate/vanadate vs. phosphate re-
flect the different natures of the chemical interactions between the catalyst and cEC (non-
similar catalyst characteristics). The more-nucleophilic character of the catalyst anion re-
sults in the sampling of a concerted pathway of the catalyst attachment. The following
section compares the calculated results with model parameters determined from kinetic
measurements to address the correlations/differences between simulations and experi-
ments.
3.2. Experimental Analysis of the Kinetics of Polymerisation of cEC, including CO
2
Release
The experimental investigations on the kinetics of the polymerisation of cEC indicate
that higher CO
2
content can be obtained by the stannate and vanadate anions, which con-
firms the better performance of these catalysts for the ROP of cEC. Notably, the kinetic
measurements confirmed that, at the beginning of the reaction, the primary source of the
CO
2
release is the cEC cleavage (decomposition). The schematic presentation of the exper-
imentally observed reaction network for SnO
32−
and VO
43−
anions is shown in Scheme 4
and for PO
43−
in Scheme 5. k
1
and k
2
in Scheme 4 represent the rate constants of ring-open-
ing polymerisation of cEC and the chain propagation with CO
2
release, respectively, and
k
3
and k
4
show the polymeric chain (PECn) decay due to decomposition or CO
2
release
from the chain, respectively.
Scheme 4. The experimentally observed reaction network of cEC ROP, catalysed by SnO
32−
and
VO
43−
anions. k
1
, k
2
, k
3
, and k
4
indicate the corresponding rate constants of ring-opening polymeri-
sation (PECn), chain propagation with CO
2
release, decomposition of the chain, and CO
2
release
from the chain, respectively.
The kinetics of these reactions can be described with first-order rate Equations (1)–
(4).
r=
k
∙c
(1)
Figure 6. The FESs of the CO
2
release mechanism and the energetics of the regions correspond to the
areas depicted in Figure 5for the three catalysts K
3
VO
4
, Na
2
SnO
3
, and K
3
PO
4
at 423 K. The energy
values (numbers) are in kcal/mol. Two different patterns can be identified in the case of K
3
VO
4
and
Na2SnO3(stepwise via an intermediate) versus K3PO4(concerted).
In summary, the different patterns observed for stannate/vanadate vs. phosphate
reflect the different natures of the chemical interactions between the catalyst and cEC
(non-similar catalyst characteristics). The more-nucleophilic character of the catalyst anion
results in the sampling of a concerted pathway of the catalyst attachment. The following
section compares the calculated results with model parameters determined from kinetic
measurements to address the correlations/differences between simulations and experiments.
3.2. Experimental Analysis of the Kinetics of Polymerisation of cEC, including CO2Release
The experimental investigations on the kinetics of the polymerisation of cEC indicate
that higher CO
2
content can be obtained by the stannate and vanadate anions, which
confirms the better performance of these catalysts for the ROP of cEC. Notably, the kinetic
measurements confirmed that, at the beginning of the reaction, the primary source of the
Polymers 2024,16, 136 10 of 24
CO
2
release is the cEC cleavage (decomposition). The schematic presentation of the experi-
mentally observed reaction network for SnO
32−
and VO
43−
anions is shown in Scheme 4
and for PO
43−
in Scheme 5. k
1
and k
2
in Scheme 4represent the rate constants of ring-
opening polymerisation of cEC and the chain propagation with CO
2
release, respectively,
and k
3
and k
4
show the polymeric chain (PECn) decay due to decomposition or CO
2
release
from the chain, respectively.
Polymers 2024, 16, x FOR PEER REVIEW 11 of 27
Figure 6. The FESs of the CO
2
release mechanism and the energetics of the regions correspond to
the areas depicted in Figure 5 for the three catalysts K
3
VO
4
, Na
2
SnO
3,
and K
3
PO
4
at 423 K. The energy
values (numbers) are in kcal/mol. Two different patterns can be identified in the case of K
3
VO
4
and
Na
2
SnO
3
(stepwise via an intermediate) versus K
3
PO
4
(concerted).
In summary, the different patterns observed for stannate/vanadate vs. phosphate re-
flect the different natures of the chemical interactions between the catalyst and cEC (non-
similar catalyst characteristics). The more-nucleophilic character of the catalyst anion re-
sults in the sampling of a concerted pathway of the catalyst attachment. The following
section compares the calculated results with model parameters determined from kinetic
measurements to address the correlations/differences between simulations and experi-
ments.
3.2. Experimental Analysis of the Kinetics of Polymerisation of cEC, including CO
2
Release
The experimental investigations on the kinetics of the polymerisation of cEC indicate
that higher CO
2
content can be obtained by the stannate and vanadate anions, which con-
firms the better performance of these catalysts for the ROP of cEC. Notably, the kinetic
measurements confirmed that, at the beginning of the reaction, the primary source of the
CO
2
release is the cEC cleavage (decomposition). The schematic presentation of the exper-
imentally observed reaction network for SnO
32−
and VO
43−
anions is shown in Scheme 4
and for PO
43−
in Scheme 5. k
1
and k
2
in Scheme 4 represent the rate constants of ring-open-
ing polymerisation of cEC and the chain propagation with CO
2
release, respectively, and
k
3
and k
4
show the polymeric chain (PECn) decay due to decomposition or CO
2
release
from the chain, respectively.
Scheme 4. The experimentally observed reaction network of cEC ROP, catalysed by SnO
32−
and
VO
43−
anions. k
1
, k
2
, k
3
, and k
4
indicate the corresponding rate constants of ring-opening polymeri-
sation (PECn), chain propagation with CO
2
release, decomposition of the chain, and CO
2
release
from the chain, respectively.
The kinetics of these reactions can be described with first-order rate Equations (1)–
(4).
r=
k
∙c
(1)
Scheme 4. The experimentally observed reaction network of cEC ROP, catalysed by SnO
32−
and
VO
43−
anions. k
1
, k
2
, k
3
, and k
4
indicate the corresponding rate constants of ring-opening polymeri-
sation (PECn), chain propagation with CO
2
release, decomposition of the chain, and CO
2
release
from the chain, respectively.
Polymers 2024, 16, x FOR PEER REVIEW 14 of 27
The determination of the conversion and the CO
2
content was performed via the
1
H-NMR
of the products dissolved in CD
2
Cl
2
or CDCl
3
and the measurement of
31
P-NMR in CD
2
Cl
2
(see Figure 8). Similar to the case of stannate, k
3,
and k
4
are also negligible for phosphate
catalysis.
Scheme 5. The experimentally observed reaction network was catalysed by PO
43−
anions. k
1
, k
2
, k
3
,
and k
4
indicate the corresponding rate constants of polymerization, propagation with CO
2
release,
decomposition of the chain, and CO
2
release from the chain, respectively. k
des
indicates the rate con-
stant of catalyst deactivation.
Table 4. Experimentally measured rate constants corresponding to Scheme 5 (for PO
43−
anion as
the catalyst). CO
2
contents were determined via
1
H-NMR in CD
2
Cl
2
.
T [K] k
1
[1/s]
k
2
[1/s]
k
des
[1/s]
k
2
/k
1
[-]
CO
2
[%]
395 0.021 0.023 0.020 1.09 17.4
405 0.0036 0.0028 0.024 0.78 27.9
415 0.0054 0.032 0.017 5.93 7.31
425 0.0071 0.026 0.029 3.66 10.64
As Table 4 shows, the ratio of k
2
/k
1
is very different from what was observed with
stannate. For the first entry, it is close to one, which means the propagation reaction with
CO
2
release and the ring-opening polymerisation have similar rates. However, for entry
two, the ring-opening polymerisation is faster than the propagation with CO
2
release. This
difference can also be seen from the high CO
2
content. However, since the reaction did not
proceed to full conversion and achieved only 7% yield, this measurement is highly sus-
ceptible to experimental error. For entries 3 and 4, the rate of the reaction with CO
2
release
is faster than cEC ring-opening polymerisation, producing a polymer with much lower
CO
2
contents. Due to the simultaneous deactivation of the phosphate catalyst, the evalua-
tion of the kinetic parameter is greatly hampered, therefore resulting in large error bars of
the deduced values.
Furthermore, the values of k
des
versus k
1
and k
2
are considered. In that case, a closer
correlation between the propagation with CO
2
release (k
2
) and catalyst deactivation can
be found, e.g., for entry four at 425 K (k
2
and k
des
are very close). In comparison to the
computational results, this is in line with the concerted mechanism in the case of phos-
phate where the catalyst attachment to the cleaved cEC appears faster than in the case of
stannate, due to a lower barrier (10 vs. 18.5 kcal/mol, respectively, Figure 6). This situation
provides conditions for a faster CO
2
release which is correlated to the lower barrier in
simulated FES of K
3
PO
4
for the diagonal path.
Considering the theoretical predictions and experimental results we have shown so
far, the main question that still needs to be addressed is how the polymerisation proceeds
through the generated molecular complex, in which the catalyst is bound to the monomer
Scheme 5. The experimentally observed reaction network was catalysed by PO
43−
anions. k
1
, k
2
, k
3
,
and k
4
indicate the corresponding rate constants of polymerization, propagation with CO
2
release,
decomposition of the chain, and CO
2
release from the chain, respectively. k
des
indicates the rate
constant of catalyst deactivation.
The kinetics of these reactions can be described with first-order rate Equations (1)–(4).
r1=k1·cn
cEC (1)
r2=k2·cn
cEC (2)
r3=k3·cn
PECn (3)
r4=k4·cn
PECn (4)
The material balances of the involved components were combined from these rate
equations and solved numerically. Using a Runge–Kutta algorithm, the rate constants
were determined via fitting the model to the experimental data. For fitting the absorption
curves, the concentrations of the corresponding species were multiplied with the extinc-
tion coefficients of these species, estimated from single component spectra (Figure 7and
Table 3). (For a detailed description of the parameter fitting, see Supporting Information,
Tables S2–S6
. For representative results of fitting the kinetic model to experimental data for
all experiments at different temperatures, see Supporting Information, Figure S13–S15).
Polymers 2024,16, 136 11 of 24
Polymers 2024, 16, x FOR PEER REVIEW 12 of 27
r=k
∙c
(2)
r=k
∙c
(3)
r=k
∙c
(4)
The material balances of the involved components were combined from these rate
equations and solved numerically. Using a Runge–Kutta algorithm, the rate constants
were determined via fitting the model to the experimental data. For fitting the absorption
curves, the concentrations of the corresponding species were multiplied with the extinc-
tion coefficients of these species, estimated from single component spectra (Figure 7 and
Table 3). (For a detailed description of the parameter fitting, see Supporting Information,
Tables S2–S6. For representative results of fitting the kinetic model to experimental data
for all experiments at different temperatures, see Supporting Information, Figure S13–
S15).
(A) (B)
Figure 7. Representative results of fitting the kinetic model to the experimental data. Experimental
conditions: TR = 170 °C, ccEC = 13.83 mol/L, cEthylenglycol = 1.38 mol/L, 0.01 eq. Potassium stannate. (A)
Concentrate against the time of the reactants and products. (B) Fit the IR adsorption bands of the
experiment to the experimental results.
Table 3. At various temperatures, experimentally measured rate constants corresponding to the
reaction network in Scheme 4 (for SnO32− anion as the catalyst). CO2 values were determined via
1H-NMR in CD2Cl2.
T
[K]
k1
[1/s]
k2
[1/s]
k3
[1/s]
k4
[1/s]
k2
/
k1
[-−]
CO2
[%]
413 0.029 0.026 5.4 × 10−7 1.0 × 10−5 0.89 26
423 0.029 0.028 6.8 × 10−6 7.4 × 10−6 0.96 26
433 0.038 0.034 1.4 × 10−6 2.3 × 10−4 0.89 26
443 0.032 0.034 1.3 × 10−5 4.9 × 10−4 1.06 24
Table 3 reports the experimentally measured rate constants corresponding to the re-
action network in Scheme 4 (for SnO32− anion as the catalyst) at various temperatures. As
seen in Table 3, k3 and k4 are negligibly low. Hence, the primary source of the CO2 release
is a result of the interaction between monomer and catalyst (CO2 release from cEC). This
result agrees with the mechanism shown in Figure 3, in which the CO2 release is due to
the catalyst attachment. In Table 3, we also show the ratio of k2/k1 to compare the rate of
CO2 release versus polymerisation of cEC. The relative values show that CO2 release oc-
curs either slightly slower than polymerisation (entries 1–3) or at a similar rate to polymer-
isation (entry 4). The k2/k1 ratio is not strongly dependent on the temperature. This exper-
imental observation agrees well with the computationally observed mechanism in Figures
Figure 7. Representative results of fitting the kinetic model to the experimental data. Experimental
conditions: T
R
= 170
◦
C, c
cEC
= 13.83 mol/L, c
Ethylenglycol
= 1.38 mol/L, 0.01 eq. Potassium stannate.
(A) Concentrate against the time of the reactants and products. (B) Fit the IR adsorption bands of the
experiment to the experimental results.
Table 3. At various temperatures, experimentally measured rate constants corresponding to the
reaction network in Scheme 4(for SnO
32−
anion as the catalyst). CO
2
values were determined via
1H-NMR in CD2Cl2.
T
[K]
k1
[1/s]
k2
[1/s]
k3
[1/s]
k4
[1/s]
k2/k1
[−]
CO2
[%]
413 0.029 0.026 5.4 ×10−71.0 ×10−50.89 26
423 0.029 0.028 6.8 ×10−67.4 ×10−60.96 26
433 0.038 0.034 1.4 ×10−62.3 ×10−40.89 26
443 0.032 0.034 1.3 ×10−54.9 ×10−41.06 24
Table 3reports the experimentally measured rate constants corresponding to the
reaction network in Scheme 4(for SnO
32−
anion as the catalyst) at various temperatures.
As seen in Table 3, k
3
and k
4
are negligibly low. Hence, the primary source of the CO
2
release is a result of the interaction between monomer and catalyst (CO
2
release from cEC).
This result agrees with the mechanism shown in Figure 3, in which the CO
2
release is
due to the catalyst attachment. In Table 3, we also show the ratio of k
2
/k
1
to compare
the rate of CO
2
release versus polymerisation of cEC. The relative values show that CO
2
release occurs either slightly slower than polymerisation (entries 1–3) or at a similar rate to
polymerisation (entry 4). The k
2
/k
1
ratio is not strongly dependent on the temperature.
This experimental observation agrees well with the computationally observed mechanism
in Figures 4and 5(for stannate and vanadate anions). Hence, the possibility of proceeding
via catalyst attachment followed by the CO
2
release or returning to the closed cEC and
further involvement in the main polymerisation path (propagation via alkoxide attack)
is identified.
Furthermore, the calculated FESs in Figure 6show similar values of the
∆
G
=
for
converting the intermediate to the CAT-CH
2
bond and forming the reactant state (18.5
vs. 18.7 kcal/mol, respectively). According to the experimental measurements, a similar
trend of the k
1
and k
2
rate constant ratios can be observed for VO
43−
, which implies
similar catalytic characteristics of the VO
43−
and SnO
32−
anions. Additionally, as shown in
Figure 6, in the middle part, for the backward path, i.e., returning to the reactant state from
the intermediate state, the
∆
G
=
is more favoured (i.e., 9.2 kcal/mol from intermediate to
the reactant state). The results of kinetic measurements of VO
43−
and the rate constants are
reported in Table S1 in SI.
Polymers 2024,16, 136 12 of 24
For the PO
43−
anion, the kinetic measurement results differ from those with SnO
32−
and VO
43−
as catalysts. The experimentally observed network of the reactions catalysed
by PO
43−
is shown in Scheme 5. In this Scheme, k
des
indicates the rate constant of catalyst
deactivation, which is observed for PO
43−
. The results of the kinetic measurements and
the rate constants are reported in Table 4. The literature (also shown in Table 4) shows that
the ROP of cEC catalysed by phosphate has lower CO
2
content in the final polymer (10%
lower than K
3
VO
4
and Na
2
SnO
3
) [
46
]. In addition to that, according to
31
P-NMR, insertion
of phosphate is observed during the polymerisation (
31
P NMR (500 MHz, CD
2
Cl
2
)
δ
17.84
(s, 1P), 1.64 (d, J = 169.7 Hz, 7P), 1.08 (s, 2P)), shown in Figure 8. The peaks at 0–2 ppm
can be assigned to potassium phosphate [
47
]. However, there is a signal at 17.84 ppm,
which does not correspond to the signals of potassium phosphate. Instead, signals related
to phosphonates can be found in this range [47].
Table 4. Experimentally measured rate constants corresponding to Scheme 5(for PO
43−
anion as the
catalyst). CO2contents were determined via 1H-NMR in CD2Cl2.
T [K] k1
[1/s]
k2
[1/s]
kdes
[1/s]
k2/k1
[−]
CO2
[%]
395 0.021 0.023 0.020 1.09 17.4
405 0.0036 0.0028 0.024 0.78 27.9
415 0.0054 0.032 0.017 5.93 7.31
425 0.0071 0.026 0.029 3.66 10.64
Polymers 2024, 16, x FOR PEER REVIEW 13 of 27
4 and 5 (for stannate and vanadate anions). Hence, the possibility of proceeding via cata-
lyst attachment followed by the CO2 release or returning to the closed cEC and further
involvement in the main polymerisation path (propagation via alkoxide attack) is identi-
fied.
Furthermore, the calculated FESs in Figure 6 show similar values of the ΔG≠ for con-
verting the intermediate to the CAT-CH2 bond and forming the reactant state (18.5 vs. 18.7
kcal/mol, respectively). According to the experimental measurements, a similar trend of
the k1 and k2 rate constant ratios can be observed for VO43−, which implies similar catalytic
characteristics of the VO43− and SnO32− anions. Additionally, as shown in Figure 6, in the
middle part, for the backward path, i.e., returning to the reactant state from the interme-
diate state, the ΔG≠ is more favoured (i.e., 9.2 kcal/mol from intermediate to the reactant
state). The results of kinetic measurements of VO43− and the rate constants are reported in
Table S1 in SI.
For the PO43− anion, the kinetic measurement results differ from those with SnO32−
and VO43− as catalysts. The experimentally observed network of the reactions catalysed by
PO43− is shown in Scheme 5. In this Scheme, kdes indicates the rate constant of catalyst de-
activation, which is observed for PO43−. The results of the kinetic measurements and the
rate constants are reported in Table 4. The literature (also shown in Table 4) shows that the
ROP of cEC catalysed by phosphate has lower CO2 content in the final polymer (10% lower
than K3VO4 and Na2SnO3) [47]. In addition to that, according to 31P-NMR, insertion of
phosphate is observed during the polymerisation (31P NMR (500 MHz, CD2Cl2) δ 17.84 (s,
1P), 1.64 (d, J = 169.7 Hz, 7P), 1.08 (s, 2P)), shown in Figure 8. The peaks at 0–2 ppm can be
assigned to potassium phosphate [48]. However, there is a signal at 17.84 ppm, which does
not correspond to the signals of potassium phosphate. Instead, signals related to phos-
phonates can be found in this range [48].
Figure 8. 31P NMR spectrum in CD2Cl2 cEC polymer prepared with 0.02 eq. potassium phosphate as
catalyst at reaction conditions: TR = 140 °C, RPM = 1000, experimental time = 5 h.
In all experiments, below 170 °C, no complete reaction could be observed. Even add-
ing fresh potassium phosphate did not lead to any further reaction. Only above 170 °C is
it possible to complete the reaction, since the polymerisation reaction is faster than the
deactivation of the catalyst. However, this leads to lower CO2 contents of the products.
Figure 8.
31
P NMR spectrum in CD
2
Cl
2
cEC polymer prepared with 0.02 eq. potassium phosphate as
catalyst at reaction conditions: TR= 140 ◦C, RPM = 1000, experimental time = 5 h.
In all experiments, below 170
◦
C, no complete reaction could be observed. Even
adding fresh potassium phosphate did not lead to any further reaction. Only above
170
◦
C is it possible to complete the reaction, since the polymerisation reaction is faster
than the deactivation of the catalyst. However, this leads to lower CO
2
contents of the
products. The determination of the conversion and the CO
2
content was performed via the
Polymers 2024,16, 136 13 of 24
1H-NMR
of the products dissolved in CD
2
Cl
2
or CDCl
3
and the measurement of
31
P-NMR
in CD
2
Cl
2
(see Figure 8). Similar to the case of stannate, k
3
, and k
4
are also negligible for
phosphate catalysis.
As Table 4shows, the ratio of k
2
/k
1
is very different from what was observed with
stannate. For the first entry, it is close to one, which means the propagation reaction with
CO
2
release and the ring-opening polymerisation have similar rates. However, for entry
two, the ring-opening polymerisation is faster than the propagation with CO
2
release. This
difference can also be seen from the high CO
2
content. However, since the reaction did
not proceed to full conversion and achieved only 7% yield, this measurement is highly
susceptible to experimental error. For entries 3 and 4, the rate of the reaction with CO
2
release is faster than cEC ring-opening polymerisation, producing a polymer with much
lower CO
2
contents. Due to the simultaneous deactivation of the phosphate catalyst, the
evaluation of the kinetic parameter is greatly hampered, therefore resulting in large error
bars of the deduced values.
Furthermore, the values of k
des
versus k
1
and k
2
are considered. In that case, a closer
correlation between the propagation with CO
2
release (k
2
) and catalyst deactivation can
be found, e.g., for entry four at 425 K (k
2
and k
des
are very close). In comparison to the
computational results, this is in line with the concerted mechanism in the case of phosphate
where the catalyst attachment to the cleaved cEC appears faster than in the case of stannate,
due to a lower barrier (10 vs. 18.5 kcal/mol, respectively, Figure 6). This situation provides
conditions for a faster CO
2
release which is correlated to the lower barrier in simulated FES
of K3PO4for the diagonal path.
Considering the theoretical predictions and experimental results we have shown so
far, the main question that still needs to be addressed is how the polymerisation proceeds
through the generated molecular complex, in which the catalyst is bound to the monomer
via the CH
2
moiety (Figures 4and 5), and what is the probability of CO
2
release alongside
the chain propagation path starting from the catalyst–monomer molecular complex.
3.3. Alternative Pathways of Chain Growth Revealed by Simulations
One of the possible chain growth paths starts with the catalyst–anion attached to the
monomer is shown in Figure 9. We have examined this path using MTD simulations with
the considered CVs depicted in Figure 9. The considered CVs correspond to the COO
−
nucleophilic attack and cEC cleavage (CV1 and CV2, respectively).
Polymers 2024, 16, x FOR PEER REVIEW 16 of 27
3.3. Alternative Pathways of Chain Growth Revealed by Simulations
One of the possible chain growth paths starts with the catalyst–anion attached to the
monomer is shown in Figure 9. We have examined this path using MTD simulations with
the considered CVs depicted in Figure 9. The considered CVs correspond to the COO
−
nucleophilic attack and cEC cleavage (CV1 and CV2, respectively).
Figure 9. The computationally considered pathway for chain growth starts from the catalyst–mon-
omer molecular complex, while the catalyst is attached to the opened cEC.
The analysis of the corresponding MTD trajectories for this path at four temperatures
for the three catalysts indicates that, since the COO
−
terminus has a relatively weak nucle-
ophilic character than the RO
−
(due to the negative charge delocalisation on the two oxy-
gen atoms), the barrier for the nucleophilic attack to the second cEC increases. On the
other hand, since the ΔG
≠
for the CH
2
-O(ethereal) bond cleavage is, on average, 8–11
kcal/mol for various molecular systems and temperatures, the possibility of CO
2
release
upon cEC cleavage enhances before COO
−
attachment. Alongside examination of this
pathway, the possibility of 30% random CO
2
release for some of the trajectories was ob-
served (four trajectories ended with CO
2
release out of 12 trajectories). Thus, to prohibit
CO
2
liberation and obtain the MTD trajectory for calculating the ΔG
≠
values, we had to fix
the corresponding bond lengths leading to CO
2
liberation. These bonds, responsible for
CO
2
release and kept fixed during simulations, are highlighted in red in Figure 9. The
calculated barriers for this reaction (ΔG
≠
) are shown in Table 5 for the rate-determining
step, which is the nucleophilic attack by COO
−
to the second cEC monomer (CH
2
moiety).
Table 5. The calculated ΔG
≠
for the COO
−
attack to the cEC (rate determining step) at the pathway
shown in Figure 9 for the three catalysts calculated at four temperatures.
Catalyst/ΔG
≠
[kcal/mol] Na
2
SnO
3
[kcal/mol]
K
3
VO
4
[kcal/mol]
K
3
PO
4
[kcal/mol]
413 K 34.0 36.0 28.1
423 K 32.3 33.5 27.2
433 K 31.8 30.8 26.0
443 K 28.9 29.2 24.0
As seen from Table 5, generally, the barriers decrease with increasing temperature,
and for K
3
PO
4
, the barriers are lowest. However, chain propagation can proceed via this
pathway without the starter ROH, with a higher barrier and more probability of CO
2
lib-
eration than the path including the starter ROH. This finding agrees with the experimental
measurements that the majority of CO
2
liberation is caused by cEC decomposition at the
beginning of the reaction. In all paths, one reaction coordinate (CV2) is responsible for the
ring opening of cEC to convert it to a nucleophile (COO
−
) that can further proceed with
the ring opening of the next cEC (propagation).
Figure 9. The computationally considered pathway for chain growth starts from the catalyst–monomer
molecular complex, while the catalyst is attached to the opened cEC.
The analysis of the corresponding MTD trajectories for this path at four temperatures
for the three catalysts indicates that, since the COO
−
terminus has a relatively weak
nucleophilic character than the RO
−
(due to the negative charge delocalisation on the
two oxygen atoms), the barrier for the nucleophilic attack to the second cEC increases.
On the other hand, since the
∆
G
=
for the CH
2
-O(ethereal) bond cleavage is, on average,
8–11 kcal/mol for various molecular systems and temperatures, the possibility of CO
2
release upon cEC cleavage enhances before COO−attachment. Alongside examination of
Polymers 2024,16, 136 14 of 24
this pathway, the possibility of 30% random CO
2
release for some of the trajectories was
observed (four trajectories ended with CO
2
release out of 12 trajectories). Thus, to prohibit
CO
2
liberation and obtain the MTD trajectory for calculating the
∆
G
=
values, we had to
fix the corresponding bond lengths leading to CO
2
liberation. These bonds, responsible
for CO
2
release and kept fixed during simulations, are highlighted in red in Figure 9. The
calculated barriers for this reaction (
∆
G
=
) are shown in Table 5for the rate-determining
step, which is the nucleophilic attack by COO
−
to the second cEC monomer (CH
2
moiety).
Table 5. The calculated
∆
G
=
for the COO
−
attack to the cEC (rate determining step) at the pathway
shown in Figure 9for the three catalysts calculated at four temperatures.
Catalyst/∆G=
[kcal/mol]
Na2SnO3
[kcal/mol]
K3VO4
[kcal/mol]
K3PO4
[kcal/mol]
413 K 34.0 36.0 28.1
423 K 32.3 33.5 27.2
433 K 31.8 30.8 26.0
443 K 28.9 29.2 24.0
As seen from Table 5, generally, the barriers decrease with increasing temperature,
and for K
3
PO
4
, the barriers are lowest. However, chain propagation can proceed via this
pathway without the starter ROH, with a higher barrier and more probability of CO
2
liber-
ation than the path including the starter ROH. This finding agrees with the experimental
measurements that the majority of CO
2
liberation is caused by cEC decomposition at the
beginning of the reaction. In all paths, one reaction coordinate (CV2) is responsible for the
ring opening of cEC to convert it to a nucleophile (COO
−
) that can further proceed with
the ring opening of the next cEC (propagation).
As an alternative possibility, the insertion of the cEC monomer to the O(CAT)-CH
2
(cEC)
bond, which is already formed in the attached catalyst–monomer complex, is examined.
With the insertion paths, we estimate if a cEC can perform as a nucleophilic species to cleave
the O(CAT)-CH
2
(cEC) bond. The insertion paths are shown in Figure 10. In the insertion
paths Iand II, we try to cleave the cEC ring and convert it to a nucleophile. However, due
to the inherent weak nucleophilicity of cEC by itself, CO
2
release alongside the insertion of
cEC into the O(CAT)-CH
2
bond is observed with a predominant possibility (in 80% of the
trajectories).
This leads us to the conclusion that cEC is inherently a more electrophilic species than
nucleophilic, especially when it interacts with the cations in the surrounding environment
that pre-activate the cEC ring. These observations introduce an alternative pathway to the
previously proposed mechanism, in which cEC attacks as a nucleophile to a second cEC
(Scheme 6) [6].
Polymers 2024, 16, x FOR PEER REVIEW 17 of 27
As an alternative possibility, the insertion of the cEC monomer to the O(CAT)-
CH
2
(cEC) bond, which is already formed in the attached catalyst–monomer complex, is
examined. With the insertion paths, we estimate if a cEC can perform as a nucleophilic
species to cleave the O(CAT)-CH
2
(cEC) bond. The insertion paths are shown in Figure 10.
In the insertion paths I and II, we try to cleave the cEC ring and convert it to a nucleophile.
However, due to the inherent weak nucleophilicity of cEC by itself, CO
2
release alongside
the insertion of cEC into the O(CAT)-CH
2
bond is observed with a predominant possibility
(in 80% of the trajectories).
This leads us to the conclusion that cEC is inherently a more electrophilic species than
nucleophilic, especially when it interacts with the cations in the surrounding environment
that pre-activate the cEC ring. These observations introduce an alternative pathway to the
previously proposed mechanism, in which cEC attacks as a nucleophile to a second cEC
(Scheme 6) [6].
Scheme 6. The previously proposed mechanism of cEC polymerisation, in which cEC attacks as a
nucleophile to a second cEC [6].
Figure 10. The computationally examined insertion paths starting from the attached CAT-cEC mo-
lecular complex, considering the second cEC as a nucleophile. The CO
2
release is the predominant
possibility when the cEC is considered a nucleophile.
In insertion path III, the S
N
2 mechanism via the nucleophilic attack by the O(CAT) to
the CH
2
moiety of a neighbouring cEC is examined. We have run the insertion paths for
Scheme 6. The previously proposed mechanism of cEC polymerisation, in which cEC attacks as a
nucleophile to a second cEC [6].
Polymers 2024,16, 136 15 of 24
Polymers 2024, 16, x FOR PEER REVIEW 17 of 27
As an alternative possibility, the insertion of the cEC monomer to the O(CAT)-
CH
2
(cEC) bond, which is already formed in the attached catalyst–monomer complex, is
examined. With the insertion paths, we estimate if a cEC can perform as a nucleophilic
species to cleave the O(CAT)-CH
2
(cEC) bond. The insertion paths are shown in Figure 10.
In the insertion paths I and II, we try to cleave the cEC ring and convert it to a nucleophile.
However, due to the inherent weak nucleophilicity of cEC by itself, CO
2
release alongside
the insertion of cEC into the O(CAT)-CH
2
bond is observed with a predominant possibility
(in 80% of the trajectories).
This leads us to the conclusion that cEC is inherently a more electrophilic species than
nucleophilic, especially when it interacts with the cations in the surrounding environment
that pre-activate the cEC ring. These observations introduce an alternative pathway to the
previously proposed mechanism, in which cEC attacks as a nucleophile to a second cEC
(Scheme 6) [6].
Scheme 6. The previously proposed mechanism of cEC polymerisation, in which cEC attacks as a
nucleophile to a second cEC [6].
Figure 10. The computationally examined insertion paths starting from the attached CAT-cEC mo-
lecular complex, considering the second cEC as a nucleophile. The CO
2
release is the predominant
possibility when the cEC is considered a nucleophile.
In insertion path III, the S
N
2 mechanism via the nucleophilic attack by the O(CAT) to
the CH
2
moiety of a neighbouring cEC is examined. We have run the insertion paths for
Figure 10. The computationally examined insertion paths starting from the attached CAT-cEC
molecular complex, considering the second cEC as a nucleophile. The CO
2
release is the predominant
possibility when the cEC is considered a nucleophile.
In insertion path III, the S
N
2 mechanism via the nucleophilic attack by the O(CAT)
to the CH
2
moiety of a neighbouring cEC is examined. We have run the insertion paths
for all three catalysts at four temperatures and, after analysing all trajectories, besides
CO
2
release, two new pathways are revealed based on the MTD trajectories. We note
that these new revealed paths were unknown. In one observed path, the second O of the
PO
43−
anion attacks the CH
2
of the second cEC, which means that chain growth can occur
through a new channel via the second cEC cleavage. In the second observed path, the
catalyst detachment for SnO
32−
and VO
43−
occurred. The two new observed paths are
shown in Figure 11A,B. Path A was mainly observed for phosphate and path B for stannate
and vanadate catalysts. The variation of the CVs versus the time evolution is depicted in
Figure 12 for the trajectories with the two alternative paths. In addition to the CVs, the
variation of the non-CV distances that lead to the new-unknown paths in Figure 11 is also
depicted in Figure 12. As can be seen in Figure 12, alternative bond formation/dissociation
can be probed in these trajectories instead of the considered CVs. Figure 12 upper part
shows the second O(CAT)-CH
2
bond is formed as soon as the cEC ring is cleaved (blue
line). As shown for vanadate and stannate, two new bonds between the O(CAT)
· · ·
CH
2+
and O
−
(cEC)
· · ·
central atom (V or Sn) are formed when the ring is cleaved (blue and
orange lines, respectively). Observation of the paths shown in Figure 11 motivated us
to examine the possibility of these methods of polymerisation and for the chain transfer
(catalyst detachment) for the three catalysts throughout the calculation of the corresponding
MD trajectories and ∆G=values.
Table 6. The barriers of the depicted pathways in Figure 13 at 423 K.
Catalyst/∆G=[kcal/mol]
T = 423 K
K3PO4
[kcal/mol]
Na2SnO3
[kcal/mol]
K3VO4
[kcal/mol]
Second nucleophilic attack 20.0 27.6 30.0
Chain transfer—catalyst
detachment 42.7 34.0 36.1
Polymers 2024,16, 136 16 of 24
Polymers 2024, 16, x FOR PEER REVIEW 18 of 27
all three catalysts at four temperatures and, after analysing all trajectories, besides CO
2
release, two new pathways are revealed based on the MTD trajectories. We note that these
new revealed paths were unknown. In one observed path, the second O of the PO
43−
anion
attacks the CH
2
of the second cEC, which means that chain growth can occur through a
new channel via the second cEC cleavage. In the second observed path, the catalyst de-
tachment for SnO
32−
and VO
43−
occurred. The two new observed paths are shown in Figure
11A,B. Path A was mainly observed for phosphate and path B for stannate and vanadate
catalysts. The variation of the CVs versus the time evolution is depicted in Figure 12 for
the trajectories with the two alternative paths. In addition to the CVs, the variation of the
non-CV distances that lead to the new-unknown paths in Figure 11 is also depicted in
Figure 12. As can be seen in Figure 12, alternative bond formation/dissociation can be
probed in these trajectories instead of the considered CVs. Figure 12 upper part shows the
second O(CAT)-CH
2
bond is formed as soon as the cEC ring is cleaved (blue line). As
shown for vanadate and stannate, two new bonds between the O(CAT)…CH
2+
and
O
−
(cEC)…central atom (V or Sn) are formed when the ring is cleaved (blue and orange
lines, respectively). Observation of the paths shown in Figure 11 motivated us to examine
the possibility of these methods of polymerisation and for the chain transfer (catalyst de-
tachment) for the three catalysts throughout the calculation of the corresponding MD tra-
jectories and ΔG
≠
values.
Figure 11. (A) The observed pathway alongside the MTD simulation (instead of the CVs showed in
Figure 10); the second oxygen of the catalyst attacks the second cEC, and the chain propagation can
also go via the second monomer. (B) The second observed path shows the catalyst detachment from
the partially formed chain.
Figure 11. (A) The observed pathway alongside the MTD simulation (instead of the CVs showed in
Figure 10); the second oxygen of the catalyst attacks the second cEC, and the chain propagation can
also go via the second monomer. (B) The second observed path shows the catalyst detachment from
the partially formed chain.
Polymers 2024, 16, x FOR PEER REVIEW 20 of 27
Figure 12. The variation of the CVs and non-CV distances alongside the simulation of the insertion
paths (Figure 11). Variation of the non-CV distances triggered us to consider the new pathways of
catalyst detachment and chain transfer, as shown in Figure 13. CV2 is the CH
2
-O(ethereal) distance
inside the cEC, which indicates the cEC cleavage.
Since the barriers of the catalyst’s second oxygen attack are lower than the catalyst
detachment (Table 6), the second nucleophilic attack begins faster and can promote the
chain transfer/catalyst detachment. This finding means that with the formation of the sec-
ond cleaved cEC and the start of a new polymerisation channel, the older generated chain
leaves the catalyst, leading to polymer chains with specific shorter lengths and constant
molecular weights.
Figure 12. Cont.
Polymers 2024,16, 136 17 of 24
Polymers 2024, 16, x FOR PEER REVIEW 20 of 27
Figure 12. The variation of the CVs and non-CV distances alongside the simulation of the insertion
paths (Figure 11). Variation of the non-CV distances triggered us to consider the new pathways of
catalyst detachment and chain transfer, as shown in Figure 13. CV2 is the CH
2
-O(ethereal) distance
inside the cEC, which indicates the cEC cleavage.
Since the barriers of the catalyst’s second oxygen attack are lower than the catalyst
detachment (Table 6), the second nucleophilic attack begins faster and can promote the
chain transfer/catalyst detachment. This finding means that with the formation of the sec-
ond cleaved cEC and the start of a new polymerisation channel, the older generated chain
leaves the catalyst, leading to polymer chains with specific shorter lengths and constant
molecular weights.
Figure 12. The variation of the CVs and non-CV distances alongside the simulation of the insertion
paths (Figure 11). Variation of the non-CV distances triggered us to consider the new pathways of
catalyst detachment and chain transfer, as shown in Figure 13. CV2 is the CH
2
-O(ethereal) distance
inside the cEC, which indicates the cEC cleavage.
To calculate the
∆
G
=
for the two observed paths, we considered the appropriate
CVs and simulated the MTD trajectories at 423 K. These CVs are shown in Figure 13,
and the calculated
∆
G
=
values are reported in Table 6. The
∆
G
=
values correlate to the
rate-determining step, corresponding to the nucleophilic attack by the O
−
(CV1) in each
pathway. In the chain transfer/catalyst-detachment path in Figure 13, we performed the
transfer of a partially formed chain with an external nucleophilic attack (RO
−
). As seen
in Table 6, the catalyst’s second oxygen attack has a lower barrier in the case of K
3
PO
4
than K
3
VO
4
and Na
2
SnO
3
, in agreement with the stronger nucleophilicity of the phosphate
anion.
The chain transfer through the catalyst detachment has a high barrier, which indicates
that the generated chain is firmly attached to the catalyst. Nevertheless, chain transfer
occurs at a specific time during the growth period of each individual chain.
Since the barriers of the catalyst’s second oxygen attack are lower than the catalyst
detachment (Table 6), the second nucleophilic attack begins faster and can promote the
chain transfer/catalyst detachment. This finding means that with the formation of the
second cleaved cEC and the start of a new polymerisation channel, the older generated
chain leaves the catalyst, leading to polymer chains with specific shorter lengths and
constant molecular weights.
Polymers 2024,16, 136 18 of 24
Polymers 2024, 16, x FOR PEER REVIEW 21 of 27
Figure 13. Upper part: is the nucleophilic attack by the second O
−
of the catalyst anion to the second
cEC, lower part: is the chain transfer and catalyst detachment path. The corresponding barriers for
these trajectories are shown in Table 6.
These insights obtained using calculations are well in line with the experimental ob-
servations and results. The calculated rate constants of the chain transfer and catalyst-
detachment (k
trans
), according to the kinetic measurements in the experiment, are reported
in Table 7 compared to the rate constants of the ring-opening polymerisation and the prop-
agation with CO
2
release (Tables 3 and 4). As seen in Table 7, the k
trans
(chain transfer rate
constant) is approximately an order of magnitude smaller than the k
1
and k
2
(the rate con-
stants of polymerisation). This observation agrees with the calculated high ΔG
≠
values for
catalyst detachment in Table 6.
For the reaction with the catalyst systems K
2
SnO
3
and K
3
PO
4
, LC-ESI-MS orbitrap
mass spectrometry measurements were made from samples withdrawn during the reac-
tion at defined time intervals. These mass spectra are shown in Figures 14 and 15.
Figure 13. Upper part: is the nucleophilic attack by the second O
−
of the catalyst anion to the second
cEC, lower part: is the chain transfer and catalyst detachment path. The corresponding barriers for
these trajectories are shown in Table 6.
These insights obtained using calculations are well in line with the experimental
observations and results. The calculated rate constants of the chain transfer and catalyst-
detachment (k
trans
), according to the kinetic measurements in the experiment, are reported
in Table 7compared to the rate constants of the ring-opening polymerisation and the
propagation with CO
2
release (Tables 3and 4). As seen in Table 7, the k
trans
(chain transfer
rate constant) is approximately an order of magnitude smaller than the k
1
and k
2
(the rate
constants of polymerisation). This observation agrees with the calculated high
∆
G
=
values
for catalyst detachment in Table 6.
Table 7. Experimentally measured rate constants for the chain transfer (k
trans
) compared to the
polymerisation and CO2release rate constant for the stannate and vanadate catalysts.
K2SnO3Na3VO4
T (K) k1k2ktrans k1k2Ktrans
413 0.029 0.026 0.0037 0.0349 0.0346 0.0061
423 0.029 0.028 0.0034 0.0524 0.0534 0.0092
433 0.038 0.034 0.0052 0.1070 0.1157 0.0196
443 0.032 0.034 0.0073 0.2071 0.2431 0.0411
Polymers 2024,16, 136 19 of 24
For the reaction with the catalyst systems K
2
SnO
3
and K
3
PO
4
, LC-ESI-MS orbitrap
mass spectrometry measurements were made from samples withdrawn during the reaction
at defined time intervals. These mass spectra are shown in Figures 14 and 15.
Polymers 2024, 16, x FOR PEER REVIEW 22 of 27
Figure 14. Plot of LC-ESI-MS spectra over time for monitoring the development of the molecular
weight of the polymer using K2SnO3 as the catalyst.
Figure 15. Plot of LC-ESI-MS spectra over time for monitoring the development of the molecular
weight of the polymer using K3PO4 as the catalyst.
It can be observed from Figures 14 and 15 that after a short reaction time, the masses
of the polymers primarily correspond to the masses present at the end of the reaction.
Substantial changes in the molecular weight during the progress of the reaction were not
observed, indicating no influence of the decreasing monomer concentration on the degree
of polymerisation. Therefore, the ring-opening polymerisation reaction behaves similarly
to a free radical polymerisation with strong transfer to monomer. Such polymerisation
reactions are primarily defined by initiation, growth, termination, and transfer reactions.
The termination itself occurs through combination and transfer reactions. If the transfer
step is the predominant step for the termination of the growth of the individual chain, the
average degree of polymerisation is given by the ratio of the propagation rate to the trans-
Figure 14. Plot of LC-ESI-MS spectra over time for monitoring the development of the molecular
weight of the polymer using K2SnO3as the catalyst.
Polymers 2024, 16, x FOR PEER REVIEW 22 of 27
Figure 14. Plot of LC-ESI-MS spectra over time for monitoring the development of the molecular
weight of the polymer using K2SnO3 as the catalyst.
Figure 15. Plot of LC-ESI-MS spectra over time for monitoring the development of the molecular
weight of the polymer using K3PO4 as the catalyst.
It can be observed from Figures 14 and 15 that after a short reaction time, the masses
of the polymers primarily correspond to the masses present at the end of the reaction.
Substantial changes in the molecular weight during the progress of the reaction were not
observed, indicating no influence of the decreasing monomer concentration on the degree
of polymerisation. Therefore, the ring-opening polymerisation reaction behaves similarly
to a free radical polymerisation with strong transfer to monomer. Such polymerisation
reactions are primarily defined by initiation, growth, termination, and transfer reactions.
The termination itself occurs through combination and transfer reactions. If the transfer
step is the predominant step for the termination of the growth of the individual chain, the
average degree of polymerisation is given by the ratio of the propagation rate to the trans-
Figure 15. Plot of LC-ESI-MS spectra over time for monitoring the development of the molecular
weight of the polymer using K3PO4as the catalyst.
It can be observed from Figures 14 and 15 that after a short reaction time, the masses
of the polymers primarily correspond to the masses present at the end of the reaction.
Substantial changes in the molecular weight during the progress of the reaction were not
observed, indicating no influence of the decreasing monomer concentration on the degree
of polymerisation. Therefore, the ring-opening polymerisation reaction behaves similarly
to a free radical polymerisation with strong transfer to monomer. Such polymerisation
reactions are primarily defined by initiation, growth, termination, and transfer reactions.
The termination itself occurs through combination and transfer reactions. If the transfer
step is the predominant step for the termination of the growth of the individual chain,
the average degree of polymerisation is given by the ratio of the propagation rate to the
transfer rate. If both reactions only depend on catalyst and monomer concentration, the
Polymers 2024,16, 136 20 of 24
concentrations cancel out, and Pn is given by the propagation rate constant to the transfer
rate constant (see Equation (5)).
Pn =
Mpolymer
MMonomer
=
kprop.
ktransfer (5)
Using LC-ESI-MS orbitrap mass spectrometry, the number average molecular weights
of the polymers can be obtained. Since an ethylene oxide unit (EO) is always incorporated
into the polymer chain during a propagation step, the mass of CO
2
can be subtracted from
the number average molar mass (M
n
) after determining the CO
2
content via
1
H-NMR.
These masses can be used to calculate the degree of polymerisation via dividing them by
44 g/mol (M ethylene oxide).
Since the propagation rate is defined by the reaction rate constants k
1
and k
2
, which
describe the propagation with and without CO
2
incorporation, resulting in the CO
2
selec-
tivity, these constants can be added to the rate constant of chain growth (k
w
). Using these
LC-ESI-MS data and Equation (5), the rate constant for the transfer step can be determined,
as shown in Table 7.
Plotting the degree of polymerisation against the catalyst concentration, it is apparent
that the degree of polymerisation is constant within the experimental error. Since the
catalysts are bases, this agrees with the observations of Rokicki et al. [
6
]. Furthermore,
the calculated barriers of nucleophilic attack by the catalyst anions to the partially formed
polymer chains are 25 and 38 kcal/mol for PO
43−
and SnO
32−
, respectively, supporting
chain degradation at high temperatures triggered by an increased rate of back binding and
acceleration of the transfer reaction. See Figure 16.
Polymers 2024, 16, x FOR PEER REVIEW 23 of 27
fer rate. If both reactions only depend on catalyst and monomer concentration, the con-
centrations cancel out, and Pn is given by the propagation rate constant to the transfer rate
constant (see Equation (5)).
Pn = M
M
=k.
k
(5)
Using LC-ESI-MS orbitrap mass spectrometry, the number average molecular
weights of the polymers can be obtained. Since an ethylene oxide unit (EO) is always in-
corporated into the polymer chain during a propagation step, the mass of CO2 can be sub-
tracted from the number average molar mass (Mn) after determining the CO2 content via
1H-NMR. These masses can be used to calculate the degree of polymerisation via dividing
them by 44 g/mol (M ethylene oxide).
Since the propagation rate is defined by the reaction rate constants k1 and k2, which
describe the propagation with and without CO2 incorporation, resulting in the CO2 selec-
tivity, these constants can be added to the rate constant of chain growth (kw). Using these
LC-ESI-MS data and Equation (5), the rate constant for the transfer step can be deter-
mined, as shown in Table 7.
Plotting the degree of polymerisation against the catalyst concentration, it is apparent
that the degree of polymerisation is constant within the experimental error. Since the cat-
alysts are bases, this agrees with the observations of Rokicki et al. [6]. Furthermore, the
calculated barriers of nucleophilic attack by the catalyst anions to the partially formed
polymer chains are 25 and 38 kcal/mol for PO43− and SnO32−, respectively, supporting chain
degradation at high temperatures triggered by an increased rate of back binding and ac-
celeration of the transfer reaction. See Figure 16.
Figure 16. Degree of polymerisation at different concentrations of catalyst. A significant reduction
in the degree of polymerisation can be observed by increasing the catalyst concentration and thereby
increasing the basicity. Tested system Starter: ethylene glycol (0.1 eq.) Catalyst: K2SnO3 T: 150 °C.
By plotting the rate constants of the propagation and transfer reaction (Figure 17), it
can be seen that the rate constant of the transfer reaction is smaller than the growth reac-
tion by a factor of about 6 to 10. This difference makes the reaction possible in the first
Figure 16. Degree of polymerisation at different concentrations of catalyst. A significant reduction in
the degree of polymerisation can be observed by increasing the catalyst concentration and thereby
increasing the basicity. Tested system Starter: ethylene glycol (0.1 eq.) Catalyst: K2SnO3T: 150 ◦C.
By plotting the rate constants of the propagation and transfer reaction (Figure 17), it
can be seen that the rate constant of the transfer reaction is smaller than the growth reaction
by a factor of about 6 to 10. This difference makes the reaction possible in the first place but
inevitably reduces the degree of polymerisation. This finding explains the variations in the
molecular weights with different catalysts already observed in the literature [6,13].
Polymers 2024,16, 136 21 of 24
Polymers 2024, 16, x FOR PEER REVIEW 24 of 27
place but inevitably reduces the degree of polymerisation. This finding explains the vari-
ations in the molecular weights with different catalysts already observed in the literature
[6,13].
Table 7. Experimentally measured rate constants for the chain transfer (ktrans) compared to the
polymerisation and CO2 release rate constant for the stannate and vanadate catalysts.
K2SnO3 Na3VO4
T (K) k1 k2 ktrans k1 k2 Ktrans
413 0.029 0.026 0.0037 0.0349 0.0346 0.0061
423 0.029 0.028 0.0034 0.0524 0.0534 0.0092
433 0.038 0.034 0.0052 0.1070 0.1157 0.0196
443 0.032 0.034 0.0073 0.2071 0.2431 0.0411
Figure 17. Left: Arrhenius plots of the growth reactions for the incorporation and non-incorporation
of CO2 and the termination reaction (transfer) when using sodium orthovanadate (Na3VO4). The
system used: catalyst (0.01 eq.), ethylene glycol (0.10 eq.), cyclic ethylene carbonate (1.00 eq.). Right:
Arrhenius plot of the growth reactions for the incorporation and non-incorporation of CO2 and the
termination reaction (transfer) when using potassium stannate (K2SnO3). The system used: catalyst
(0.01 eq.), ethylene glycol (0.10 eq.), cyclic ethylene carbonate (1.00 eq.).
4. Conclusions
In this work, we have investigated alternative paths of the ring-opening polymerisa-
tion of cEC monomer with and without CO2 release, originating from the catalyst–mono-
mer interaction in the propagation step. The kinetic measurements, in combination with
metadynamics simulations, indicate that the mechanism of chain growth (starting from
the catalyst–monomer molecular complex with the catalyst attached to the cleaved cEC
monomer) has a higher barrier with a more significant probability for CO2 release than
the two-step mechanism involving the ROH starter molecule in the initiation step. We
examined the insertion of the cEC monomer as a nucleophile into the CAT-CH2 molecular
complex and the detachment of the catalyst from the chain, followed by the chain transfer.
The barriers of chain transfer are high; this finding is in accordance with the increase of
the possibility of CO2 release throughout the cEC ring cleavage. Our mechanistic investi-
gations confirmed that the cEC monomer can rarely be considered as a nucleophile, but
instead predominantly as an electrophilic species, which is in accordance with the anionic
ROP of cEC proposed in the literature. These computational findings are supported by
the experimental kinetic measurements and the results that the calculated rate constants
of the chain transfer and catalyst detachment are almost an order of magnitude higher
than the chain propagation and CO2 release. The calculated barriers confirm that the nu-
cleophilic attack throughout the second anionic oxygen of the catalyst anion can promote
Figure 17. Left: Arrhenius plots of the growth reactions for the incorporation and non-incorporation of
CO2and the termination reaction (transfer) when using sodium orthovanadate (Na3VO4). The system
used: catalyst (0.01 eq.), ethylene glycol (0.10 eq.), cyclic ethylene carbonate (1.00 eq.). Right: Arrhenius
plot of the growth reactions for the incorporation and non-incorporation of CO
2
and the termination
reaction (transfer) when using potassium stannate (K
2
SnO
3
). The system used: catalyst (0.01 eq.),
ethylene glycol (0.10 eq.), cyclic ethylene carbonate (1.00 eq.).
4. Conclusions
In this work, we have investigated alternative paths of the ring-opening polymerisa-
tion of cEC monomer with and without CO
2
release, originating from the catalyst–monomer
interaction in the propagation step. The kinetic measurements, in combination with metady-
namics simulations, indicate that the mechanism of chain growth (starting from the catalyst–
monomer molecular complex with the catalyst attached to the cleaved cEC monomer) has
a higher barrier with a more significant probability for CO
2
release than the two-step
mechanism involving the ROH starter molecule in the initiation step. We examined the
insertion of the cEC monomer as a nucleophile into the CAT-CH
2
molecular complex and
the detachment of the catalyst from the chain, followed by the chain transfer. The barriers
of chain transfer are high; this finding is in accordance with the increase of the possibility of
CO
2
release throughout the cEC ring cleavage. Our mechanistic investigations confirmed
that the cEC monomer can rarely be considered as a nucleophile, but instead predominantly
as an electrophilic species, which is in accordance with the anionic ROP of cEC proposed
in the literature. These computational findings are supported by the experimental kinetic
measurements and the results that the calculated rate constants of the chain transfer and
catalyst detachment are almost an order of magnitude higher than the chain propagation
and CO
2
release. The calculated barriers confirm that the nucleophilic attack throughout
the second anionic oxygen of the catalyst anion can promote chain transfer and catalyst de-
tachment. This realisation means that with the formation of the second cleaved cEC and the
start of a new polymerisation channel, the former generated chain leaves the catalyst. This
transfer can lead to polymer chains with specific shorter lengths and keeps the molecular
weights constant. The results obtained in this study can help in designing novel catalysts
with improved characteristics to produce polymers with higher CO
2
content. For higher
molecular weights of the polymers, catalysts with higher basicity should be applied, with a
lower tendency to detach from the growing chain.
Supplementary Materials: The following supporting information can be downloaded at: https://
www.mdpi.com/article/10.3390/polym16010136/s1. Figures S1–S3 show the variation of the two
CVs versus each other for the three catalysts at four temperatures; Figures S4–S6 show the variation
of CV1 and CV2 versus time corresponding to Figure 3in the main text; Figures S7–S9 show the
variations of the CV1 and CV2 for the MD trajectories corresponding to Figure 9in the main text;
Polymers 2024,16, 136 22 of 24
Figure S10 shows the variation of the CV1 and CV2 for the trajectories of the upper part of Figure 13
and Figure S11 shows the variation of the CV1 and CV2 for the lower part of Figure 13; Table S1
shows experimental measurements and rate constants regarding vanadate catalyst. From Section
S3 onwards, the supporting information for kinetic measurements is presented. The reactor used
is shown in Figure S12. The experimental parameters used are shown in Tables S2–S6. Section S4
deals with the results of the simulation, showing both the fit of the calculated adsorption of the
kinetic model to the experimental values and the resulting concentration-time curves for the catalysts
stannates (Figure S13), phosphates (Figure S14), and orthovanadates (Figure S15). Section S5 shows
the NMR spectra of the
1
H-NMR of the products of the kinetic measurement. The NMR spectra for
stannates are in Figures S16–S19 for phosphates, Figures S20–S23, and for the orthovanadate catalyst,
Figures S24–S27.
Author Contributions: Conceptualization, D.B. and M.H.; methodology, D.B. and M.H.; software, D.B.
and M.H.; validation, D.B. and M.H.; formal analysis, D.B. and M.H.; investigation, D.B. and M.H.;
resources, R.S. and M.R.M.; writing—original draft preparation, D.B. and M.H.; writing—review and
editing, R.S. and M.R.M.; visualization, D.B. and M.H.; supervision, R.S.; project administration, R.S.;
funding acquisition, R.S. All authors have read and agreed to the published version of the manuscript.
Funding: We gratefully acknowledge the RWTH-HPC (project number p0020212) for computation
time and technical support. This research work has been carried out within the project DreamRe-
sourceConti (033R222A), which has received public funding from the BMBF (Bundesminustrerim
für Bildung und Forschung). Technical University of Berlin work has been carried out within the
project “Dream ResourceConti” (033R222C). The German Federal Ministry of Education and Research
(BMBF) funded the project within the funding priority “CO
2
Plus—Stoffliche Nutzung von CO
2
zur
Verbesserung der Rohstoffbasis”.
Data Availability Statement: The data will be made available upon request.
Acknowledgments: We gratefully thank the BMBF for its financial support.
Conflicts of Interest: The authors declare no conflicts of interest.
References
1.
Quadrelli, E.A.; Armstrong, K.; Styring, P. Carbon Dioxide Utilisation, 1st ed.; Elsevier: Amsterdam, The Netherlands, 2015;
ISBN 9780444627469.
2.
Machat, M.R.; Marbach, J.; Schumacher, H.; Raju, S.; Lansing, M.; Over, L.C.; Adler, L.; Langanke, J.; Wolf, A.; Leitner, W.; et al.
Turning CO/CO
2
-containing industrial process gas into valuable building blocks for the polyurethane industry. React. Chem. Eng.
2022,7, 580–589. [CrossRef]
3.
Langanke, J.; Wolf, A.; Hofmann, J.; Böhm, K.; Subhani, M.A.; Müller, T.E.; Leitner, W.; Gürtler, C. Carbon dioxide (CO
2
) as
sustainable feedstock for polyurethane production. Green Chem. 2014,16, 1865–1870. [CrossRef]
4.
Wolf, A.; Grasser, S.; Gürtler, C.; Hofmann, J. Verfahren zur Herstellung von Polyethercarbonatpolyolen Mit Doppelmetallcyanid-
katalysatoren und in Gegenwart von Metallsalzen. WO Patent WO2011101403, 25 August 2011.
5.
Gürtler, C.; Hofmann, J.; Müller, T.E.; Wolf, A.; Grasser, S.; Köhler, B. Verfahren zur Herstellung von Polyethercarbonatpolyolen.
WO Patent WO2011117332, 29 September 2011.
6. Rokicki, G. Aliphatic cyclic carbonates and spiroorthocarbonates as monomers. Prog. Polym. Sci. 2000,25, 259–342. [CrossRef]
7.
Rokicki, G.; Piotrowska, A. A new route to polyurethanes from ethylene carbonate, diamines and diols. Polymer 2002,43,
2927–2935. [CrossRef]
8.
Takata, T.; Endo, T. Recent advances in the development of expanding monomers: Synthesis, polymerization and volume change.
Prog. Polym. Sci. 1993,18, 839–870. [CrossRef]
9.
Bailey, W.J.; Sun, R.L. Polymer Preprints (American Chemical Society, Division of Polymer Chemistry); American Chemical Society,
Division of Polymer Chemistry: Dallas, TX, USA, 1972; pp. 281–286, ISBN 0032-3934.
10.
Brady, R.F. Volume Change in the Polymerization of Spirocyclic Monomers: Causes, Effects, and Applications. J. Macromol. Sci.
Part C Polym. Rev. 1992,32, 135–181. [CrossRef]
11.
Sadhir, R.; Luck, R. Expanding monomers. In Expanding Monomers; CRC Press: Boca Raton, FL, USA, 1992; p. 152, ISBN 9780849351563.
12.
Lopes, E.; Ribeiro, A.; Martins, L. New Trends in the Conversion of CO
2
to Cyclic Carbonates. Catalysts 2020,10, 479. [CrossRef]
13.
von Seggern, N.; Schindler, T.; Naumann, S. Dual Catalytic Ring-Opening Polymerization of Ethylene Carbonate for the
Preparation of Degradable PEG. Biomacromolecules 2020,21, 2661–2669. [CrossRef]
14.
Harris, R.F. Structural features of poly(alkylene ether carbonate) diol oligomers by capillary gas chromatography. J. Appl. Polym.
Sci. 1989,37, 183–200. [CrossRef]
15.
Harris, R.F. Effect of catalyst on the molecular weight advancement of poly(ethylene ether carbonate) polyols. J. Appl. Polym. Sci.
1990,40, 1265–1279. [CrossRef]
Polymers 2024,16, 136 23 of 24
16.
Yadav, N.; Seidi, F.; Crespy, D.; D’Elia, V. Polymers Based on Cyclic Carbonates as Trait d’Union Between Polymer Chemistry and
Sustainable CO2Utilization. ChemSusChem 2019,12, 724–754. [CrossRef]
17.
Lee, J.-C.; Litt, M.H. Ring-Opening Polymerization of Ethylene Carbonate and Depolymerization of Poly(ethylene oxide-co-
ethylene carbonate). Macromolecules 2000,33, 1618–1627. [CrossRef]
18.
Kiesewetter, M.K.; Shin, E.J.; Hedrick, J.L.; Waymouth, R.M. Organocatalysis: Opportunities and Challenges for Polymer Synthesis.
Macromolecules 2010,43, 2093–2107. [CrossRef]
19.
Jang, J.H.; Ha, J.H.; Kim, I.; Baik, J.H.; Hong, S.C. Facile Room-Temperature Preparation of Flexible Polyurethane Foams from
Carbon Dioxide Based Poly(ether carbonate) Polyols with a Reduced Generation of Acetaldehyde. ACS Omega 2019,4, 7944–7952.
[CrossRef] [PubMed]
20.
Schäfer, P.; Ag, C.D. Bauen Mit CO
2
-Technologie Wird Möglich. Available online: https://www.covestro.com/press/de/bauen-
mit-co-technologie-wird-moeglich/ (accessed on 4 October 2023).
21.
Zhang, S.-G.; Wei, Y.-D.; Yin, S.-F.; Luo, S.-L.; Au, C.-T. Superbasic sodium stannate as catalyst for dehydrogenation, Michael
addition and transesterification reactions. Appl. Catal. A Gen. 2011,406, 113–118. [CrossRef]
22.
Haba, O.; Tomizuka, H.; Endo, T. Anionic Ring-Opening Polymerization of Methyl 4,6-O-Benzylidene-2,3-O-carbonyl-
α
-d-
glucopyranoside: A First Example of Anionic Ring-Opening Polymerization of Five-Membered Cyclic Carbonate without
Elimination of CO2.Macromolecules 2005,38, 3562–3563. [CrossRef]
23.
Tezuka, K.; Komatsu, K.; Haba, O. The anionic ring-opening polymerization of five-membered cyclic carbonates fused to the
cyclohexane ring. Polym. J. 2013,45, 1183–1187. [CrossRef]
24.
Piccini, G.; Lee, M.-S.; Yuk, S.F.; Zhang, D.; Collinge, G.; Kollias, L.; Nguyen, M.-T.; Glezakou, V.-A.; Rousseau, R. Ab initio
molecular dynamics with enhanced sampling in heterogeneous catalysis. Catal. Sci. Technol. 2022,12, 12–37. [CrossRef]
25.
Daru, J.; Bakó, I.; Stirling, A.; Pápai, I. Mechanism of Heterolytic Hydrogen Splitting by Frustrated Lewis Pairs: Comparison of
Static and Dynamic Models. ACS Catal. 2019,9, 6049–6057. [CrossRef]
26.
Heshmat, M. Alternative Pathway of CO
2
Hydrogenation by Lewis-Pair-Functionalized UiO-66 MOF Revealed by Metadynamics
Simulations. J. Phys. Chem. C 2020,124, 10951–10960. [CrossRef]
27.
Liu, L.; Lukose, B.; Ensing, B. A Free Energy Landscape of CO
2
Capture by Frustrated Lewis Pairs. ACS Catal. 2018,8, 3376–3381.
[CrossRef]
28.
Ensing, B.; Laio, A.; Parrinello, M.; Klein, M.L. A Recipe for the Computation of the Free Energy Barrier and the Lowest Free
Energy Path of Concerted Reactions. J. Phys. Chem. B 2005,109, 6676–6687. [CrossRef] [PubMed]
29.
Stirling, A.; Iannuzzi, M.; Parrinello, M.; Molnar, F.; Bernhart, V.; Luinstra, G.A.
β
-Lactone Synthesis from Epoxide and CO:
Reaction Mechanism Revisited. Organometallics 2005,24, 2533–2537. [CrossRef]
30.
Piccini, G.; Polino, D.; Parrinello, M. Identifying Slow Molecular Motions in Complex Chemical Reactions. J. Phys. Chem. Lett.
2017,8, 4197–4200. [CrossRef] [PubMed]
31. Ghoussoub, M.; Yadav, S.; Ghuman, K.K.; Ozin, G.A.; Singh, C.V. Metadynamics-Biased ab Initio Molecular Dynamics Study of
Heterogeneous CO2Reduction via Surface Frustrated Lewis Pairs. ACS Catal. 2016,6, 7109–7117. [CrossRef]
32.
Dong, Y.; Ghuman, K.K.; Popescu, R.; Duchesne, P.N.; Zhou, W.; Loh, J.Y.Y.; Jelle, A.A.; Jia, J.; Wang, D.; Mu, X.; et al. Tailoring
Surface Frustrated Lewis Pairs of In
2
O
3−x
(OH)
y
for Gas-Phase Heterogeneous Photocatalytic Reduction of CO
2
by Isomorphous
Substitution of In3+ with Bi3+.Adv. Sci. 2018,5, 1700732. [CrossRef]
33.
Dong, Y.; Duchesne, P.; Mohan, A.; Ghuman, K.K.; Kant, P.; Hurtado, L.; Ulmer, U.; Loh, J.Y.Y.; Tountas, A.A.; Wang, L.; et al.
Shining light on CO
2
: From materials discovery to photocatalyst, photoreactor and process engineering. Chem. Soc. Rev. 2020,49,
5648–5663. [CrossRef]
34.
Brüggemann, D.; Shojamejer, T.; Tupinamba Lima, M.; Zukova, D.; Marschall, R.; Schomäcker, R. The Performance of Carbonate-
Modified Nonionic Surfactants in Microplastic Flotation. Water 2023,15, 1000. [CrossRef]
35.
VandeVondele, J.; Krack, M.; Mohamed, F.; Parrinello, M.; Chassaing, T.; Hutter, J. Quickstep: Fast and accurate density functional
calculations using a mixed Gaussian and plane waves approach. Comput. Phys. Commun. 2005,167, 103–128. [CrossRef]
36.
Perdew, J.P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996,77, 3865–3868.
[CrossRef]
37.
Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A consistent and accurate ab initio parametrization of density functional dispersion
correction (DFT-D) for the 94 elements H-Pu. J. Chem. Phys. 2010,132, 154104. [CrossRef] [PubMed]
38. Laio, A.; Parrinello, M. Escaping free-energy minima. Proc. Natl. Acad. Sci. USA 2002,99, 12562–12566. [CrossRef] [PubMed]
39.
Invernizzi, M.; Parrinello, M. Rethinking Metadynamics: From Bias Potentials to Probability Distributions. J. Phys. Chem. Lett.
2020,11, 2731–2736. [CrossRef] [PubMed]
40.
Bonomi, M.; Branduardi, D.; Bussi, G.; Camilloni, C.; Provasi, D.; Raiteri, P.; Donadio, D.; Marinelli, F.; Pietrucci, F.; Broglia, R.A.; et al.
PLUMED: A portable plugin for free-energy calculations with molecular dynamics. Comput. Phys. Commun. 2009,180, 1961–1972.
[CrossRef]
41. Humphrey, W.; Dalke, A.; Schulten, K. VMD: Visual molecular dynamics. J. Mol. Graph. 1996,14, 33–38. [CrossRef]
42.
Hediger, H.J. Infrarotspektroskopie: Grundlagen, Anwendungen, Interpretation mit zahlreichen Tabellen und 32 Spektren. In
Infrarotspektroskopie: Grundlagen, Anwendungen, Interpretation mit Zahlreichen Tabellen und 32 Spektren; Akademische Verlagsge-
sellschaft: Frankfurt, Germany, 1971; pp. 1–25, 29–169, 173–195, ISBN 3-400-00052-3.
Polymers 2024,16, 136 24 of 24
43.
Cammann, K. Infrarotspektroskopie. In Instrumentelle Analytische Chemie; Verfahren, Anwendungen, Qualitätssicherung; Spektrum
Akademischer Verlag: Heidelberg/Berlin, Germany, 2001; pp. 5–51, ISBN 3-8274-0057-0.
44. Fox, T. Hesse-Meier-Zeeh Infrarot- und Ramanspektren. In Spektroskopische Methoden in der Organischen Chemie; Bienz, S., Bigler,
L., Fox, T., Meier, H., Eds.; George Thieme Verlag: Stuttgart, Germany, 2016; pp. 37–74, ISBN 978-3-13-576109-1.
45.
Schwedt, G. Infrarot- und Raman-Spektroskopie. In Taschenatlas der Analytik; George Thieme Verlag: Stuttgart, Germany, 1992;
pp. 114–121, ISBN 3-13-759301-8.
46.
Heshmat, M.; Leven, M.; Linker, O.; Sebastian, M.; Gürtler, C.; Machat, M.R. A DFT-metadynamics study disclosing key properties
of ring-opening polymerization catalysts to produce polyethercarbonate polyols from cyclic ethylene carbonate as part of an
emerging CCU technology. Phys. Chem. Chem. Phys. 2023,25, 20485–20494. [CrossRef]
47.
Reich, H.; Hans Reich’s Collection. NMR Spectroscopy. Available online: https://organicchemistrydata.org/hansreich/resources/
nmr/?index=nmr_index/31P_shift (accessed on 24 April 2023).
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.
