Catalyst design criteria and fundamental limitations in the electrochemical synthesis of dimethyl carbonate

Article abstract of DOI:10.1039/c9gc02265a

Dimethyl carbonate is an environmentally friendly precursor in various chemical reactions and is currently synthesized by hazardous processes. An electrocatalytic approach could result in a process abiding to the principles of Green Chemistry. Herein we demonstrate how density functional theory (DFT) calculations and experiment advance our understanding of electrocatalytic production of chemicals. Using density functional theory, we form design criteria for dimethyl carbonate electrosynthesis on metallic surfaces. The criteria are based on adsorption free energies of reactants and reaction energies of possible products. The design criteria allow us to identify copper as an interesting candidate for the electrode material as it is classified as being selective to dimethyl carbonate and requires ≈1 V lower potential than a gold electrode. By further addressing electrode stability copper was found to dissolve and produce copper-carbonyl species which lead to dimethyl carbonate as a consequence of a reaction in the solution, therefore not occurring by surface electrocatalysis. This shows that the design criteria presented herein are necessary but not sufficient requirements that the ideal electrode should satisfy.

Full text of DOI:10.1039/c9gc02265a

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Green Chemistry
View Article Online
View Journal | View Issue
PAPER
Catalyst design criteria and fundamental
limitations in the electrochemical synthesis
of dimethyl carbonate
Cite this: Green Chem., 2019, 21,
6200
Manuel Šarić,†a Bethan Jane Venceslau Davies, †a Niels Christian Schjødt,b
Søren Dahl,b Poul Georg Moses,b María Escudero-Escribano, a Matthias Arenz
and Jan Rossmeisl*a
c
Dimethyl carbonate is an environmentally friendly precursor in various chemical reactions and is currently
synthesized by hazardous processes. An electrocatalytic approach could result in a process abiding to the
principles of Green Chemistry. Herein we demonstrate how density functional theory (DFT) calculations
and experiment advance our understanding of electrocatalytic production of chemicals. Using density
functional theory, we form design criteria for dimethyl carbonate electrosynthesis on metallic surfaces.
The criteria are based on adsorption free energies of reactants and reaction energies of possible products.
The design criteria allow us to identify copper as an interesting candidate for the electrode material as it is
classified as being selective to dimethyl carbonate and requires ≈1 V lower potential than a gold elec-
trode. By further addressing electrode stability copper was found to dissolve and produce copper-carbo-
nyl species which lead to dimethyl carbonate as a consequence of a reaction in the solution, therefore
not occurring by surface electrocatalysis. This shows that the design criteria presented herein are necess-
ary but not sufficient requirements that the ideal electrode should satisfy.
Received 3rd July 2019,
Accepted 24th October 2019
DOI: 10.1039/c9gc02265a
rsc.li/greenchem
pounds such as phosgene, currently used in various indust-
rially important chemical processes. DMC meets the increas-
1 Introduction
Today most chemicals are derived from oil which also provides ing demand in today’s chemical industry for less hazardous
the necessary energy to drive chemical synthesis reactions. Oil versatile reagents. Thus, DMC constitutes an important
is a limited resource and at some point we will need to move example of a chemical, whose use adheres to the principles of
to a sustainable way of producing chemicals. Therefore, Green Chemistry.3–7 DMC is mostly used as a precursor in
electrochemical and electrocatalytic production of chemicals chemical synthesis although it can also be used as a solvent,
receives increasing attention. The energy to drive reactions can electrolyte, fuel or fuel additive.1,2,8–12
be obtained from renewable sources (e.g. wind or solar power).
The chemical methods of producing DMC often involve
If an abundant feedstock is used, very attractive alternatives to toxic precursors or waste products and/or explosive mixtures
standard processes seem possible. Prominent examples are of carbon monoxide and oxygen. Development of a new
water and carbon dioxide electrolysis to produce fuels. green chemistry production method for DMC is desirable.
However, at such scale the technology is not yet mature Electrocatalysis offers an alternative, Green synthesis method
enough to compete with fossil fuels and therefore relies on for the production of chemicals without the use of chemical
economical subsidies.
oxidants. The reaction energy can be controlled by the poten-
Dimethyl carbonate (DMC) is an environmentally benign tial which means that energy rich, hazardous reaction con-
and versatile chemical1,2 with potential to replace toxic com- ditions can be avoided. An electrochemical method of synthe-
sising DMC was reported on gold by Funakawa et al.:13,14
aDepartment of Chemistry, University of Copenhagen, Universitetsparken 5,
2CH3OH þ CO Ð CH3OCOOCH3 þ 2ðHþ þ eꢀÞ
ð1Þ
2100 Copenhagen, Denmark. E-mail: Jan.Rossmeisl@chem.ku.dk
bHaldor Topsøe A/S, Nymøllevej 55, 2800 Kgs. Lyngby, Denmark
cDepartment of Chemistry and Biochemistry, University of Bern, Freiestrasse 3,
3012 Bern, Switzerland
In situ spectroelectrochemical measurements have also
demonstrated the ability of Au to produce DMC15 from anodic
carbonylation of methanol. The similarity of IR adsorption of
†These authors contributed equally to this work.
6200 | Green Chem., 2019, 21, 6200–6209
This journal is © The Royal Society of Chemistry 2019

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
View Article Online
Green Chemistry
Paper
DMC and DMO however limits the deconvolution of product adsorbed on the catalyst surface were modeled by placing
selectivity verses applied potential.16 By performing electrolysis them on the surface models. The most stable adsorption site
experiments and theoretical calculations, we present a funda- for each adsorbate was determined by performing calculations
mental understanding of the potential-controlled selectivity.
We have recently reported on the formation of DMC via an preferred adsorption does not change from metal to metal.
electrogenerated copper carbonyl species17 as well as palla-
All simulations are performed without representing the
on copper. It was assumed that, for a specific adsorbate, the
dium though a combination of quantitative analysis and electrolyte. Including the electrolyte normally gives rise to an
in situ spectroelectrochemistry.16 Additional electrochemical approximately constant shift of all the binding energies and
methods of producing DMC using transition metal complex would therefore not influence the conclusions31
mediators, metal–organic hybrids, CuCl2 in gas phase and pal-
The Brillouin zone of systems consisting of surfaces was
ladium based electrodes have also been reported by various sampled with 4, 4, 1 k-points in the x, y and z directions
authors.18–25 Although these studies report successful synth- respectively while for the molecules 1 k-point was used in each
eses of DMC they use complex catalyst materials.
In order to facilitate studying electrocatalytic DMC pro- the largest interatomic force was smaller than or equal to
duction computationally, we focus on transition metal sur- 0.05 eV Å−1 while molecules were relaxed up to 0.01 eV Å−1
direction. All systems consisting of surfaces were relaxed until
.
faces, thus favoring the approach from Funakawa et al.13,14 To During optimization of the systems consisting of surfaces the
avoid trial-and-error experiments and gain direction in catalyst bottom two layers of the surface slab were fixed in bulk posi-
development, a fundamental understanding of the process is tion using the calculated lattice constant. The DFT electronic
required. At present still largely unutilised, the combination of energies were corrected for zero-point energy and thermo-
computer simulations with experiments holds promise in the dynamic contributions. Zero point energies and entropic con-
search for understanding in electrocatalysis for chemical pro- tributions at room temperature were obtained by calculating
duction. Rational design is necessary in the development of vibration frequencies and employing the harmonic approxi-
complex catalytic systems.26,27
mation for adsorbates and the ideal gas approximation for gas
Herein we show how density functional theory (DFT) and phase molecules as implemented in ASE/GPAW.32,33 The free
experimental synthesis leads to the understanding of funda- energies of adsorbed species (X*), molecules (X) and surfaces
mental design criteria and requirements for the electrocataly- (*) were calculated as:
sis of DMC. On this basis we establish a theoretical framework
form which suitable catalysts can be selected. The framework
gives direction towards activity, in terms of the potential
required, and selectivity, in terms of co-product formation or
surface poisoning.
GX* ¼ EX* þ ZPEX* þ ΔUX0*;T ꢀ TSTX*
pX
p°
GX ¼ EX þ ZPEX þ ΔHX0;T ꢀ TSXT þ kT ln
G* ¼ E*
ð2Þ
where E is the electronic energy calculated with DFT, ZPE is
the zero point energy, ΔU0,T and ΔH0,T are the changes in
internal energy and enthalpy from 0 K to T, ST is the entropy at
temperature T, k is the Boltzmann constant, p is pressure and
p° is the standard pressure.
The adsorption free energies of carbon monoxide (CO),
methoxy (CH3O) and methyl formate (CH3OCO) were calcu-
lated using methanol, carbon monoxide and hydrogen as
references:
2 Materials and methods
2.1 Density functional theory
The GPAW DFT code was employed in finite difference mode
with a grid spacing of 0.18 Å in order to perform total energy
and vibration frequency calculations.28,29 To describe exchange
and correlation effects the BEEF-vdW and RPBE functionals
were used in two separate sets of calculations.‡ Catalyst sur-
faces were modeled as periodic (3 × 3) fcc 111 slabs of four
atomic layers surrounded by 14 Å of vacuum in the z direction.
The 111 facet is a representative surface as the variations in
binding energies between low index facets is small compared
to the differences between the different metals in the present
article. The trend between different metals which is in focus
would be the same regardless of which low index facet the cal-
culations were performed on. As previously demonstrated for
Cu by Bagger et al.30
ΔGCO ¼ GCO* ꢀ GCO ꢀ G*
ΔGCH O ¼ GCH O* þ μðHþþeꢀÞ ꢀ GCH OH ꢀ G*
ð3Þ
3
3
3
ΔGCH OCO ¼ GCH OCO* þ μðHþþeꢀÞ ꢀ GCH OH ꢀ GCO ꢀ G*
3
3
3
In Fig. 3b the adsorption free energies are shown relative to
the adsorbate on copper which is calculated as follows:
Cu
ΔGrXel:Cu ¼ GMX* ꢀ G þ GC* u ꢀ G*M
ð4Þ
X*
Molecules were modeled by placing them in a cell and sur-
rounding with 8 Å of vacuum on each side. A single molecule
was used per super cell in the simulations. Adsorbates
Cu
where GMX* is the free energy of adsorbate X on metal M, G is
X*
the free energy of adsorbate X on copper, GC* u is the energy of
the copper surface and GM* is the energy of the surface of M. To
treat the relation between the applied potential and the chemi-
cal potential of a proton–electron pair, the computational
‡Unless otherwise noted, BEEF-vdW values are shown throughout the text.
This journal is © The Royal Society of Chemistry 2019
Green Chem., 2019, 21, 6200–6209 | 6201

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
View Article Online
Paper
Green Chemistry
hydrogen electrode was employed, changing the proton–elec- (HS-GC-MS) was used to detect DMC, dimethyl oxalate (DMO),
tron chemical potential by −eU:34,35
dimethoxy methane (DMM), molecular methyl formate (mMF)
and dimethyl ether (DME) in the electrolyte solution. Product
identification was achieved by mass spectra and retention time
matches of reference solutions. Quantification was performed
by external calibration for DMC, DMO, mMF and DMM.
Faradaic efficiency (FE) is calculated by the equation below for
DMC and DMO:
1
2
μ
ꢀÞðUÞ ¼ GH ꢀ eU
ð5Þ
ðHþþe
2
thus being able to account for the effect of the applied poten-
tial on the energy of potential dependent steps.
When working with the BEEF-vdW exchange–correlation
functional the total energy calculation yields an ensemble of
2000 energies. The thermodynamic corrections and adsorption
energy calculations described from eqn (2)–(4) are applied
element-wise on the ensemble.36 This results in 2000 adsorp-
tion free energies for each adsorbate on each metal. The mean
value of the ensemble corresponds to the BEEF-vdW calculated
value while the standard deviation of the ensemble represents
the uncertainty of the calculation.
The uncertainty of the calculation holds information on
how much the calculated adsorption energy§ depends on the-
choice of functional at the generalized gradient approximation
level of accuracy. All structures with total energies are avail-
able on Jan Rossmeisl’ group homepage: “https://nano.ku.dk/
english/research/theoretical-electrocatalysis/katladb/dmc_
electrosynthesis/”.
F
Q
FEi ¼ 2ni ꢁ
ð6Þ
where i is DMC or DMO, n is quantity in mol, F is Faraday’s
constant in C mol−1 and Q is the charge in C.
The potential used for synthesis (vs. SCE) should not be
compared directly to the values shown in the DFT work (vs.
RHE). Different reference electrode potentials have been used
for practical purposes and conversion between the two presents a
challenge when working with a non-aqueous electrolyte solution.
3 Results and discussion
3.1 Mechanisms and surface species
As the first step in our theoretical analysis of the synthesis of
DMC, two simple reaction mechanism are suggested. One
leads to the formation of DMC:
2.2 Experimental
The electrochemical tests were conducted in a custom made
glass H-type cell based on the design of Funakawa et al.13 The
working and counter compartments were separated with a
glass frit. A saturated calomel reference electrode (SCE) poten-
tial was placed in a separate compartment and connected to
the working electrode compartment and close to the working
electrode by the use of a Luggin capillary. The working elec-
trode consisted of a gold wire (Premion® 99.9985% Alfa Aesar)
and was stabilised with a holder made of PFA tubing. The geo-
metric surface area of the gold wire was 0.186 cm2. A Nordic
EC ECi-200 potentiostat/galvanostat was used to apply poten-
tial and record current.
All measurements were performed at room temperature. For
the electrochemical tests the counter and working electrode
compartments each contained 20 mL of electrolyte solution of
0.1 M NaClO4·H2O dissolved in methanol (ultrapure, spectro-
photometric grade 99.8% Alfa Aesar). The electrolyte solution in
the counter compartment was purged with argon gas continu-
ously. The electrolyte was chosen based on previous studies.14,25
The electrolyte solution in the working compartment was
initially purged with argon gas for 5–10 minutes to create an
inert atmosphere. Carbon monoxide at atmospheric pressure
was subsequently bubbled into the electrolyte to saturate the
solution prior to the electrochemical synthesis tests.
CO þ * Ð CO*
CH3OH þ * Ð CH3O * þ ðHþ þ eꢀÞ
ð7Þ
methoxy
2CH3O* þ CO* Ð CH3 OCOOCH3 þ 3*
DMC
and the other leads to the co-product, DMO:
CO þ * Ð CO*
CH3OH þ CO* Ð CH3OCO* þðHþ þ eꢀÞ
ð8Þ
methyl formate
2CH3OCO* Ð CH3 OCOCOO CH3 þ 2*
DMO
The initial step is assumed to be the adsorption of carbon
monoxide which is a chemical step, followed by an electro-
chemical step in which methanol is activated. Methanol can
be activated either by direct adsorption on the surface as
methoxy as shown in eqn (7), or by co-adsorbing with carbon
monoxide to form methyl formate as shown in eqn (8). Both of
these steps are electrochemical as they are accompanied by the
exchange of a proton–electron pair. As a consequence it is
possible to lower the free energy of these electrochemical steps
by applying potential. In the suggested mechanisms DMC
forms when two methoxy species meet one carbon monoxide
species while DMO is formed when two methyl formate
species meet. The difference in the reaction mechanisms
leading to DMC and DMO is further summarized in Fig. 1.
It can be seen from the reaction mechanisms that in order
to describe the energetics of all the steps it is necessary to
work with three variables that differ from catalyst to catalyst,
namely the adsorption free energies of carbon monoxide,
Constant potential electrolysis tests were conducted each
for 20 minutes. Between tests, all components of the cell were
rinsed with methanol. A headspace sampler connecting to a
gas chromatograph with
a mass spectrometer detector
§The uncertainty is estimated only from the 0 K total energy calculation,
without the zero-point energy and thermal contributions.
6202 | Green Chem., 2019, 21, 6200–6209
This journal is © The Royal Society of Chemistry 2019

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
View Article Online
Green Chemistry
Paper
Fig. 1 Summary of the reaction paths towards DMC (above) and DMO
(below). DMC requires methoxy to be present on the surface while DMO
is made from methyl formate species. The yellow, grey, red and cyan
circles represent gold, carbon, oxygen and hydrogen atoms respectively.
methoxy and methyl formate. In addition to the adsorption
free energies it is also necessary to consider the reaction ener-
gies towards the different products. Fig. 2 shows the adsorp-
tion free energies of all the surfaces species plotted against
each other while the calculated reaction energies for DMC,
DMO and other co-products identified experimentally are
shown in Table 1.
From Fig. 2b it appears that there is correlation between the
adsorption energy of methyl formate and carbon monoxide. By
zooming in on the adsorption energy of methyl formate against
the adsorption energy of carbon monoxide and fitting a linear
model to the ensemble values, Fig. 3 is obtained. Weakening/
strengthening the adsorption of carbon monoxide weakens/
strengthens the adsorption of methyl formate in return. This
observation is called
a linear scaling relation between
adsorbates35,37–39 and can be explained by the fact that both
adsorbates bind onto surfaces through a carbon atom. As the
adsorption energy of methyl formate can be expressed as a func-
tion of the adsorption energy of carbon monoxide it is possible
to reduce the number of variables required to describe the syn-
thesis of DMC from three to two, namely the adsorption free
energies of carbon monoxide and methoxy.
By comparing Fig. 2a and b it can be seen that when using
the adsorbate on a specific metal as the reference instead of
the gas phase reactants, the uncertainty of the calculations is
smaller and the ensemble values shown by the ellipses follow
the line more closely, resulting in a R2 value closer to 1. In
other words, it is possible to more accurately describe how
much the adsorption free energies varies from one surface to
another than the “absolute” values of the adsorption free ener-
gies relative to the gas phase reference.
Fig. 2 Adsorption energies of all the surface species plotted relative
to each other, (a) methoxy vs. CO (b) methyl formate vs. CO and
(c) methoxy vs. methyl formate. The ellipses show two standard devi-
ations of the BEEF ensembles for each of the catalysts considered herein
while the points show the ensemble means.
3.2 Thermodynamic analysis
Using the adsorption free energies of carbon monoxide and
methoxy as descriptors, a thermodynamic analysis can be
performed as shown in Fig. 4. In the thermodynamic analysis, the different surfaces will be related with much smaller error
catalysts that satisfy set of design criteria for DMC bars due to error-cancellation.
a
production are identified. These design criteria are energy
3.2.1 Energy efficiency. Energy efficiency assumes that the
efficiency, no surface poisoning and selectivity towards DMC.
potential required to perform the synthesis is low. The step
The observed activity trends as shown in Fig. 4 are clear that determines the energy efficiency towards producing DMC
even including the BEEF error bars. The differences between is the formation of methoxy as shown in eqn (7). In Fig. 4 the
This journal is © The Royal Society of Chemistry 2019
Green Chem., 2019, 21, 6200–6209 | 6203