See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/233901054

Catalysis of the electrochemical reduction of carbon dioxide

Article in Chemical Society Reviews · March 2013

DOI: 10.1039/c2cs35360a · Source: PubMed

CITATIONS READS

1,418 5,220

3 authors, including:

Cyrille Costentin Marc Robert

Université Grenoble Alpes Université de Paris
212 PUBLICATIONS 12,607 CITATIONS 252 PUBLICATIONS 17,993 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Marc Robert on 21 May 2014.

The user has requested enhancement of the downloaded file.

 Chem Soc Rev
View Article Online
REVIEW ARTICLE View Journal | View Issue

Catalysis of the electrochemical reduction of carbon

Cite this: Chem. Soc. Rev., 2013,
dioxide†
42, 2423
Cyrille Costentin, Marc Robert and Jean-Michel Savéant*
Published on 11 December 2012 on http://pubs.rsc.org | doi:10.1039/C2CS35360A

The direct and catalyzed electrochemistry of CO2 partake in the contemporary attempts to reduce this
inert molecule to fuels by means of solar energy, either directly, after conversion of light to electricity,
or indirectly in that all elements of comprehension derived from electrochemical experiments can be
Downloaded by SCD Université Paris 7 on 25 February 2013

used in the design and interpretation of photochemical experiments. Following reviews of the activity
in the field until 2007–2008, the present review reports more recent findings even if their interpretation
remains uncertain. It also develops useful notions that allow analyzing and comparing more rigorously
Received 30th August 2012 the performances of existing catalysts when the necessary data are available. Among the general trends
DOI: 10.1039/c2cs35360a that transpire presently and are likely to be the object of active future work emphasis is put on the
favorable role of acid addition in homogeneous catalytic systems and on the crucial chemical role of
www.rsc.org/csr the electrode material in heterogeneous catalysis.

1 Introduction negative as 1.97 V vs. SHE in an aprotic solvent like N,N 0 -

dimethylformamide (DMF),6 and 1.90 in water.7 Catalytic
Among the contemporary energy objectives, the reduction of strategies have thus been developed to avoid the intermediacy
carbon dioxide is a major issue. Ideally, one dreams to convert of the CO2 anion radical for obtaining the various products one
CO2 into fuels by means of solar energy. One way is to directly may expect from the electrochemical reduction of CO2 at lesser
use light to achieve this conversion. The other involves the energetic costs.8 Most of these products appear in the diagram
preliminary conversion of light into electricity and then the represented in Fig. 1. Not represented on the diagram are
electrochemical reduction of CO2. The first of these approaches multi-carbon products starting with oxalate and other even
has been the object of recent reviews1–3 with particularly note- more important condensation products. Fig. 1 is actually a
worthy examples.4,5 We will therefore concentrate on the electro- set of Pourbaix diagrams relating the standard potential of each
chemical reduction of CO2, having however in mind that the data, CO2 conversion reaction to the pH in water as the solvent,
models and all elements of mechanistic comprehension derived pointing to the fact that in most cases not only electrons but
from electrochemical experiments can be used in the design and also protons are involved. This observation should be borne
interpretation of photochemical experiments. in mind when discussing these energetically more favorable
Before addressing the strategies developed for catalyzing the strategies, in water as well as in non-aqueous solvents. These
electrochemical reduction of CO2, it is useful to remind what strategies essentially consist of catalyzing the target reaction
products are obtained by direct reduction on an electrode by means of a homogeneous or a heterogeneous catalyst.
material that does not interfere in the reaction and the factors Homogeneous catalysis has essentially, but not exclusively,
that govern the competition between their formations. This will involved reduced states of transition metal complexes. Following
be the object of the second section of this review. the pioneering work on Ni and Co macrocyclic complexes as
Direct reduction on such an ‘‘outersphere electrode’’ entails potential catalysts for electrochemical CO2 reduction,10 many
the initial formation of the highly energetic CO2 anion radical. metal complexes have been used for this purpose. Recent
The standard potential of the CO2 =CO 2 couple is indeed as reviews11,12 have listed them, sometimes in great detail, until
2007–2008, with however rather cursory descriptions of their
performances. In particular, comparison between the various
Université Paris Diderot, Sorbonne Paris Cité, Laboratoire d’Electrochimie
catalysts could not be performed on sound bases. These are the
Moléculaire, Unité Mixte de Recherche Université – CNRS No 7591,
Bâtiment Lavoisier, 15 rue Jean de Baı̈f, 75205 Paris Cedex 13, France.
reasons that the present review pursues the following goals: (i)
E-mail: saveant@univ-paris-diderot.fr report more recent findings even if their interpretation remains
† Part of the solar fuels themed issue. uncertain on the one hand and to analyze and (ii) compare

This journal is c The Royal Society of Chemistry 2013 Chem. Soc. Rev., 2013, 42, 2423--2436 2423
 View Article Online

Review Article Chem Soc Rev

Scheme 1 Reduction of CO2 at an inert electrode.

Published on 11 December 2012 on http://pubs.rsc.org | doi:10.1039/C2CS35360A
Downloaded by SCD Université Paris 7 on 25 February 2013

Fig. 1 Variation of the apparent standard potential, E0ap, with pH (Pourbaix

diagrams) for the reductive conversion of CO2 into the following products.
In the order of decreasing values of E0ap at pH = 0: CH4, CH3OH, CH2O, CO
(full line), HCO2H, HCO2 (dashed line), C2O4H2, C2O4H, C2O42.9 The slopes
are RTln10/F with the exception of the horizontal lines and of the formation
of C2O4H (RTln10/2F).

more rigorously the performances of the catalysts when the

necessary data are available. The latter objective requires recalling
or describing of some notions that are deemed useful to this
purpose. This is the object of Section 3 in which emphasis will
be put on the relationships that link turnover frequency and
overpotential for preparative scale experiments. It will also
describe the way in which cyclic voltammetry can provide an
access to these relationships in the framework of a quick Fig. 2 Oxalate yield in the preparative electrolysis of CO2 in DMF at a current
catalyst screening. Section 4 reviews the homogeneous catalysts density of 1.6 mA cm2 at 0 1C as a function of CO2 concentration.16
that convert CO2 essentially to CO (which can be further
converted to fuels, like, e.g., methanol or hydrocarbons).
The favorable role of acid addition is emphasized and the
section closes with an attempt to benchmark the various in Scheme 1.13–16 They are the results of the homogeneous
catalysts described in the literature. Section 5 is devoted to chemistry of the CO2 anion radical. Dimerization, leading to
the homogeneous catalysts that give rise to products other than oxalate, is a manifestation of the radical character of CO 2
CO, i.e., formate and oxalate. Attempts to produce CO + H2 entailing a one-electron stoichiometry.
mixtures in prescribed proportions are discussed. The last CO and carbonate, on the one hand, and formate, on the
section is dedicated to heterogeneous catalysts putting empha- other, result from CO 2 being a Lewis base and a Brønsted
sis, thanks to a few examples, on the role of the metal of the base, respectively, leading to a two-electron stoichiometry. The
electrode. dimerization of the anion radical is fast in DMF6 and in water
(from pulse radiolysis experiments17), close to the diffusion
limit. This mechanism was substantiated by a careful analysis
2 Direct electrochemical reduction at of the product ratio oxalate/CO as a function of the concen-
tration of CO2. The oxalate/CO product ratio (in the absence
inert electrodes
of purposely added water), is an increasing function of the
Mercury and lead may be considered as the best approximation krr I
following parameter: prr rs ¼ (I is the cur-
of electrode materials that do not interfere in the chemistry of ðkrs Þ3=2 ½CO2 3=2 D1=2 F
CO2 reduction. Three products are formed upon electrolysis rent density, D the CO2 diffusion coefficient. krr and krs are
in DMF at the level of the very negative cyclic voltammetric defined in Scheme 1). The good agreement between the theo-
CO2 reduction wave,6 namely carbon monoxide, oxalate retical predictions and experiments provides (Fig. 2) a rigorous
and formate, with a quantitative global faradaic yield. The proof of the mechanism, a rare event in direct and catalyzed
reaction pathways that lead to these three products are shown electrochemical reduction of CO2.

2424 Chem. Soc. Rev., 2013, 42, 2423--2436 This journal is c The Royal Society of Chemistry 2013
 View Article Online

Chem Soc Rev Review Article

3 Useful notions
We recall in this section notions concerning catalysis in general
that may be useful in the mechanism analysis and practical
design of catalytic systems for the electrochemical reduction of
CO2. The distinction between redox catalysis and chemical
catalysis is not a recently uncovered notion but seems worth
recalling each time a booming, and necessarily somewhat
anarchic, attention is paid to a new catalysis field as is presently
the reduction of CO2. This is the object of the following first
subsection. Re-examining the notion of turnover number, and
turnover frequency, and their possible dependence on over-
potential is also a timely task for the same reasons. Clarifica-
Published on 11 December 2012 on http://pubs.rsc.org | doi:10.1039/C2CS35360A

tion of these questions in preparative-scale conditions for both

heterogeneous and homogeneous catalysis is the object of the
Fig. 3 Redox and chemical catalysis of electrochemical reactions. 1: redox
following second subsection. Fast screening of the catalytic
catalysis, 2: one-step chemical catalysis, 3: two-step chemical catalysis.
properties of molecules by means of a non-destructive technique
Downloaded by SCD Université Paris 7 on 25 February 2013

such as cyclic voltammetry, serving as guidelines for preparative-

scale electrolyses, is also a timely question, which has recently
made significant progresses. This is the object of following third esters and nitriles, which give rise to a quasi-redox catalysis
subsection. The notions developed in these two last subsections, leading exclusively to oxalate.21
provide the bases of meaningful comparison between the various
catalytic molecules that have been proposed for the electrochemical 3.2 Overpotential and turnover frequency in homogeneous
reduction of CO2 as discussed in the following sections. and heterogeneous catalysis22
The reaction sequence depicted in Scheme 2, complemented
3.1 Redox and chemical catalysis with the condition k 0 c k, represents an almost general two-
In redox catalysis,18 the catalyst simply shuttles electrons between electron stoichiometry reaction scheme.23 The catalytic current
the electrode and the substrate, acting as an outersphere electron in preparative-scale electrolysis then results from the combi-
transfer reagent. The very reason that catalysis is obtained is nation of electrode electron transfers, chemical reactions and
more physical than chemical: the electrons to be transferred steady-state linear diffusive transport sketched in Fig. 4, for
are dispersed in a three-dimensional space instead of being both heterogeneous catalysis (where a film of catalytic mole-
confined within a two-dimensional space. Electrochemists are cules has been deposited on the electrode surface) and homo-
keen designing porous electrodes, so porous that they are geneous catalysis (where the catalytic molecules are dispersed
sometimes named ‘‘volumic electrodes’’. Redox catalysis is in the whole cell compartment).24,25 It follows that:
the equivalent of such an electrode where ‘‘volumization’’  
dCCb SDA dCC IS
reaches the molecular scale. ¼ ¼
dt V dx x¼d 2FV
In contrast with redox catalysis, where the catalyst acts as an
outersphere electron transfer agent, chemical catalysis involves
more intimate interactions between the active form of the In practically interesting systems, the catalytic reaction is
catalyst and the substrate. Two situations may arise according I 0
fast, in which case: ¼ kCA GQ in the heterogeneous case and
to the strength of these interactions. In the first of these, 2F
I q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
bonded interactions in the transition state confer to electron ¼ 2kCA 0 D ðC Þ
P Q x¼0 in the homogeneous case.
transfer an innersphere character, which results in a lower F
activation energy than for an outersphere electron transfer of Using the classical definition of the turnover number, TON, as
same driving force (Fig. 3). Stronger interactions may involve the ratio between the number of molecules transformed by the
the formation of an adduct (Fig. 3), which should not however catalytic reaction and the number of catalyst molecules employed to
be too stable for an efficient catalysis to operate. achieve this transformation, one obtains in the heterogeneous case:
Among others, the reductive conversion of vicinal dihalides mol C Cb  V 0 GQ
into the corresponding olefins has provided a particularly clear TON ¼ ¼ C0 ¼ kCA t ð1Þ
molðP þ QÞ GP  S G0P
illustration of redox and chemical catalysis.19,20 In the case of
CO2 reduction, unambiguous examples of redox catalysis could and thus, the turnover frequency:
not be established because of deactivation of the catalyst in the GQ
0
course of the catalytic process. Indeed, aromatic molecules, the TOF ¼ kCA ð2Þ
G0P
anion radicals of which usually serve as redox catalysts, undergo
carboxylation, preventing them to act as electron mediators. In the homogeneous case, the above expression of the current
A noteworthy exception to this rule is provided by aromatic density is valid for fast catalytic processes, such as the ones of

This journal is c The Royal Society of Chemistry 2013 Chem. Soc. Rev., 2013, 42, 2423--2436 2425
 View Article Online

Review Article Chem Soc Rev

Published on 11 December 2012 on http://pubs.rsc.org | doi:10.1039/C2CS35360A
Downloaded by SCD Université Paris 7 on 25 February 2013

Scheme 2 General two-electron stoichiometry catalytic reaction scheme (P/Q: catalyst couple, A substrate, C product, B: intermediate).

interest in the present discussion. Then ‘‘pure kinetic conditions’’ Counting only the catalyst molecules present in the reaction
are achieved corresponding to the establishment, close to the layer is the basis of a fair and meaningful comparison between
electrode surface, of a steady-state, which results from the mutual the performances of heterogeneous and homogeneous catalysts.
compensation of catalytic reaction and catalyst diffusion: This approach is at variance with current practice, which takes
instead the catalyst present in the whole solution into account in
d2 CQ 0
DP  2kCA CQ ¼ 0; the determination of the TON and TOF (see Section 4). This not
dx2
  only unduly disadvantages homogeneous catalysis toward hetero-
dCQ geneous catalysis, but also does not reflect the actual properties of
with ðCQ Þx¼1 ¼ 0; ¼0
dx x¼1 the catalyst. The volume-to-surface ratio of the cell would then
interfere, making the comparison between the various reports
Within this reaction–diffusion layer, the Q-profile, obtained by
meaningless. We note that the catalysts present in the solution
space integration of the above differential equation and boundary
outside the reaction layer may serve as a useful stock when
conditions is:
0 sffiffiffiffiffiffiffiffiffiffiffiffi 1 deactivation of the catalyst in the course of electrolysis occurs, a
2kCA 0 possibility that is not available in heterogeneous catalysis.
CQ ðxÞ ¼ ðCQ Þx¼0 exp@ xA In a number of cases, electron transfer between the electrode
DP
and the catalyst is so fast that the Nernst law is obeyed. Then, in
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi both the heterogeneous and homogeneous cases:
0
corresponding to an approximate thickness m ¼ DP =2kCA .
0
Keeping the same definition of TON as in the case of hetero- kCA
TOF ¼ 0
ð3Þ
geneous catalysis, we consider the total amount of the catalyst, 1 þ exp½f ðE  Ecat Þ
including both forms, per unit surface area in the reaction where f = F/RT and E0cat is the catalyst standard potential, i.e. the
layer: standard potential of the P/Q couple.
sffiffiffiffiffiffiffiffiffiffiffiffi
DP 0 In a more general case the kinetics of electron transfer to
molðP þ QÞm ¼ S 0 P
C the catalyst is not so fast and may therefore interfere. Then,
2kCA
assuming the validity of the Butler–Volmer rate law with a 0.5
leading to exactly the expressions of TON and TOF similar but transfer coefficient:27
not identical to their heterogeneous counterparts (eqn (1) and F 1
(2), respectively): ¼ 0 G0 exp½f ðE  E 0 Þ
I 2kCA P cat
mol C 0 ðCQ Þx¼0
TON ¼ ¼ kCA t ð1 0 Þ 1 1
molðP þ QÞm CP0 þ þ 0 0
khet;cat
S G0P exp½f ðE  Ecat
0
Þ=2 2kCA GP

0 ðCQ Þx¼0 in the heterogeneous case, where khet,cat is the heterogeneous

TOF ¼ kCA ð2 0 Þ S
CP0 electron transfer standard rate constant (corresponding to a

2426 Chem. Soc. Rev., 2013, 42, 2423--2436 This journal is c The Royal Society of Chemistry 2013
 View Article Online

Chem Soc Rev Review Article

in the heterogeneous case:

qffiffiffiffiffiffiffiffiffi  
0 fZ
1 expðf ZÞ 2 kCA exp  2
¼ þ het;cat pffiffiffiffiffiffiffiffiffiffiffiffiffi
TOF TOF0 kS TOF0
ð4Þ
0 0
exp½f ðEAC  Ecat Þ
þ
TOF0
and in the homogeneous case:
 
pffiffiffiffiffiffiffiffiffi exp f Z
1 expðf ZÞ 2DP 2
¼ þ cat pffiffiffiffiffiffiffiffiffiffiffiffiffi
TOF TOF0 kS TOF0
0 0
exp½f ðEAC  Ecat Þ
þ ð5Þ
Published on 11 December 2012 on http://pubs.rsc.org | doi:10.1039/C2CS35360A

TOF0
with:
0 0 0
TOF0 ¼ kCA exp½f ðEAC  Ecat Þ
Downloaded by SCD Université Paris 7 on 25 February 2013

in both cases.
It thus appears that there is a definite relationship between
turnover frequency and overpotential for each particular catalyst.
This recently emerged concept allows a precise comparison
between catalysts in contrast with the previous notion that each
catalyst was characterized by two independent parameters, its
TOF and its overpotential, leading to the rather trivial aphorism
that a good catalyst is characterized by a large TOF and a small Z
and vice versa. An example of the log TOF–Z curve is shown in
Fig. 4 Catalysis of electrochemical reactions. Electrode electron transfer,
Fig. 5, where three linear variation zones can be delineated:
chemical reactions and steady state linear diffusive transport in heterogeneous independence from Z at large values of Z, linear variation with
and homogeneous molecular catalysis of an electrochemical reaction according an F/RTln10 slope at small values of Z, and, in between, an
to Scheme 2 (with k 0 c k), in the case where the substrate bulk concentration, F/2RTln10 slope linear variation corresponding to the kinetic
C0A, is maintained constant during the whole course of electrolysis.26 x: distance
contribution of electron transfer to the catalyst. The latter segment
to the electrode surface, m: film (top) or reaction layer (bottom) thickness,
d: diffusion layer thickness. Dsubscript: diffusion coefficients. Gsubscript: surface
decreases in size and eventually vanishes as electron transfer
concentrations. Csubscript: volume concentrations. G0P, C0P: total surface and becomes faster and faster.
volume concentrations of the catalyst. S: electrode surface area, V: cell It thus appears that one of the best ways to characterize the
compartment volume. I: current density, k: catalytic second order rate intrinsic catalytic properties of a molecule is the value of its
constant (see Scheme 2).
TOF0, the turnover frequency at overpotential zero.
The low overpotential linear variation zone in eqn (4) may look
at first sight as a form of a Tafel plot. A ‘‘Tafel plot’’ is a classical
reactant attached to the electrode surface, dimension of time1),
and:

F 1
¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
I 0 0 0 Þ
2kCA DP CP exp½f ðE  Ecat

1 1
þ þ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
kcat
S CP
0 0
exp½f ðE  Ecat Þ=2 2kC 0 D C 0
A P P

in the homogeneous case, where kcat S is the electron transfer

standard rate constant corresponding to a reactant in solution
(dimension of space  time1)
If the overpotential, Z, defined as Z = E0AC  E (where E is the
electrode potential at which electrolysis is run and E0AC
the standard potential corresponding to the transformation of
Fig. 5 An example of turnover frequency vs. overpotential relationship.
the substrate A into the product C), is introduced in the pffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffi
preceding equations:
0
EAC 0
 Ecat = 0.64 V, log TOF0 ðs1 Þ = 4.6, 2DP =kS ¼ 2 kCA 0
=khet;cat
S ¼ 0:03 s1=2 .

This journal is c The Royal Society of Chemistry 2013 Chem. Soc. Rev., 2013, 42, 2423--2436 2427
 View Article Online

Review Article Chem Soc Rev

linear approximation of the relationship between the current, to be used to estimate the overpotential and the catalytic efficiency,29
or current density, and the overpotential of an electrochemical successful strategies, as detailed below, for derivation of the TOF–Z
reaction, kinetically controlled by electron transfer. It is char- relationship from cyclic voltammetric data are recent.22
acterized by a transfer coefficient, a, giving rise to the slope In the homogeneous case, Fig. 6 shows a zone diagram
aF/RTln10 and an exchange current density I0 defined as depicting the various diffusion/reaction regimes that can be
the current density at zero overpotential. In the present hetero- observed by means of cyclic voltammetry according to the various
geneous case: intervening parameters (these have been defined in the caption of
0 0 0 0 Fig. 4, n is the scan rate).30,31 The catalytic systems of interest are
a ¼ 1 and I0 =2F ¼ kCA GP exp½f ðEAC  Ecat Þ
those in which k is large enough for the representative point of
¼ TOF0  G0P the system to stand in one of the three upper zones in which
‘‘pure kinetic conditions’’ are achieved. Pure kinetic conditions
are exactly the same as already introduced in preparative-scale
Published on 11 December 2012 on http://pubs.rsc.org | doi:10.1039/C2CS35360A

It is seen that, unlike TOF0, I0 is not a molecular intrinsic conditions, corresponding to the same steady-state situation.
expression of the catalytic capabilities of the catalyst since it The most classical cyclic voltammetric response is the S-shaped
depends on the amount of catalyst one has been able to deposit wave in the upper right-hand zone of Fig. 6, of equation:32,33
on the electrode surface. Moreover, Tafel plots with unity transfer qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
coefficients are quite rare for electrochemical reactions kineti- i 2kCA 0
DP CP0
Downloaded by SCD Université Paris 7 on 25 February 2013

cally controlled by electron transfer. This recalls to us that eqn (5) ¼ 0 Þ

ð6Þ
FS 1 þ exp½f ðE  Ecat
is not an actual Tafel plot. It is rather a turnover frequency vs.
overpotential plot characterizing a deposited molecular catalyst (i: current, S: electrode surface area). It is independent of scan rate
in the low overpotential region. and the reverse trace is identical to the forward trace. This type of
Likewise in the homogeneous case where: response corresponds to a fast catalytic reaction with nevertheless
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi negligible consumption of the substrate in the reaction–diffusion
0 0
I0 =F ¼ 2DP =kCA CP  TOF0 ; layer, and a fast electron transfer between the catalyst and the
electrode. Analysis of experimental results by means of eqn (6) is
eqn (5) and the plots in Fig. 5, 8 and 10 are not actual helped by dividing the catalytic current by the peak current, i0p,
Tafel plots. of the reversible P/Q wave that is observed in the absence of
catalysis (e.g. in the absence of substrate):34,35
3.3 Analysis of catalytic responses in cyclic voltammetry
ip0 pffiffiffiffiffiffipffiffiffiffi
Carrying out a systematic series of electrolyses to evaluate a ¼ 0:446  CP0 DP fv ð7Þ
catalyst, determine the best conditions of its use, avoiding side- FS
phenomena such as ohmic losses, self-inhibition, deactivation
etc. . . is an exhausting task. This is the reason that non-destructive The normalization procedure that thus consists of replacing
techniques, such as cyclic voltammetry,28 are currently used to eqn (6) by eqn (8):36
rapidly gauge what can be the catalytic significance of a rffiffiffiffiffiffiffiffiffi
molecule and establish the best conditions for preparative scale 2kC 0
i 2:24 fv
A

electrolyses. In spite of previous attempts to define the ¼ ð8Þ

0
ip0 1 þ exp½f ðE  Ecat
potential of the cyclic or rotating-disk electrode voltammogram Þ

is a straightforward manner of avoiding the determination of

the catalyst diffusion coefficient and the electrode surface area.
A representation of eqn (8) is given in Fig. 7a.
Eqn (6) has been used extensively to measure the global
catalytic rate constant and thereof to analyze mechanism
through determination of reaction orders in catalyst, CO2
or molecules favoring catalysis such as protons donors (see
Section 4.1).37–42 Caution has sometimes been taken to use this
equation when it is indeed applicable, i.e., when the cyclic
voltammetric response is (or is close to) an S-shaped curve
independent from the direction of the potential scan and from
the scan rate.41 In a large number of cases it has nonetheless
been unduly applied when the catalytic response is peak-shaped
and – coming together – scan-rate dependent. The odd situation
of a scan-rate dependent catalytic rate ‘‘constant’’ thus arises.40
One may consequently wonder what the reliability of reactions
Fig. 6 Diffusion/reaction regimes and shape of the cyclic voltammetric responses orders derived in these conditions is and thereof of the mecha-
in catalytic reactions of the type represented in Scheme 2. nistic conclusions drawn.37–40,43–46 Various side-phenomena

2428 Chem. Soc. Rev., 2013, 42, 2423--2436 This journal is c The Royal Society of Chemistry 2013
 View Article Online

Chem Soc Rev Review Article

Published on 11 December 2012 on http://pubs.rsc.org | doi:10.1039/C2CS35360A
Downloaded by SCD Université Paris 7 on 25 February 2013

Fig. 7 Catalytic cyclic voltammetry responses (left) and foot-of-the-wave analyses

(right) for a typical system, v = 0.1 V s1, DP = 105 cm2 s1, C0P = 1 mM, T = 298 K.
a, a 0 : C0A = 1 M, 2k = 100 s1, no substrate consumption. b, b 0 : substrate
consumption. From blue to yellow: C0A (M) = 1, 0.1, 0.01, 0.005; 2k = 100 s1.

have been evoked to explain the appearance of a peak-shaped

response when an S-shaped response is expected.40,47,48
The most obvious of these side-phenomena is substrate
consumption. Passing indeed from the upper-right zone of
Fig. 6 to the two left-hand zones entails an increasing Fig. 8 Catalysis of the electrochemical reduction of CO2 to CO by
consumption of the substrate, the diffusion of which becomes electrogenerated Fe(0)TPP. Cyclic voltammetry of FeTPP (1 mM) in DMF + 0.1 M
the sole rate-determining step in ‘‘total catalysis’’ conditions. In n-Bu4NBF4, in the presence of 0.23 M CO2 and 0.1 (a,a 0 ), 3 (b,b 0 ) M PhOH at a scan
the intermediate zone, the interference of substrate consump- rate of 0.1 V s1. a, b: cyclic voltammetric responses. Solid line: experiments;
dashed line: simulation of the corresponding hypothetical unperturbed catalytic
tion decreases when one approaches the foot of the catalytic
responses. a 0 , b 0 : foot-of-the-wave analyses. c: Resulting turnover frequency vs.
wave. This observation suggests a strategy consisting of plotting overpotential relationships (thick lines) derived from the foot-of-the-wave
i=ip0 against f1 þ exp½f ðE  Ecat
0
Þg1 . In the case where the catalytic analyses at 0.1, 1, 10, 50 V s1 (green, red, magenta and blue respectively). The
three yellow dots represent the results obtained by means of preparative scale
reaction is not perturbed by side phenomena, a straight line is thus
electrolysis in the same conditions as in a.
obtained over the whole range of f1 þ exp½f ðE  Ecat 0
Þg1 , as
0
pictured in Fig. 7a, a . This is no longer the case if substrate
consumption comes into play (Fig. 7b, b 0 ), but remains true charge is increased, e.g. inhibition by products,48 or deactivation
at the foot of the wave within an interval that decreases when of the catalyst.22 The ‘‘foot-of-the-wave’’ strategy applies to these
this side-phenomenon interferes increasingly. The slope of this cases too and to any other side-phenomena that interfere
linear portion may then be used for determining of k and finally increasingly upon increasing the faradaic charge passed.
of the TOF–Z relationship for any catalyst without taking sub- An illustrative example is given in Fig. 8 where the catalyst
strate consumption into account. This is due to the fact that at for CO2-to-CO reduction is electrogenerated iron(0)tetraphenyl-
the foot of the wave, the faradaic charge passed is small and porphyrin (Fe(0)TPP) in the presence of phenol22 (the favorable
therefore the amount of substrate transformed is also small. role of adding acid will be further discussed in Section 4.1). It is
Raising the scan rate also results in a decrease of the faradaic worth noting that the results of preparative-scale electrolysis
charge passed in a given potential interval, thus extending the are consistent with the predictions made on the basis of the
linear portion of the plot.22 There are other phenomena, which foot-of-the-wave analysis, thus validating this strategy.
may be considered as side-phenomena, vis à vis the catalytic Extension to the case where electron transfer between the
reaction, which interfere in an increasing manner as the faradaic catalyst and the electrode is not unconditionally fast is possible.

This journal is c The Royal Society of Chemistry 2013 Chem. Soc. Rev., 2013, 42, 2423--2436 2429
 View Article Online

Review Article Chem Soc Rev

This may become necessary even with fast electron exchanging of CO2 to CO.49 However, the catalyst deactivates after a few
catalysts when catalysis is so fast that electron transfer with the turnovers. Addition of Lewis50,51 and Brønsted acids47,52 consider-
electrode starts to interfere in the overall kinetics. Then, the ably boosts catalysis. In the latter case, weak acids such as
0
function that should be plotted against f1 þ exp½f ðE  Ecat Þg1 is alcohols, particularly trifluoroethanol, were used to avoid catalysis
no longer i=ip0 but, rather: of proton reduction as observed with the same Fe(0)TPP, when an
acid as strong as Et3NH+ is used.53 It has been shown more
i
0 ip0 recently that addition of moderately strong acids, such as phenols,
FITðE  Ecat Þ ¼ pffiffiffiffiffipffiffiffiffi h i
DP
0 Þ
f ðEEcat also boosts catalysis while still yielding essentially CO as
1  0:446ii0 fv exp 2
p kS product.22 A mechanism, based on reaction orders derived from
rffiffiffiffiffiffiffiffiffi cyclic voltammetric responses close to the canonical S-shape, has
2kCA 0
2:24 fv been proposed (Scheme 3 in ref. 47). More reliable and more
¼ 0 detailed mechanistic pictures are expected from the systematic
1 þ exp½f ðE  Ecat Þ
Published on 11 December 2012 on http://pubs.rsc.org | doi:10.1039/C2CS35360A

application of the foot-of-the-wave analysis of cyclic voltammetric

ks is the standard rate constant of electron transfer between the responses as illustrated in Section 3.3 (Fig. 8) with, precisely,
catalyst and the electrode, as derived from the cyclic voltam- the example of phenol addition.54
metry of the catalyst couple in the absence of catalysis. k is then Similar studies of the effect of weak Brønsted acids on the
obtained from the slope of the initial linear portion, of the FIT catalysis of the conversion of CO2 to CO by means of electro-
Downloaded by SCD Université Paris 7 on 25 February 2013

plot, leading finally to the TOF–Z relationship. An example of reduced [Re(bpy)(CO)3(py)]2 have been performed,55 in aceto-
this more sophisticated application of the foot-of-the-wave- nitrile, following the pioneering studies of the Re(bpy)(CO)3
strategy is given in Section 4.2. catalyst.56 Mechanistic conclusions were similar to those drawn
Application of this strategy has two outcomes: (i) determina- in the case of Fe(0) porphyrins. The same experiments have been
tion of the global catalytic rate constants as a function of the repeated recently57 using the same weak Brønsted acids, with
concentration of the various reactants, and hence the mechanism determination of reaction orders and of H/D kinetic isotope
of catalysis by means of the reaction orders thus determined; (ii) effects. Application of eqn (8) was considered as safe because the
establishment of the TOF–Z relationship as a guide to select the cyclic voltammograms are scan-rate independent even though
most efficient conditions of preparative-scale electrolysis and to they do not plateau. In fact, the two characteristics cannot be
compare the intrinsic properties of the catalysts proposed. separated, but the results are most probably reliable because the
peaks are very shallow, close to a plateau behavior.
4 Homogeneous catalysis of the reduction
4.2. Attaching the acid groups to the catalyst molecule leads
of CO2 to CO to a very efficient catalysis
4.1 Favorable role of acids Based on the above observations of the favorable role of acids,
In the series of palladium–phosphine catalysts, addition of an iron(0) tetraphenylporphyrin with OH substituents in ortho,
acid has been shown to be beneficial since the beginning.37–42 It is ortho 0 positions of the phenyl groups (Fig. 9, the phenyl groups
remarkable that in spite of the choice of a strong acid, HBF4, to are in fact perpendicular to the porphyrin ring) proved to be a
accelerate CO2 conversion, CO remains the main product with no particularly efficient catalyst of the CO2 to CO electroreduc-
or little hydrogen production. In spite of the above-mentioned tion.58 Fig. 9a shows the very large catalytic current obtained
uncertainties on reaction orders, the mechanism is likely to be with this molecule. As in many aforementioned cases, it does not
of the type shown in Scheme 3 of ref. 42. plateau off. This is therefore a typical case where the foot-of-the-
Electrogenerated iron(0) porphyrins, essentially tetraphenyl- wave analysis shows its usefulness. The catalysis rate constant
porphyrin (TPP) has been shown to be a catalyst for the conversion can be derived from the linear portion of the plot (Fig. 9b) as well

Fig. 9 An iron(0)porphyrin catalyst with the acid groups attached to the molecule. a: cyclic voltammetry of FeTDHPP (1 mM) in DMF + 0.1 M n-Bu4NBF4, in the absence
(black) and presence (red) of CO2 (0.23 M) at a scan rate of 0.1 V s1. b: Foot-of-the-wave analysis (see text).

2430 Chem. Soc. Rev., 2013, 42, 2423--2436 This journal is c The Royal Society of Chemistry 2013
 View Article Online

Chem Soc Rev Review Article

5 Homogeneous catalysis of the reduction

of CO2 to other products
5.1 Formate (vs. CO, vs. H2)
Formate is also an interesting product since formic acid may be
used for hydrogen storage or as a fuel.63 When the reduction of
CO2 takes the anion radical pathway, formate results from the
reaction with water and is indeed the only product of electro-
lysis in water on an inert electrode. In homogeneous catalysis,
formate appears as a side-product in many cases but there are
catalysts that give a more specific access to its production. For
Fig. 10 Correlation between turnover frequency and overpotential for the series example, substantial amounts of formate were obtained with
Published on 11 December 2012 on http://pubs.rsc.org | doi:10.1039/C2CS35360A

of CO2-to-CO electroreduction catalysts listed in Table 1. Thick red and blue Rh(diphos)2 (diphos = 1,2-bis(diphenylphosphino)ethane) as
segments: TOF values derived from ‘‘foot-of-the-wave-analysis’’ of the cyclic catalyst at a potential where the intermediacy of the rhodium
voltammetric catalytic responses of FeI/0TDHPP and FeI/0TDMPP, respectively,
hydride appears likely,64 based on previous investigations of
in the presence of 2 M H2O. Dashed lines: plots for Fe0TDHPP (top) and
FeI/0TDMPP (bottom). The star indicates a preparative-scale experiment with the electrochemistry of this complex.65 Substantial amounts
Fe0TDHPP as catalyst. of formate were obtained together with CO and H2 with
Downloaded by SCD Université Paris 7 on 25 February 2013

Ru(bpy)2(CO)2 as catalyst in acetonitrile or methanol in the

presence of a proton donor.66 It is interesting to note that the
as the TOF–Z relationship. This is shown in Fig. 10 where it is yield in formate increases with the pKa of the proton donor
compared to other catalysts of the CO2 to CO electroreduction present. Similarly important yields of formate were also
(see Section 4.3). Because of the high catalytic rate, electron obtained with reduced cis-[Rh(bpy)2X2]+ (X = Cl or the trifluoro-
transfer kinetics between the electrode and the catalyst had to methane-sulfonate anion)67 or with another rhodium complex
be taken into account in the analysis as detailed in Section 3.3. [(Z5-Me5C5)Rh(bpy)Cl]+68 in acetonitrile. Recent use of the
In this case too, the result of preparative scale electrolysis catalyst shown in Scheme 3a, in acetonitrile + 5% water,
(star in Fig. 10) falls in line with the prediction of the foot-of- allowed an even better specificity, with still a non-negligible
the-wave analysis. That the remarkable catalytic properties of production of hydrogen (15%).69
this molecule are related to the presence of the phenolic OH A much better selectivity is obtained when using a tungsten-
groups in the molecule close to the iron(0) catalytic center is containing formate dehydrogenase enzyme adsorbed to the
confirmed by comparison with the porphyrin molecule where electrode surface in water with a high turnover frequency and
all the OH groups have been replaced by methoxy substituents a very low overpotential.70
(FeTDMPP). Rather high TOF values are also obtained but Several indications from the papers cited above point to a
at the price of a much larger overpotential, resulting in a first competition between the addition of CO2 and of H+ on the
TOF0 value one billion times smaller than with FeTDHPP. active reduced form of the catalyst. The former leads to CO and
Comparison with the kinetics of catalysis by Fe(0)TPP in the the latter to a metal hydride. The hydride may then react
presence of phenol (Fig. 8c) shows that the eight phenolic competitively with CO2 and H+, yielding formate and H2,
groups in Fe(0)TDHPP are equivalent to a 150 M concentration respectively. One would thus expect that increasing the
of phenol in solution. medium acidity would make the product distribution pass
from CO to formate and finally to H2. This is indeed what
seems to happen with the catalyst shown in Scheme 3b.71
4.3 Catalysts benchmarking Formate is the ‘‘major product’’ (no figures provided) in the
The intrinsic catalytic properties of most of the proposed presence of benzoic acid whereas hydrogen is the only product
homogeneous catalysts are compared in Fig. 10, based on the in the presence of the stronger toluenesulfonic acid. In this
value of the turnover frequency and overpotential as derived case, the three cornered competition between CO, formate and
from preparative-scale experiments described in the literature H2 formation under the effect of added acids of various
(the list is limited to catalysts with a reasonable stability, strengths clearly requires further investigation. This is also
at least several hours and to electrode materials that do not the case with iron(0) porphyrin, which appears as an excellent
interfere in the catalytic process). The intrinsic catalytic proper- catalyst for H2 evolution in the presence of a strong acid,
ties of the various molecules may be compared according to protonated triethylamine,53 whereas addition of weak (e.g.
the location of the representative point in the log TOF vs. trifluoroethanol), and moderately weak (e.g. phenol) acids
Z plane. The characteristics of each catalyst are summarized accelerates the formation of CO. In other words, the conditions
in Table 1, in which comparison can also be made through the for producing syngaz mixtures with adjustable proportions of
TOF0 values. It is seen that the two most efficient catalysts CO and H2 using a single catalyst are not yet at hand. Another
are FeTDHPP and Re(bpy)(CO)3, with a slight advantage for the approach to the same goal consists of the combination of two
former, which also involves a must more common metal than catalyst systems, as in ref. 72 (Re(bpy-tBu)(CO)3Cl and p-Si
the second. under illumination).

This journal is c The Royal Society of Chemistry 2013 Chem. Soc. Rev., 2013, 42, 2423--2436 2431
 View Article Online

Review Article Chem Soc Rev

Table 1 Catalysis of CO2 reduction into CO. Correlation between turnover frequency and overpotential for the series of catalysts listed

Solvent Catalyst
0
ECO 2 =CO
(V vs. SHE) E0cat (V vs. SHE) Z (V) log TOF (s1) log TOF0 (s1) Ref.
0
DMF + 2M H2O Fe TDHPP 0.41–0.56 2.0–3.9 4.9 58
0.690 1.333
Fe0TDMPP 0.89–0.99 1.0–2.2 14.2
1.69
Re(bpy)(CO)3 0.57 3.0 6.1 56
1.25

DMF + HBF4 {m-(triphos)2[Pd(CH3CN)]2} 0.80 0.37 7.8 59

0.260a 0.76

CH3CN + 5% H2O 1.16 0.51 0.35 8.7 60

Published on 11 December 2012 on http://pubs.rsc.org | doi:10.1039/C2CS35360A

0.650
Downloaded by SCD Université Paris 7 on 25 February 2013

CH3CN 1.30 0.87 1.2 9.8 61

0.650

1.25 0.81 1.2 9.1

1 : 4 H2O CH3CN 1.30 0.55 1.9 7.4 62

0.650

a 0
The large change in ECO2 =CO
is due to the presence of a strong acid, HBF4, much stronger than (CO2 + H2O).

quantitative in acetonitrile on a glassy carbon electrode at a very

low overpotential, prior to the reduction of dioxygen, allowing
the capture and reduction of CO2 from air selectively to a C2
product, oxalate. It should however be noted that no simple
route to reductively transform oxalate or oxalic acid to valuable
carbon derivatives is available so far.

Scheme 3 Catalysts for the reduction of CO2 to formate.

6 Heterogeneous catalysis of the reduction
of CO2
5.2 Oxalate 6.1 Reduction to CO by means of transition metal complexes
Oxalate is formed in high yields together with CO in the direct at mercury electrodes
reduction of CO2 on inert electrodes in aprotic media, by means In spite of the fact that reduced nickel cyclams have been
of dimerization of the initial CO2 anion radical (Section 3). It is shown to catalyze the reduction of CO2 to CO on an inert
formed exclusively in the same medium with anion radicals of electrode such as glassy carbon,62,74 the same complexes give a
aromatic esters and nitriles as catalysts (Section 3.1). In both much better catalysis on mercury making these systems ones of
cases, a rather large overpotential is required to make the the most selective and efficient catalysts for the CO2 to CO
reaction go. A remarkable catalytic system in this connection conversion.75–78 Careful studies have shown that the most likely
is shown in Scheme 4.73 The formation of oxalate is practically catalyst is the nickel(I) complex chemisorbed on mercury.74,79,80

2432 Chem. Soc. Rev., 2013, 42, 2423--2436 This journal is c The Royal Society of Chemistry 2013
 View Article Online

Chem Soc Rev Review Article

Published on 11 December 2012 on http://pubs.rsc.org | doi:10.1039/C2CS35360A
Downloaded by SCD Université Paris 7 on 25 February 2013

Scheme 4 A remarkable catalyst for the reduction of CO2 to oxalate.

The chemical nature of this adduct and the reasons that make 0.07–0.1 eV84,85 (the 0.65 eV stabilization mentioned in ref. 81

it such a good catalyst are however not known at present. Its surface is for O2
2 not for O2 ). This a first clue that the results obtained
concentration is likewise unknown making it impossible for in ref. 81 are not simply due to ion pairing of CO 2 by the
the moment a specific characterization of its intrinsic catalytic 1-ethyl-3-methylimidazolium cation but that the electrode mate-
properties of the type described in Section 3.2 for homogeneous rial, platinum or silver, is likely to be involved. The fact that
catalysts. the behavior of an inert electrode such as glassy carbon under the
same conditions is not mentioned seems to indicate that the
6.2 Reduction in the presence of pyridinium and other
same spectacular results have not been achieved in this case. We
organic cations on non-innocent electrodes
may thus conclude that the mechanism of the reaction should be
Two recently proposed catalytic systems give interesting and reconsidered in depth having in mind the likely chemical role of
puzzling results. the noble metals taken as electrode material. From a practical
The simpler of the two consists of merely adding to the aqueous viewpoint, the fact that noble metals are required as electrode
solution a large concentration of a salt generally used as an ionic materials tempers the interest of such systems.
liquid for electrochemical purposes, namely 1-ethyl-3-methylimid- Equally interesting and puzzling is the reductive conversion
azolium tetrafluoroborate in very high concentration (18 mol%). of CO2 to formic acid methanol thanks to the addition of proto-
The faradaic yield in CO is practically quantitative, the rate nated pyridine to the solution.86–90 A first report mentioned a 30%
is quite respectable at a rather low overpotential.81 These spec- methanol faradaic yield at a palladium electrode in water at a pH of
tacular results were interpreted as resulting from a strong 5.4 during galvanostatic electrolyses where the electrode potential
interaction between the CO2 anion radical and the 1-ethyl-3- remains close to 0.5 V vs. SHE in the course of electrolysis.86
methylimidazolium cation allowing the formation of the corre- Formic acid and methanol faradaic yields of 11 and 22% respec-
sponding ion pair at a very positive potential as compared to tively were further reported at platinum or palladium electrode in
usual values of the standard potential of the CO2 =CO 2 couple similar solution and electrolysis conditions.88 The rest of the charge
(1.97 V vs. SHE in an aprotic solvent such as DMF6 and 1.90 passed apparently corresponds to H2 generation. The replacement
in water7). The influence of ion-pairing on the thermodynamics of pyridine by 4-tert-butylpyridine resulted in the absence of formic
and kinetics of electrochemical reactions in general is a reason- acid and of a decreased amount of methanol in the reduction
ably well-documented issue (see ref. 82 and 83 and references products.88 Even better methanol yields, close to 100% are reported
therein). The peak potentials reported in this study (Fig. S5 and on a p-GaP electrode under illumination.87 Much lower yields of
S6 in the supporting information of ref. 81), indicate that the formic acid (3%) with no methanol formation were recently
gain in driving force resulting from ion pairing of CO 2 by the reported at an illuminated iron pyrite electrode.90 It is noteworthy
1-ethyl-3-methyl imidazolium cation would be of the order of that no such methanol and/or formic acid occurs at a carbon
1.2 eV, implying that electron transfer and ion pairing should be electrode, indicating that some interaction with the electrode
concerted. While such a situation is theoretically conceivable, material plays an important role.88 In spite of this observation
such a considerable strength of the ion pair is quite unlikely. In detailed mechanism analyses based on cyclic voltammetric
the analogous case of the ion pair formed by the superoxide responses were carried out as if protonated pyridine was interfering
anion radical, taken as a motivation example by the study only through homogeneous reactions.88,89 Unfortunately most of
in ref. 81, the free energy of interaction is not more than these analyses were carried out at very low scan rates (5 mV s1)

This journal is c The Royal Society of Chemistry 2013 Chem. Soc. Rev., 2013, 42, 2423--2436 2433
 View Article Online

Review Article Chem Soc Rev

where the linear diffusion conditions required by the simula- molecules that are contained in the reaction layer adjacent
tions may not be achieved because of the interference of to the electrode and therefore effectively participate in the
natural convection.45 It is remarkable in this respect that when catalytic process. These notions apply to preparative-scale condi-
the scan rate is raised to the reasonable value of 0.1 V s1 (Fig. 6 tions, i.e., conditions in which the substrate, CO2, is effectively
in ref. 89), the wave under examination seems perfectly rever- converted to the target product as fast as possible with the lowest
sible in the chemical sense and consequently does not contain possible overpotential. Evaluation of the performances of a
any usable information about the chemical step associated with given catalyst does require that the TOF–Z relationship be
electron transfer. We may thus conclude that, at this stage, the derived in preparative-scale conditions with minimization of
mechanism that leads to these interesting results will remain the various side-factors, starting with ohmic drop. For the
mysterious until the role of the electrode material has been purpose of catalyst screening such tedious procedures may be
deciphered.91,92 replaced by a quicker and simpler approach based on the
analysis of the cyclic voltammetric responses of the same catalytic
6.3 Copper and copper oxide electrodes
Published on 11 December 2012 on http://pubs.rsc.org | doi:10.1039/C2CS35360A

system. As at preparative-scale, various side-phenomena (substrate

Another example of the importance of the electrode material is consumption, catalyst deactivation, self-inhibition etc. . .) affect
provided by the results obtained with copper and copper oxides these responses resulting in deviations from the canonical S-shape
in water. Copper has long been noticed to be a peculiar and curves. These deviations that may be so strong as to result in
particularly interesting electrode material for CO2 reduction as peak-shaped responses increase as the current increases. They
Downloaded by SCD Université Paris 7 on 25 February 2013

allowing the production of methane and even ethylene.93–95 are thus minimized at the foot of the cyclic voltammetric curves.
The products however vary considerably according to the purity The foot-of-the-wave analysis thus allows a quick derivation of
and physical state of the surface as well as to the operating the intrinsic properties of the catalyst. One drawback of the
conditions.95 CO, formic acid and methanol can indeed be method is that it requires a careful control of the potential scale
obtained in variable amounts. A recent study reports that (ohmic drop, reference electrode drift, etc.. . .) precisely because
electrodes prepared by annealing thick layers of Cu2O obtained it treats foot-of-the-wave current–potential responses. Having
by air oxidation of Cu foils followed by electrochemical thus derived the intrinsic characteristics of the catalyst one
reduction of the Cu2O layer gives much better results in terms may decide whether or not it is worth considering for preparative
of rates and overpotentials than plain Cu electrodes.96 The scale investigations. Turnover frequency and overpotential may
main products were CO and formic acid with a few percents of then be selected so as to minimize the same side-phenomena that
CH4 and C2H4. Although these improved performances may be affect the high-current cyclic voltammetric responses. Particular
due to the high degree of roughness of the electrodes prepared attention should then be paid to ohmic drop or more generally to
in this way, it was noted that remaining Cu2O particles or other the potential distribution at the working electrode surface, which,
CuI species may play an important role. In this respect, catalysis if not mastered, may result in unexpected competitive product
of CO2 reduction by Cu2O is well documented, but the product formation. The question of the durability of the catalyst is also
is then methanol.97,98 The role of low valent oxides in the essential and can only be treated at the preparative scale. The
reduction of CO2 also appears important with other metals selection of an anodic reaction is also an important one especially
such as tin.99 if one wishes to go beyond the laboratory scale.
The most investigated homogeneous systems so far are
7. Conclusions and perspectives those that lead to high CO faradaic yields. Application of the
above tools has allowed benchmarking of these catalysts.
As a part of the response to contemporary energy challenges, The favorable role of acid addition has been shown to be
catalysis of the electrochemical reduction of CO2 is the object of important in several cases, leading to an efficient catalyst in
a current intense attention made of proposition of new systems which the acid is attached to the molecule itself. However, how
and re-evaluation of older ones. these acids interfere in the course of the catalytic reaction
In this context, new tools are being made available for char- remains to be uncovered. Reaction orders analysis susceptible
acterizing the catalytic properties of the molecules or systems that to give the answer should indeed be freed of undue application
are proposed. It has been shown that turnover frequency, TOF, of relationships that are relevant to the canonical S-shape
and overpotential, Z, of molecular catalysts are bound by a defined responses in cases where they are not observed. The foot-of-
relationship that depends upon its standard potential, standard the-wave analysis is worth considering in these conditions. One
rate constant of electron exchange with the electrode and a could then address by means of systematic investigations,
catalytic kinetic factor that may be expressed as its TOF0, fundamental mechanistic questions. For example, what is the
i.e., the value of the TOF at zero overpotential. The latter respective role of acid vs. H-bonding properties? In the CO2-to-
characterization is particularly pertinent for low Z values where CO conversion, how the breaking of one C–O bond is coupled
the TOF–Z relationship has the form of a Tafel law. This with proton and electron transfer? Proton (P) coupled (C)
approach applies to heterogeneous as well as to homogeneous electron (E) transfer (T) reactions are currently receiving a lot
molecular catalysts. A meaningful and fair comparison between of attention,100,101 but so far PCET coupled with heavy atom
the two situations implies taking into account in the definition bond breaking has been fully analyzed only in one case, that of
of turnover numbers and frequencies in the latter case only the an O–O bond.102 One may expect an answer to this problem

2434 Chem. Soc. Rev., 2013, 42, 2423--2436 This journal is c The Royal Society of Chemistry 2013
 View Article Online

Chem Soc Rev Review Article

once the kinetics of relevant steps have been extracted from the 15 C. Amatore, L. Nadjo and J. M. Savéant, New J. Chem., 1984, 8,
global catalysis kinetics. 565–566.
16 A. Gennaro, A. A. Isse, M.-G. Severin, E. Vianello, I. Bhugun and
Homogeneous catalysis by transition metal complexes may J.-M. Savéant, J. Chem. Soc., Faraday Trans., 1996, 92, 3963–3968.
be applied to the formation of products other than CO, notably 17 P. Neta, R. E. Huie and A. B. Ross, J. Phys. Chem. Ref. Data, 1988, 17,
formate and oxalate. Formate may indeed be considered as an 1027–1284.
18 J.-M. Savéant, Elements of Molecular and Biomolecular Electrochemistry,
interesting intermediate or, directly, for energy storage. In Wiley-Interscience, New York, 2006, ch. 4, pp. 251–252.
several cases, formate is deemed to result from the reaction 19 D. Lexa, J. M. Savéant, K. B. Su and D. L. Wang, J. Am. Chem. Soc.,
of an hydride donor intermediate with CO2. Systematic investiga- 1987, 109, 6464.
20 D. Lexa, J. M. Savéant, H. J. Schäfer, K. B. Su, B. Vering and
tions aiming at the determination of the factors that govern the D. L. Wang, J. Am. Chem. Soc., 1990, 112, 6162.
respective formation of formate, CO and hydrogen would certainly 21 A. Gennaro, A. A. Isse, J. M. Savéant, M. G. Severin and E. Vianello,
be welcome. Oxalate is formed specifically with aromatic esters J. Am. Chem. Soc., 1996, 118, 7190–7196.
22 C. Costentin, S. Drouet, M. Robert and J.-M. Savéant, J. Am. Chem.
anion radicals as catalysts. It is obtained at a remarkably much Soc., 2012, 134, 11235–11242.
Published on 11 December 2012 on http://pubs.rsc.org | doi:10.1039/C2CS35360A

lower overpotential – so low that the reduction can be performed in 23 (a) Scheme 1 is for a reductive process. Transposition to oxidation
the presence of dioxygen – by means of a dimeric CuI sulfide is straightforward; (b) A stoichiometric factor of 2 was unduly
omitted in the derivations and expressions of ref. 22. Correction
complex. Reduction of the oxalic acid thus obtained without submitted.
breaking the C–C bond is a definitely worthwhile objective. 24 C. Amatore and J.-M. Savéant, J. Electroanal. Chem., 1981, 123, 189–201.
In terms of heterogeneous catalysis of CO2 reduction, copper is 25 Ref. 18, ch. 2, pp. 132–139.
Downloaded by SCD Université Paris 7 on 25 February 2013

26 Other regimes may be analyzed along the same lines22.

also remarkable as an electrode material in the sense that methane 27 Ref. 18, ch. 1, pp. 28–32.
and even ethylene may be obtained. Other interesting products 28 In cyclic voltammetry or in related techniques about one millionth
may also be obtained and the role of CuI oxide has been of the substrate is consumed during each run at usual slow scan
rates (0.1 V s1), making these techniques ‘‘non-destructive’’ as
emphasized either as a precursor of activated Cu0 or, directly, as
opposed to bulk electrolyses, where the objective is instead to
the catalyst as observed when methanol is the target product. transform as much as possible of the substrate in the shortest
Methanol is also the main product with the remarkably possible time.
29 V. Fourmond, P.-A. Jacques, M. Fontecave and V. Artero, Inorg.
simple addition of protonated pyridine to the solution. The
Chem., 2010, 49, 10338–10347.
reaction does not occur on an inert electrode such as glassy 30 J.-M. Savéant and K.-B. Su, J. Electroanal. Chem., 1984, 171, 341–
carbon, leading to the idea that the electrode material, Pt or 349.
GaP under illumination, plays an essential role. The same is 31 Ref. 18, ch. 2, pp. 108–111.
32 J.-M. Savéant and E. Vianello, in Advances in Polarography, ed.
true with the addition of a liquid electrolyte, which allows the I. S. Longmuir, Pergamon Press, Cambridge, UK, 1959, pp. 367–374.
fast formation of CO at low overpotentials. 33 J.-M. Savéant and E. Vianello, Electrochim. Acta, 1965, 10, 905–920.
The role of the electrode material – mercury in this case – in 34 J. E. B. Randles, Trans. Faraday Soc., 1948, 44, 327–328.
35 A. Sevcik, Coll. Czech. Chem. Commun., 1948, 13, 349–377.
the remarkable efficiency of nickel cyclam as a catalyst of the 36 C. P. Andrieux, C. Blocman, J.-M. Dumas-Bouchiat, F. M’Halla and
CO2-to-CO conversion, known for a long time, also transpires in J.-M. Savéant, J. Electroanal. Chem., 1980, 113, 19–40.
recent studies in the area. 37 D. L. DuBois, A. Miedaner and R. C. Haltiwanger, J. Am. Chem. Soc.,
1991, 113, 8753–8764.
Overall, good theoretical and empirical starting points exist 38 P. R. Bernatis, A. Miedaner, R. C. Haltiwanger and D. L. DuBois,
but much work remains to be done before mechanisms are Organometallics, 1994, 13, 4835–4843.
clarified and efficient practical procedures are designed. 39 A. Miedaner, B. C. Noll and D. L. DuBois, Organometallics, 1997, 16,
5779–5791.
40 D. Steffey, C. J. Curtis and D. L. DuBois, Organometallics, 1995, 14,
References 4937–4943.
41 J. W. Raebiger, J. W. Turner, B. C. Noll, C. J. Curtis, A. Miedaner,
1 H. Takeda and O. Ishitani, Coord. Chem. Rev., 2010, 254, 346–354. B. Cox and D. L. DuBois, Organometallics, 2006, 25, 3345–3351.
2 D. Windle and R. N. Perutz, Coord. Chem. Rev., 2012, 256, 2562–2570. 42 M. Rakowski Dubois and D. L. Dubois, Acc. Chem. Res., 2009, 42,
3 Y. Izumi, Coord. Chem. Rev., DOI: 10.1016/j.ccr.2012.04.018. 1974–1982.
4 T. W. Woolerton, S. Sheard, E. Reisner, E. Pierce, S. W. Ragsdale 43 Another example of similar, and even more severe, inconsistency is
and F. A. Armstrong, J. Am. Chem. Soc., 2010, 132, 2132–2133. the application of the expression of the catalytic plateau, i=FS ¼
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
5 M. A. Méndez, P. Voyame and H. H. Girault, Angew. Chem., Int. Ed., 0
2kCA DP CP0 to the height of a sharp peak, calling this expression,
2011, 50, 7391–7394.
6 E. Lamy, L. Nadjo and J. M. Savéant, J. Electroanal. Chem., 1977, 78, the ‘‘Randles-Sevcik equation’’, while the actual Randles-Sevcik
403–407. equation pertains to a simple Nernstian diffusion controlled
7 H. A. Schwarz and R. W. Dodson, J. Phys. Chem., 1989, 93, 409–414. response (eqn (7)).
8 The use of visible light in photochemical processes, imposes rather 44 Z. Chen, C. Chen, D. R. Weinberg, P. Kang, J. J. Concepcion,
stringent constraints in terms of potentials accessible by visible D. P. Harrison, M. S. Brookhart and T. J. Meyer, Chem. Commun.,
light irradiations of dyes or semiconductors. 2011, 47, 12607–12609.
9 K. W. Frese Jr, in Elecrochemical and Electrocatalytic Reactions 45 Sometimes very low scan rates are used to obtain the expected
of Carbon Dioxide, ed. B. P. Sullivan, K. Krist and H. E Guard, canonical plateau-shaped wave as, e.g., in ref. 46 where scan rates
Elsevier, New York, ch. 6, 1993. as low as 2 mV s1 have been used. In these conditions the plateau
10 B. J. Fisher and R. Eisenberg, J. Am. Chem. Soc., 1980, 102, 7361–7363. shape may as well result from natural convection, which makes the
11 J.-M. Savéant, Chem. Rev., 2008, 108, 2348–2378. current–potential curves look like rotating disk electrode voltam-
12 E. E. Benson, C. P. Kubiak, A. J. Sathrum and J. M. Smieja, Chem. mograms in all circumstances.
Soc. Rev., 2009, 38, 89–99. 46 M. D. Rail and L. A. Berben, J. Am. Chem. Soc., 2011, 133, 18577–18579.
13 C. Amatore and J. M. Savéant, J. Am. Chem. Soc., 1981, 103, 5021–5023. 47 I. Bhugun, D. Lexa and J.-M. Savéant, J. Am. Chem. Soc., 1996, 118,
14 C. Amatore and J. M. Savéant, J. Electroanal. Chem., 1981, 125, 1769–1776.
23–39. 48 I. Bhugun and J.-M. Savéant, J. Electroanal. Chem., 1996, 408, 5–14.

This journal is c The Royal Society of Chemistry 2013 Chem. Soc. Rev., 2013, 42, 2423--2436 2435
 View Article Online

Review Article Chem Soc Rev

49 M. Hammouche, D. Lexa, J.-M. Savéant and M. Momenteau, 74 M. Fujihira, Y. Hirata and K. Suga, J. Electroanal. Chem., 1990, 292,
J. Electroanal. Chem., 1988, 249, 347–351. 199–215.
50 M. Hammouche, D. Lexa, M. Momenteau and J.-M. Savéant, J. Am. 75 M. Beley, J. P. Collin, R. Ruppert and J. P. Sauvage, J. Chem. Soc.,
Chem. Soc., 1991, 113, 8455–8466. Chem. Commun, 1984, 1315–1316.
51 I. Bhugun, D. Lexa and J.-M. Savéant, J. Phys. Chem., 1996, 100, 76 M. Beley, J. P. Collin, R. Ruppert and J. P. Sauvage, J. Am. Chem.
19981–19985. Soc., 1986, 108, 7461–7467.
52 I. Bhugun, D. Lexa and J.-M. Savéant, J. Am. Chem. Soc., 1994, 116, 77 J. P. Collin, A. Jouaiti and J. P. Sauvage, Inorg. Chem., 1988, 27,
5015–5016. 1986–1990.
53 I. Bhugun, D. Lexa and J.-M. Savéant, J. Am. Chem. Soc., 1996, 118, 78 J. Schneider, H. Jia, K. Kobiro, D. E. Cabelli, J. T. Muckerman and
3982–3983. E. Fujita, Energy Environ., 2012, 5, 9502–9510.
54 Electrogenerated Fe(0)TPP is sometimes given the reputation of 79 G. B. Balazs and F. C. Anson, J. Electroanal. Chem., 1992, 322,
having a good current efficiency but requiring a hopelessly large 325–345.
overpotential and to require mercury as the electrode.2,11 The 80 G. B. Balazs and F. C. Anson, J. Electroanal. Chem., 1993, 361,
TOF–Z curve for this catalyst represented in Fig. 8c shows that 149–157.
the first of these assertions is inaccurate. Mercury is not required. 81 B. A. Rosen, A. Salehi-Khojin, M. R. Thorson, W. Zhu,
The results are closely the same on glassy carbon. Mercury was D. T. Whipple, P. J. A. Kenis and R. I. Masel, Science, 2011, 334,
Published on 11 December 2012 on http://pubs.rsc.org | doi:10.1039/C2CS35360A

preferred in most cases because scan-rate dependency investiga- 643–644.

tions are easier than on glassy carbon. The situation is in this sense 82 J.-M. Savéant, J. Phys. Chem. B, 2001, 105, 8995–9001.
quite different from that of catalysts for which mercury is essential 83 J.-M. Savéant, J. Am. Chem. Soc., 2008, 130, 4732–4741.
to their performances as discussed in Section 6.1. 84 A. S. Barnes, E. I. Rogers, I. Streeter, L. Aldous, C. Hardacre,
55 K.-Y. Wong, W.-H. Chung and C.-P. Lau, J. Electroanal. Chem., 1998, G. G. Wildgoose and R. G. Compton, J. Phys. Chem. C, 2008, 112,
13709–13715.
Downloaded by SCD Université Paris 7 on 25 February 2013

453, 161–169.
56 J. Hawecker, J. M. Lehn and R. Ziessel, J. Chem. Soc., Chem. 85 M. Islam and T. Ohsaka, J. Phys. Chem. C, 2008, 112, 1269–1275.
86 G. Seshadri, C. Lin and A. B. Bocarsly, J. Electroanal. Chem., 1994,
Commun., 1984, 328–330.
57 M. Smieja, E. E. Benson, B. Kumar, K. A. Grice, C. S. Seu, A. J. M. 372, 145–150.
87 E. E. Barton, D. M. Rampulla and A. B. Bocarsly, J. Am. Chem. Soc.,
Miller, J. M. Mayer and C. P. Kubiak, Proc. Natl. Acad. Sci. U. S. A.,
2008, 130, 6342–6344.
2012, 109, 15646–15650.
88 E. Barton Cole, P. S. Lakkaraju, D. M. Rampulla, A. J. Morris,
58 C. Costentin, S. Drouet, M. Robert and J.-M. Savéant, Science, 2012,
E. Abelev and A. B. Bocarsly, J. Am. Chem. Soc., 2010, 132,
338, 90–94.
11539–11551.
59 J. W. Raebiger, J. W. Turner, B. C. Noll, C. J. Curtis, A. Miedaner,
89 A. J. Morris, R. T. McGibbon and A. B. Bocarsly, ChemSusChem,
B. Cox and D. L. DuBois, Organometallics, 2006, 25, 3345–3351.
2011, 4, 191–196.
60 M. Bourrez, F. Molton, S. Chardon-Noblat and A. Deronzier, Angew. 90 A. B. Bocarsly, Q. D. Gibson, A. J. Morris, R. P. L’Esperance,
Chem., Int. Ed., 2011, 50, 9903–9906. Z. M. Detweiler, P. S. Lakkaraju, E. L. Zeitler and T. W. Shaw,
61 Z. Chen, C. Chen, D. R. Weinberg, P. Kang, J. J. Concepcion, ACS Catal., 2012, 2, 1684–1692.
D. P. Harrison, M. S. Brookhart and T. J. Meyer, Chem. Commun., 91 Recent quantum chemical calculations of the characteristics of the
2011, 47, 12607–12609. pyH+/pyH couple conclude that it cannot be involved in the
62 J. D. Froehlich and C. P. Kubiak, Inorg. Chem., 2012, 51, 3932–3934. reaction.93 Such calculations were not in fact necessary: it suffices
63 T. C. Johnson, D. J. Morris and M. Wills, Chem. Soc. Rev., 2010, 39, to note that the reaction does not occur at inert electrodes. Moreover
81–88. these calculations do not seem to reproduce correctly the experi-
64 S. Slater and J. H. Wagenknecht, J. Am. Chem. Soc., 1984, 106, mental behavior of pyH+ reduction at an inert electrode.
5367–5368. 92 J. A. Keith and E. A. Carter, J. Am. Chem. Soc., 2012, 134, 7580–7583.
65 G. Pilloni, E. Vecchi and M. Martelli, J. Electroanal. Chem., 1973, 45, 93 Y. Hori, K. Kikuchi and S. Suzuki, Chem. Lett., 1985, 1695–1698.
483–487. 94 Y. Hori, K. Kikuchi, A. Murata and S. Suzuki, Chem. Lett., 1986,
66 H. Ishida, H. Tanaka, K. Tanaka and T. Tanaka, J. Chem. Soc., 897–898.
Chem. Commun., 1987, 131–132. 95 Y. Hori, in Modern Aspects of Electrochemistry, ed. C. G. Vayenas,
67 C. M. Bolinger, N. Story, B. P. Sullivan and T. J. Meyer, Inorg. Chem., R. E. White and M. E. Gamboa-Aldeco, Springer, New York, 2008,
1988, 27, 4582–4587. vol. 42, p. 89.
68 C. Caix, S. Chardon-Noblat and A. Deronzier, J. Electroanal. Chem., 96 C. W. Li and M. W. Kanan, J. Am. Chem. Soc., 2012, 134, 7231–7234.
1997, 434, 163–170. 97 K. W. Frese, J. Electrochem. Soc., 1991, 138, 3338–3344.
69 P. Kang, C. Cheng, Z. Chen, C. K. Schauer, T. J. Meyer and 98 M. Le, M. Ren, Z. Zhang, P. T. Sprunger, R. L. Kurtz and J. C. Flake,
M. Brookhart, J. Am. Chem. Soc., 2012, 134, 5500–5503. J. Electrochem. Soc., 2011, 158, E45–E49.
70 T. Reda, C. M. Plugge, N. J. Abram and J. Hirst, Proc. Natl. Acad. Sci. 99 Y. Chen and M. W. Kanan, J. Am. Chem. Soc., 2012, 134, 1986–1989.
U. S. A., 2008, 105, 10654–10658. 100 Chem. Rev. thematic issue ‘‘Proton-Coupled Electron Transfer’’
71 M. D. Rail and L. A. Berben, J. Am. Chem. Soc., 2011, 133, 18577–18579. 2010, 110, 6937–7100.
72 B. Kumar, J. M. Smieja, A. F. Sasayama and C. P. Kubiak, Chem. 101 Energy Environ. Sci. themed issue ‘‘Proton-Coupled Electron Transfer’’
Commun., 2012, 48, 272–274. 2012, 5, 7677–7694.
73 R. Angamuthu, P. Byers, M. Lutz, A. L. Spek and E. Bouwman, 102 C. Costentin, V. Hajj, M. Robert, J.-M. Savéant and C. Tard, Proc.
Science, 2010, 327, 313–315. Natl. Acad. Sci. U. S. A., 2011, 108, 8559–8564.

2436 Chem. Soc. Rev., 2013, 42, 2423--2436 This journal is c The Royal Society of Chemistry 2013

View publication stats