Sie sind auf Seite 1von 16

View Article Online / Journal Homepage / Table of Contents for this issue

Chem Soc Rev Dynamic Article Links

Cite this: Chem. Soc. Rev., 2012, 41, 2036–2051


Published on 14 December 2011. Downloaded by INDIAN INSTITUTE OF TECHNOLOGY BOMBAY on 10/10/2016 10:04:09.

www.rsc.org/csr TUTORIAL REVIEW


Thermodynamics and kinetics of CO2, CO, and H+ binding to the metal
centre of CO2 reduction catalysts
Jacob Schneider,*a Hongfei Jia,b James T. Muckermana and Etsuko Fujita*a
Received 6th October 2011
DOI: 10.1039/c1cs15278e

In our developing world, carbon dioxide has become one of the most abundant greenhouse gases
in the atmosphere. It is a stable, inert, small molecule that continues to present significant
challenges toward its chemical activation as a useful carbon end product. This tutorial review
describes one approach to the reduction of carbon dioxide to carbon fuels, using cobalt and
nickel molecular catalysts, with particular focus on studying the thermodynamics and kinetics of
CO2 binding to metal catalytic sites.

1. Introduction of gases, and 1 M solutes. Because the one-electron reduction


of CO2 to CO2  requires nearly two eV of free energy, a
There are both thermodynamic and kinetic barriers to overcome proton-assisted approach to CO2 reduction, such as those in
for successful reduction of CO2 to carbon products. The former eqn (2)–(6), lowers the thermodynamic barrier significantly.
are outlined in eqn (1)–(6). Eqn (7) shows the thermodynamic Transition-metal complexes can have accessible multiple redox
potential of a competing reaction for CO2 reduction, and that is states that promote such proton-assisted, multi-electron path-
the reduction of H+ to hydrogen. Here Eo0 is the formal potential ways. Molecular systems that are capable of catalysing the
vs. the normal hydrogen electrode (NHE) under standard reduction of CO2 typically give the products CO and/or formic
conditions in aqueous solution: pH 7, 25 1C, 1 atmosphere acid (HCO2H), but examples that reduce CO2 to higher energy
products such as methanol and methane are rare, typically
a
Chemistry Department, Brookhaven National Laboratory, Upton, inefficient, and not selective. Fig. 1 shows a Latimer–Frost
NY 11973-5000, USA. E-mail: jschneider@bnl.gov, fujita@bnl.gov diagram1 for the multi-electron, multi-proton reduction of
b
Materials Research Department, Toyota Research Institute of North
America, Toyota Motor Engineering & Manufacturing North CO2 in homogeneous aqueous solution at pH 7 (red points);
America, Inc., 1555 Woodridge, Ann Arbor, MI 48105, USA the dashed lines correspond to the potentials in eqn (2)–(6).

Jacob Schneider obtained his BS Hongfei Jia is a Senior


degree from the University of Research Scientist in the
North Carolina at Asheville, Materials Research Depart-
graduating in 2003 with distinc- ment at Toyota Research
tion as a University Research Institute of North America.
Scholar. He earned his PhD He received his BE and ME
from the University of Rochester degrees in Chemical Engineering
in 2008, under the direction of from Tianjin University, and
Richard Eisenberg, and went on his PhD also in Chemical
to study with Robin N. Perutz as Engineering from the Univer-
an EPSRC Postdoctoral Fellow sity of Akron with Ping Wang
at the University of York in in 2005. After graduation, he
York, England. Jacob joined joined Toyota first as a Post-
the photocatalysis group at doc Researcher and then
Jacob Schneider BNL in January 2011 as a Hongfei Jia became a scientist in 2007.
Research Associate, and is He is the recipient of 2005
currently interested in photo- and electrocatalytic reduction of small Chemstress Outstanding Graduate Student Award and also the
molecules toward sustainable fuels. co-organiser of the symposium on solar cells and solar fuels at
239th ACS national meeting. While his earlier research focused
mainly on biocatalysis and biofuel cells, currently he is working
on solar energy utilisation, CO2 recycling and functional
coatings.

2036 Chem. Soc. Rev., 2012, 41, 2036–2051 This journal is c The Royal Society of Chemistry 2012
View Article Online

The intermediates in such a diagram, all of which lie well CO2 + e - CO2  Eo 0 = 1.90 V (1)
above the appropriate dashed line (except for CO as an
Published on 14 December 2011. Downloaded by INDIAN INSTITUTE OF TECHNOLOGY BOMBAY on 10/10/2016 10:04:09.

intermediate for the reaction of CO2 to H2CO), depend on CO2 + 2H+ + 2e - CO + H2O Eo 0 = 0.53 V (2)
the properties of the catalyst (if any). For any given product,
CO2 + 2H+ + 2e - HCO2H Eo 0 = 0.61 V (3)
the intermediate (point) that lies farthest in DG/n above the
corresponding dashed line (i.e., the distance in free energy CO2 + 4H + 4e - HCHO + H2O
+  o0
E = 0.48 V (4)
above a dashed line divided by the number of electrons added
from the starting point, CO2) determines the intrinsic thermo- CO2 + 6H+ + 6e - CH3OH + H2O Eo 0 = 0.38 V
dynamic overpotential for that process. Alternative intermedi- (5)
ates are indicated in Fig. 1 by the blue points and labels.
CO2 + 8H+ + 8e - CH4 + H2O Eo 0 = 0.24 V (6)
Odd-electron intermediates tend to be relatively high-energy
species compared to even-electron intermediates,2 so, apart 2H+ + 2e - H2 Eo 0 = 0.41 V (7)
from CO2 , we have listed and shown only the even-electron
intermediates. One of the key steps in catalytic CO2 reduction involves the
binding of CO2 to a vacant coordination site at the metal
centre of a reduced catalyst (eqn (8) and (9)). The kinetics and
thermodynamics of the process (eqn (10)) can have a major
effect on the overall mechanism and efficiency of the catalytic
cycles. Therefore, it is important that the binding step be fully
characterised and understood. This is particularly apparent
when inert, stable molecules such as CO2 are used as the
substrate. Carbon dioxide is a linear molecule that usually
binds to a metal catalyst through the electrophilic carbon
atom. This causes a change in geometry of CO2 from linear
to bent, a difficult rearrangement that may require a large
excess of energy, especially in the absence of protons. The
solubility of CO2 in organic and aqueous solutions is also a
limiting factor for studying it as a substrate for catalytic
activation. For example, the solubility of CO2 in aqueous
Fig. 1 Latimer–Frost diagram for the multi-electron, multi-proton
solution is ca. 0.03 M while that in MeCN is roughly ten times
reduction of CO2 in homogeneous aqueous solution at pH 7 based greater. Further, CO2 equilibria in aqueous solution are
partially on the potentials in eqn (1)–(6). Any species lying above complicated to say the least.3 Equilibrium and rate constants
the straight line joining two adjacent points (e.g., CO2  between CO2 that favour the product of a given CO2 reduction reaction (e.g., CO)
and HCOOH) is thermodynamically unstable with respect to can have an overall adverse effect on catalytic ability. However,
disproportionation. kinetic parameters for CO2 binding (eqn (10)) need to be small

Jim Muckerman, a Senior Etsuko Fujita is a Senior


Chemist and the Deputy Chair Chemist and the group leader
for Strategic Planning in the in photocatalysis group in the
Chemistry Department at Chemistry Department at
BNL, received his BA degree BNL and the recipient of
in Chemistry from Carleton the 2008 BNL Science and
College, and his PhD in Technology Award for out-
Physical Chemistry from the standing research in solar fuels
University of Wisconsin with generation. She received a BS in
Richard B. Bernstein in 1969. Chemistry from Ochanomizu
Since that time he has been a University, Tokyo, and a
staff scientist in theoretical PhD in Chemistry from the
physical chemistry at BNL, Georgia Institute of Technology.
where he served as the Her research interests span
James T. Muckerman group leader in ‘‘Gas-Phase Etsuko Fujita solar fuels generation
Molecular Dynamics’’ for a including water splitting and
number of years, and is currently working in the area of solar CO2 utilization, mechanistic inorganic chemistry, and thermo-
fuels. He has been a John Simon Guggenheim Fellow, and the dynamics/kinetics of small molecule binding/activation. She is
organizer of several conferences and symposia. an advisory board member for several solar energy conversion
projects including the NSF funded the Center for Chemical
Innovation Project ‘‘Powering the Planet’’; the Japan Science
and Technology, PRESTO (i.e., Precursory Research for
Embryonic Science and Technology) Project entitled ‘‘Light
Energy and Chemical Conversion’’.

This journal is c The Royal Society of Chemistry 2012 Chem. Soc. Rev., 2012, 41, 2036–2051 2037
Published on 14 December 2011. Downloaded by INDIAN INSTITUTE OF TECHNOLOGY BOMBAY on 10/10/2016 10:04:09. View Article Online

Scheme 1 Proposed mechanism(s) of CO2 and water reductions (photo- or electrocatalytic) by M (Co or Ni) ‘cyclam-like’ complexes; H2
production (blue), HCO2 production (red), CO formation (brown), and putative intermediates (green). D is a highly reduced, carbon-centred
radical species, such as Et2NC HCH3 or (HOC2H4)2N(C HCH2OH).7

enough not to form a stable ML(CO2)+ adduct, yet large between the two processes are different but the catalytic steps in
enough for catalysis to occur. both photo- and electrocatalytic systems are comparable.
MIIL2+ + e " MIL+ (8)
kf
1.1. Scope of review
MI Lþ þ CO2 Ð MLðCO2 Þþ ð9Þ
kr This tutorial review considers experiments that deal with the
+ I + isolation and characterisation of complex intermediates and
KCO2 = [ML(CO2) ]/{[M L ][CO2]} = kf/kr (10)
their kinetics, as they pertain to the binding of substrates to
In this review, we describe both photo- and electrocatalytic reduced metal macrocycles, and how the data correlate with
CO2 reduction systems. In the photocatalytic systems (see catalytic activity. There are a number of recent reviews that
Scheme 1), light is required to excite ground-state chromo- describe the features of photo- and electrocatalytic reduction
phores that, once in the excited-state, are reductively quenched of CO2 using metal catalysts, e.g., catalytic efficiencies; product
by a sacrificial electron donor, typically a tertiary amine, and turnover numbers, frequencies, and selectivities; catalyst stability;
transfer an electron to a separate, molecular catalyst (MIIL2+) etc.,4–10 and interested readers are encouraged to seek those out
to produce a reduced state of the catalyst (MIL+, see eqn (8)). for more information. Because the literature has been extensively
The reduced species reacts further with the substrate, CO2 (or reviewed in this context, we have chosen to focus on thermo-
H+, CO, etc.). If a ML(H)2+ is formed (eqn (11)) CO2 may insert dynamic and kinetic experiments using metal macrocyclic
into the M–H bond to give the formate adduct, M–OCHO catalysts and how those parameters relate to CO2 reduction
(eqn (12)). The CO, formate (HCO2), and H2 yields depend (and H+ reduction), for which there are no current accounts.
on the binding equilibria and the rates of various competing In particular, we will describe properties associated with
reactions; CO and HCO2 remain desirable end products of CO2 cobalt and nickel complexes of the macrocyclic ligand
reduction and H2 a desirable side product. 1,4,8,11-tetraazacyclotetradecane (cyclam) and its derivatives
in Chart 1. The importance of determining parameters such as
MIL+ + H+ - ML(H)2+ (11) the equilibrium constant of substrate binding, KS (S =
ML(H) 2+
+ CO2 - ML(OCHO) 2+
(12) substrate such as CO2, CO, or H+), to catalytic centres will
be discussed. These studies can be important for developing
In an electrocatalytic system, there is no need for a sacrificial efficient molecular catalysts, and we outline below how to
reagent or even a chromophore. The reducing equivalents for determine such parameters from five example techniques.
eqn (8) come directly from the electrode surface and electrons are While much of the seminal work in this field has postulated
transferred either following diffusion of the catalyst to the elec- the involvement of intermediate species, little spectroscopic
trode surface or by catalyst adsorption onto the electrode surface. evidence for such intermediates was obtained. It should be
Electricity obtained by photovoltaic or dye-sensitized solar cells understood that although intermediates cannot always be
can be used for the catalytic reduction of CO2. The initial steps isolated, kinetic and thermodynamic studies can provide

2038 Chem. Soc. Rev., 2012, 41, 2036–2051 This journal is c The Royal Society of Chemistry 2012
Published on 14 December 2011. Downloaded by INDIAN INSTITUTE OF TECHNOLOGY BOMBAY on 10/10/2016 10:04:09. View Article Online

Chart 1 Molecular structures of the macrocycle derivatives of 1,4,8,11-tetraazacyclotetradecane (L1, cyclam).

better insight into the existence of postulated reaction sweep methods, e.g., cyclic voltammetry (CV), in solutions that
intermediates, such as those suggested in Scheme 1. have been purged with an inert gas (Ar or N2). When CO2 or
Metallo-porphyrins, phthalocyanines, corroles, and chlorins have CO is introduced into the solution the E1/2red—usually the MII/I
also shown evidence for electro- and photocatalytic reduction of couple—of the metal macrocycle may shift or become an
CO2, but there have been no reports of CO2 adducts of such irreversible process. When electron transfer and forward (kf)
metallo-macrocycles due to the highly reactive nature of the and reverse (kr) reactions (e.g., eqn (8)–(10)) are sufficiently
unsaturated system(s).7 However, it has been suggested that the rapid, and KS is relatively large so that the system is always in
catalytic reaction pathways using metal ‘cyclam-like’ catalysts are equilibrium, KS can be determined by measuring the change in
similar to the reaction pathways for metal ‘porphyrin-like’ catalysts, E1/2red in solutions that contain different amounts of the
except for the involvement of catalytically active species such as M0 dissolved substrate. A value for KS can be obtained by taking
for ‘porphyrin-like’, instead of MI for ‘cyclam-like’ catalysts.7 the exponential of the intercept from a linear plot of E1/2red vs.
ln[S], derived from the modified Nernst relationship in eqn (13),
where Eo 0 is the MII/I reduction potential in the absence of S
2. Overview of methods (i.e., under Ar or N2), R is the gas constant in J mol1 K1, T is
It is far beyond the extent of this review to cover every the temperature in degrees Kelvin, n is the number of electrons
possible method/technique that may be used for isolation and involved in the reduction process (one for the case of MII/I), F is
characterisation of reaction intermediates, but the ability to use the Faraday constant in C mol1, and q is the number of ligands
multiple techniques is certainly advantageous. This section describes involved in binding to the metal centre (one in the case of
five methods that have proven useful. Electrochemical methods are macrocyclic complexes).11,12 This new half-potential, E1/2red,
suitable for every chemistry laboratory with regard to instrumenta- modified by the binding of the substrate that removes MI from
tion, but the technique and analysis can be quite demanding. the solution, refers to the MII/MIS couple with [MII] = [MIS].
Spectroscopic methods involving chemical reduction, laser flash E1/2red = Eo 0 + (RT/nF)ln(KS) + q(RT/nF)ln[S] (13)
photolysis, and pulse radiolysis are also considered. Finally,
theoretical studies based on DFT calculations have become An example of the CO2 binding determination is shown in
ubiquitous, and hybrid DFT calculations with large basis sets Fig. 2. From a plot of E1/2red vs. ln[CO2], Gangi and Durand
are now quite feasible and can be used for predicting free- obtained KCO2 of 7  104 M1 in dimethylsulfoxide (DMSO)
energy profiles of catalytic reactions in addition to geometric for CoIL6+ (L6 = 5,7,7,12,14,14-hexamethyl-1,4,8,11-tetra-
and electronic structures. Other methods of characterising azacyclotetradeca-4,11-diene, Chart 1). Throughout this
intermediates are presented throughout the text. review values of KCO2 (and KCO) are quoted in units of M1,
as the solubility of CO2 (and CO) in a given solvent was taken
2.1. Electrochemical
from the literature or determined by known methods.13–17
One typically measures the reduction potentials (E1/2red) for Electrochemistry is an excellent method for determining the
cobalt or nickel ‘cyclam-like’ macrocycles using electrochemical binding constants for metal macrocycles with CO2 or CO, but

This journal is c The Royal Society of Chemistry 2012 Chem. Soc. Rev., 2012, 41, 2036–2051 2039
Published on 14 December 2011. Downloaded by INDIAN INSTITUTE OF TECHNOLOGY BOMBAY on 10/10/2016 10:04:09. View Article Online

Fig. 3 UV-Vis spectra of CoIL6+ (A) and CoL6(CO2)+ (B) in


MeCN at 25 1C. Inset: ln(KCO2) vs. 1000/T. The KCO2 values at
Fig. 2 Cyclic voltammograms for 8.5  104 M CoIIL62+ dissolved in 25–77 1C were calculated from the 530 and 680 nm absorbances of
0.10 M tetraethylammonium perchlorate in DMSO at a graphite electrode 0.1–1 mM CoL6(CO2)+ solution containing 0.01–0.1 M CO2. The
(area 0.17 cm2); scan rate 200 mV s1. (A) Under nitrogen; (B) under CO2, figure has been reprinted with permission from ref. 19. Copyright 1988
and (C), more negative scan under CO2. The figure has been reprinted American Chemical Society.
with permission from ref. 12. Copyright 1986 Royal Society of Chemistry.
unwanted colloids, or reduction of the macrocycle itself can
often there is no shift in the E1/2red in the presence of either gas. lead to demetalation and product decomposition. Further,
Such behaviour suggests that the complex does not interact Na(Hg) is thermodynamically capable of direct reduction of
with the substrate on the time scale of the CV, or the binding CO2 to CO2 . The biggest drawback to Na(Hg) reduction is
constant is too low to be measured by electrochemical the experimental set-up. Methods used in our laboratory
methods. In some other cases, the CV becomes irreversible require special skills in glass-blowing, Schlenk line techniques,
upon introduction of CO2, and only the cathodic component and the handling of reactive Na and toxic Hg vapour.
of the voltammogram may be observed. This can occur, for An example study investigated the reversible binding of CO2
example, when electron transfer and forward reactions are to CoIL6+ in MeCN (eqn (9)).19 When solutions of
sufficiently rapid (and KCO2 is also large) but the reverse CoL6(CO2)+ were heated at reduced CO2 pressure, partial
reaction is slow. Simulation of CVs can be utilized to obtain dissociation to the parent complex was observed. The KCO2
the thermodynamic and kinetic parameters using commercially values and other thermodynamic parameters (DGo(298) =
available programs such as BASi DigiSims Simulation (5.5  1.0) kcal mol1, DHo = (5.4  1.0) kcal mol1,
Software for Cyclic Voltammetry or DigiElch Electrochemical DSo = (+0.4  3) cal K1 mol1) were characterised by
Simulation Software.18 UV-Vis spectroscopy (Fig. 3) using eqn (10) and (14).
DGo = DHo  TDSo = RTln(KCO2) (14)
2.2. Spectroscopic
4 1
The binding constant KCO2 = (1.2  0.5)  10 M at 298 K
2.2.1. Electrochemical and chemical reduction. Spectro-
in MeCN is slightly smaller than that reported for DMSO
scopy is an alternative approach to obtain KS. Bulk solutions
(7  104 M1), as determined from CV experiments.16
of MI macrocycles may be generated by electrochemical or
chemical reduction of MII (or from MIII for cobalt). It is then 2.2.2. Laser flash photolysis. Transient-absorption spectra,
possible to introduce various concentrations of the substrate lifetimes of various reactive intermediates MI and M–CO2
and study the resulting species (e.g., M–CO2 or M–CO) by (or M–CO), and KCO2 can be measured by flash-quench
spectroscopic methods. Bulk electrolysis is used to create experiments using a photosensitiser (PS), a quencher (i.e., an
solutions of MI macrocycles electrochemically, but isolation electron donor, D) and an MII species (eqn (9) and (15)–(19)).
of intermediates formed via this method is complicated by the
presence of high concentrations of electrolyte in solution. PS + hn - PS* (15)
Chemical reduction of bulk solutions in aprotic solvents, such
PS* + D - PS + D  +
(16)
as acetonitrile (MeCN) and tetrahydrofuran (THF), is frequently
accomplished using sodium amalgam (Na(Hg)) as a reducing PS + MIIL2+ - PS + MIL+ (17)
agent. A mixture of Na and Hg with 0.5–1% Na has reducing
power of ca. 2.0 V vs. NHE. The solvents must be rigorously KCO2 = [ML(CO2)+]/{[MIL+][CO2]} = DA/{AN[CO2]}
dried, but only small quantities of the complex are necessary for (18)
spectroscopic studies. As with bulk electrolysis at very negative kobs = kf[CO2] + kr (19)
potentials, there are disadvantages to reduction by Na(Hg) as
well. The strong reducing nature of Na(Hg) can reduce the metal Fig. 4 (left) shows an example of a transient decay curve for
macrocycles beyond their catalytically active MI state yielding CoIL6+ monitored at 670 nm in a MeCN/methanol (MeOH)

2040 Chem. Soc. Rev., 2012, 41, 2036–2051 This journal is c The Royal Society of Chemistry 2012
Published on 14 December 2011. Downloaded by INDIAN INSTITUTE OF TECHNOLOGY BOMBAY on 10/10/2016 10:04:09. View Article Online

Fig. 4 (left) Transient decay curve of CoIL6+ monitored at 670 nm for a solution containing 0.1 mM terphenyl, 1 mM CoIIL62+ and 0.53 mM
CO2 in MeCN/MeOH. (right) Kinetic behaviour of the CoIL6+ in the presence of CO2, as shown by a plot of the pseudo-first-order rate constant
vs. [CO2] to give kobs = 1.7  108 M1 s1. The figure has been reprinted with permission from ref. 20. Copyright 1995 American Chemical Society.

solution containing 0.1 mM terphenyl, 1 mM CoIIL62+, and From these data it is possible to apply conditions necessary
0.53 mM CO2; and (right) kinetic behaviour of the CoIL6+ in to generate a reactive species for study. For example, in
the presence of CO2.20 When CO2 is introduced into the system order to investigate the reduction chemistry of CoIIL62+ the
containing terphenyl and CoIIL62+, the lifetime of CoIL6+ addition of  OH scavengers such as tert-butanol (t-BuOH) is
changes dramatically. The variation of the observed rate constants essential, since CoIIL62+ reacts very rapidly with the  OH
for CoIL6+ decay with CO2 concentrations (0.53 to 7.2 mM) is radical. Formate (HCO2) can also scavenge  OH to form the
shown in the right panel of Fig. 4. The CO2 binding rate constant CO2 radical, CO2 , a strong reducing reagent (eqn (1)). The
of 1.7  108 M1 s1 is obtained from the slope of the line-fit in CO2  can also be produced by a reaction of dissolved CO2
the figure. As shown in eqn (18), the binding equilibrium constant with eaq. While CoIIL62+ can be reduced to CoIL6+ using
of 1.1  104 M1 can be calculated using the decay traces because eaq, CO2  can attach to the metal centre to form
the initial and final absorbances of CoIL6+ at 670 nm can give the CoL6(CO2)+ in addition to producing a small amount of
ratio of the total CoL6(CO2)+ complex (DA) to the unreacted CoIL6+ (Scheme 2). Creutz et al. observed that in the CoIL6+
CoIL6+ complex (AN) using the fact that the CO2 adducts have system (which has found application in both the photo-
no significant absorption in the visible region of the spectrum. reduction of water to H2, and electroreduction of CO2 to CO),
Alternatively, KCO2 can be obtained by measuring kobs using all three adducts CoL6(H)2+, CoL6(CO2)+, and CoIL6(CO)+
solutions with various [CO2] (eqn (10) and (19)). However, in are present at equilibrium (on the time scale of seconds), and
the case of CoIL6+, the error in kr is too large and the KCO2 could equilibration is achieved exclusively via dissociation to
not be determined with this method. CoIL6+.22 The KCO2, KCO and KH+ for CoIL6+ will be
discussed in Section 3.
2.2.3. Pulse radiolysis. Pulse radiolysis is a unique type of The benefit of pulse radiolysis lies in the fact that it is a
spectroscopy that complements laser flash-quench experiments. simple and clean technique used to generate and study the
Although the technique is somewhat limited to use in aqueous reactivity of metal macrocycles. While the chemistry behind
solution, it is certainly possible to perform experiments in pulse radiolysis is quite complex, significant quantities of
organic media, but the possibility of radical reactions in organic data may be obtained from experiments simply by applying
solvents only complicates the species being studied. Since pulse different reaction conditions. Examples of output data include:
radiolysis does not require a photosensitiser, the reduction UV-Vis spectra of transient species and their extinction
chemistry of species that do not have high extinction coefficients coefficients, pKa isotherms, binding constants, and reaction
such as Co–H and NiI or Ni–S (S = CO2, H) complexes of rate constants.
‘cyclam-like’ macrocycles can be conveniently investigated.
Through the radiolysis of water (eqn (20)), which occurs by
rapid electron injection generated by an accelerator (e.g., a Van
de Graaff generator) or by radioactive rays (e.g., 60Co g-rays),
free radicals are created and their concentrations are shown as
the number in parentheses representing the G-value (number of
molecules formed per 100 eV) of each species.

H2O eaq (2.6),  H (0.6),  OH (2.7), H2 (0.45), H2O2 (0.7)


(20)

The rate constants for the free radicals generated in eqn (20) are Scheme 2 Reactions of CoIIL62+ with eaq and CO2 , and reac-
well documented, as are their reactions with a number of reagents tions of CoIL6+ with CO2, CO, and H+, as studied by the pulse
to create reactive, unstable organic and inorganic radicals.21 radiolysis technique.

This journal is c The Royal Society of Chemistry 2012 Chem. Soc. Rev., 2012, 41, 2036–2051 2041
View Article Online

2.3. Theoretical It should be noted that the absolute standard reduction


potential so computed does not correspond to the standard
Published on 14 December 2011. Downloaded by INDIAN INSTITUTE OF TECHNOLOGY BOMBAY on 10/10/2016 10:04:09.

There are a number of quantum chemistry/density functional


convention of electrochemistry which references all reduction
programs in widespread use, and an even greater variety in the
potentials to that of the normal hydrogen electrode (NHE).
details of how they are used to elucidate the reaction pathways
For consistency, the absolute standard reduction potential of
and energetics of molecular catalysis. An explanation of the
the NHE is computed,
many levels and methods of theory exceeds the limits of this
review. A good starting point is to carry out computations H+(s) + e(g) - (1/2)H2(g) (25)
with hybrid density functional methods using, e.g., the Gaussian 09 
EoNHE,abs = [12G*(H2(g)) +
 G*(H(s) )  G*(e(g) )]/nF
suite of programs.23 Geometry optimisations and vibrational
(26)
frequency calculations are required in order to compute the
absolute free energy of each species involved in the chemical using a calculated value of G*(H2(g)) with n = 1. The resulting
reaction under study. In the context of metal macrocycles, standard reduction potential is then Eored vs. NHE = Eored,abs 
calculations of all species in a polarisable continuum model of EoNHE,abs.
the solution phase (CPCM) seem more reliable than calculations The value of Go(e(g)) = 0.868 kcal mol1 24 for the
in vacuum. For example, as discussed in Section 4.1, CO2 does standard free energy of the gas-phase electron and the values
not bind to the Ni centre in RSRS-NiL1(CO2)+ in the gas phase, G*(H+(s)) = 266.5 kcal mol1 in MeCN and 272.2 kcal mol1
but it does so in a CPCM solvation model of water, probably in water are recommended in the literature.25–28
because the interaction with the dielectric continuum stabilises
the charge separation between the Ni centre and the CO2 moiety, 3. Cobalt tetraazacyclam macrocycles
while the vacuum does not.
Once the absolute free energies in solution (except for gases) of Cobalt ‘cyclam-like’ complexes have been studied as CO2
all relevant species are computed, the problem of calculating reduction catalysts for many years. In fact, CoIL6+ and
standard free energies of reaction, equilibrium (or binding) CoIL1+ act as catalysts for the photochemical29–31 and
constants, pKa values, and standard reduction potentials may electrochemical29,32 reduction of CO2 in aqueous and organic
be addressed (eqn (21)–(26)). The first step is to calculate the media. Product selectivity is low however, with the complexes
standard free energy change for the reaction under consideration. producing CO and HCO2 as CO2 reduction products, and H2
This is given by as a side product. The most studied cobalt macrocycle is
X X CoIIL62+, which can be isolated in its N-rac and N-meso
DGrxn ¼ Gi  Gj ð21Þ isomeric forms (Chart 2), each having different spectroscopic
products;i reactants;j
characteristics. Equilibration of the two isomers is slow at
where Gi is the calculated absolute free energy of species i with room temperature (o2  107 s1) in organic solvents and in
the appropriate standard state (1 atm gas, 1 M in solution, or acidic aqueous solutions; equilibration is, however, rapid in
pure liquid). A species i may be a gas-phase electron (in the case alkaline media, where the N-rac isomer is favored at
of a reduction half-reaction), a solvated proton (in the cases of a equilibrium.33 While infrared (IR) spectra of the perchlorate
proton-coupled reduction half-reaction or an acid dissociation salts provide a reliable means of distinguishing the two
reaction), or any other species involved in the reaction. Once the diastereomers in the solid state, the intensity of the solvent-
DGrxn has been determined, the equilibrium constant for a dependent d–d band in the near-IR (1400–1700 nm) region
chemical reaction, provides a probe of the isomeric composition in media such as
Nujol, DMSO, D2O/H2O, or MeCN. The near-IR band is two
Keq ¼ expðDGrxn =RTÞ ð22Þ to three times more intense for the N-rac than for the N-meso
isomer. It is possible to observe isomeric mixtures of the
the pKa value for an acid dissociation reaction,
N-rac- and N-meso-CoIL6+ complexes using 1H NMR
pKa ¼ DGa =½RT lnð10Þ ð23Þ spectroscopy in CD3CN,17 and at room temperature it was
found that the mixture was 85 : 15 (N-rac : N-meso). Similar
or an absolute standard reduction potential, can be computed, ratios were observed for other isomeric ligands such as
o
Ered;abs ¼ DGred =nF ð24Þ CoIL4+17 in CD3CN and NiIIL62+ in D2O.34,35 The N-rac
isomer also has positional isomers with both five- or six-
where n is the number of electrons transferred, and F is the coordinate geometry as shown in Chart 2. The isomerisation of
Faraday constant. N-meso-CoIIL62+ to the N-rac isomer is slow but once reduced

Chart 2 The two stereoisomers of CoIIL62+ and three isomers of (Sol)CoL6(CO2)+, Sol = solvent.

2042 Chem. Soc. Rev., 2012, 41, 2036–2051 This journal is c The Royal Society of Chemistry 2012
View Article Online

to the CoI complex, the isomerisation is rapid. In pulse radiolysis The binding of CO2 via both its electrophilic centre (carbon
investigations of the thermodynamics and kinetics of CO2 binding, atom) and its nucleophilic centres (oxygen atoms) is thus
Published on 14 December 2011. Downloaded by INDIAN INSTITUTE OF TECHNOLOGY BOMBAY on 10/10/2016 10:04:09.

pure isomers of CoIIL62+ were used, however, in many other implicated in the stabilisation of the CO2 complex at low
studies the equilibrium mixtures of CoIIL62+ were utilised. temperatures. Also, a dramatic nCO2 shift from 1710 cm1 to
1544 cm1 upon cooling is consistent with a significant charge
3.1. Binding of CO2 transfer from the CoI centre to the bound CO2. While both
Gangi and Durand were among the first to report KCO2 for the five- and six-coordinate CO2 adducts are diamagnetic, six-
reversible binding of CO2 to CoIL6+ to form CoL6(CO2)+ coordinate CoI complexes are normally high-spin, paramagnetic
using an electrochemical method as discussed in Section 2.1.12 species. XANES measurements of these CO2 adducts in MeCN
They determined KCO2 to be 7  104 M1 in dry DMSO were used to identify the oxidation state and coordination
solution with 0.1 M Et4NClO4 as the supporting electrolyte. geometry of the complex. The cobalt main edge energy in the
Since the KCO2 is relatively large, Fujita, Creutz and coworkers XANES of the CO2 adduct at room temperature is similar to that
characterised the species in MeCN using UV-Vis, FT-IR, of CoIIL62+, indicating that it is a CoIIL6(CO2)+ complex,
1
H NMR, XANES (X-ray absorption near edge structure), having transferred one net electron from the CoIL6+ species to
laser flash photolysis, and pulse radiolysis.17,19,20,36–39 As the bound CO2 ligand. However, the XANES of the CO2 adduct
mentioned in Section 2.2.1, the KCO2, together with other at low temperature unambiguously showed a considerable edge
thermodynamic parameters (i.e., DHo and DSo) can also be shift toward higher energy, suggesting the formation of a CoIII
determined spectroscopically using eqn (14), or by heating the carboxylate, (MeCN)CoIIIL6(CO22)+, by transferring two
solution containing the species and a known amount of CO2 in electrons from CoI to the bound CO2. Metal carboxylates
a closed vessel. However, when a purple solution of have been postulated to be important intermediates in photo- and
CoL6(CO2)+ is cooled, the solution becomes brownish yellow electrochemical CO2 reduction and the water gas shift reaction,
and freezes as a yellow solid. The purple colour returns when but this study clearly showed that cobalt(I) macrocycles can
the solution is warmed to room temperature. This thermo- transfer two electrons to the bound CO2, and thereby facilitate
chromism, which is due to the addition of a solvent molecule CO2 reduction, avoiding an energetically unfavorable one-electron
to five-coordinate CoL6(CO2)+ as shown in eqn (27), was pathway.
studied in MeCN, butyronitrile (PrCN, Fig. 5), and a CD3CN/ This was a very important result, but it was hard to under-
THF mixture. stand how a closed-shell species could transfer first one and
then a second electron from the metal centre to a bound CO2
CoL6(CO2)+ + Sol " (Sol)CoL6(CO2)+, KSol (27) molecule. The electron donation in CoL6(CO2)+ was investigated
with B3LYP/6-31+G(d,p) calculations in a CPCM treatment of
The FT-IR spectra measured over the range 25 to 75 1C in an MeCN solution with and without an explicit solvent molecule
a CD3CN/THF mixture indicate the existence of the six- coordinated in the sixth position. The results explain the strange
coordinate species with intramolecular hydrogen bonds apparent sequential one-electron transfer. The calculated
between the bound CO2 and the amine hydrogens of the TD-DFT UV-Vis spectra of the five- and six-coordinate CO2
macrocyclic ligand (nCO2 = 1544 cml, nNH = 3145 cm1). adducts are shown in Fig. 5. The calculated spectrum of the

Fig. 5 (left) Temperature-dependent spectrum of CoL6(CO2)+ in PrCN. The peak at 530 nm diminishes, while the peak at ca. 430 nm increases
upon cooling. The spectra were taken at 40.0, 23.8, 14.8, 0.0, 9.0, 20.0, 30.5, 40.8, 51.5, 60.2, 73.6, 84.5, and 110.5 1C. Inset:
Relationship between 1000/T and ln(KSol) that gives DHo = 13.8  3 kcal mol1, DSo = 30  3 cal K1 mol1. The figure has been reprinted
with permission from ref. 17. Copyright 1991 American Chemical Society; (right) Calculated spectra for CoL6(CO2)+ (red) and (MeCN)CoL6-
(CO2)+ (blue).

This journal is c The Royal Society of Chemistry 2012 Chem. Soc. Rev., 2012, 41, 2036–2051 2043
View Article Online

In this manner, i.e., the sharing of two pairs of electrons


originally associated with the reduced metal centre between
Published on 14 December 2011. Downloaded by INDIAN INSTITUTE OF TECHNOLOGY BOMBAY on 10/10/2016 10:04:09.

the metal centre and the C atom of the CO2 ligand, we can
explain the apparent sequential one-electron transfer from CoI to
CO2 in the closed-shell singlet state upon first the binding of the
CO2 and then the coordination of a MeCN solvent molecule.
The calculated standard binding free energy of CO2 to CoIL6+
to form the five-coordinate adduct is 5.70 kcal mol1, yielding
a value for KCO2 of 1.5  104 atm1 (or 5.4  104 M1), which is
between the values obtained by the electrochemical and spectro-
scopic methods discussed above.
Let us now discuss more details about CO2 binding constants
to CoIL6+. Schmidt et al.16 extended the electrochemical KCO2
measurements in various solvents using pure N-rac-CoIIL62+
and Creutz et al.36 determined KCO2 values of N-rac and
N-meso-CoIL6+ in water using pulse radiolysis (Table 1).
The KCO2 value of the N-rac-CoIIL62+ isomer is 100 times
larger than that of the N-meso-CoIIL62+ isomer in both
DMSO and H2O. A weak correlation is found between KCO2
of the N-rac-CoIIL62+ isomer and the solvent dielectric
constant,16 and may be attributed to the ‘solvation’ of CO2
in each solvent tried.19 Furthermore, the values in water for
both N-rac and N-meso-CoIIL62+ are 1000 times larger than
those in MeCN or DMSO, indicating efficient stabilisation by
a network of hydrogen bonding between the bound CO2
molecule and free water molecules. The UV-Vis spectrum of
Fig. 6 Calculated doubly-occupied highest occupied molecular orbital N-rac-CoIIL62+ at room temperature also indicates that
(HOMO) of five-coordinate CoL6(CO2)+, top, and the calculated C2 five- and six-coordinate species exist with a ratio of 1 : 2,36
symmetry structure of the same view of the complex, bottom. The CO2 in contrast to almost all five-coordinate species in MeCN. It
adduct is stabilised by hydrogen bonds (distance 1.978 Å) between its
should be noted that the difference in KCO2 values for CoIL5+
two oxygen atoms and the two N–H protons of the L6 ligand. The
and CoIL6+ in MeCN, which have the same reduction
Co–C bond distance is 1.981 Å and the O–C–O bond angle is 132.431.
Co (light blue), C (gray), H (white), N (blue), O (red).
potential for CoII/I, is surprisingly large. These results indicate
that effective hydrogen-bonding interactions between the
bound CO2 and amine macrocycle N–H protons may serve
former shows a metal-to-ligand charge transfer (MLCT) transition to additionally stabilise the adduct in some cases, while steric
at ca. 480 nm in reasonable agreement with the experimentally repulsion by methyl groups on the tetradecane ring may
observed transition at ca. 530 nm. The calculated spectrum for destabilise the adducts, depending upon the complex.
the six-coordinate CO2 adduct exhibits a weaker transition Electrochemically, CoIIL22+ and CoIIL32+ show no change
at ca. 440 nm in good agreement with the experimentally when the solution is saturated with CO2, indicating that either
observed peak at ca. 430 nm. The observed change in the the KCO2 o 0.5 M1 or the forward reaction rate of the
electronic structure upon coordination by a solvent molecule is complex with CO2 is very small so no chemical reaction takes
thus reflected by the calculations. place. The complexes CoIIL42+, CoIIL52+, and N-meso-
As shown in Fig. 6, the electron pair in the highest occupied CoIIL62+ exhibit behaviour consistent with an electron transfer
molecular orbital (HOMO) of the five-coordinate CO2 adduct followed by a chemical reaction, or an ErCr mechanism.41
to CoIL6+, originally in the dz2 orbital of the metal centre, is Under CO2, the reversibility of CoII/I couples for CoIIL42+,
shared to make a s-bond between the metal centre and the CoIIL52+, N-meso-CoIIL62+, and CoIIL72+ is maintained but
C atom of the CO2 adduct. Since this is a shared pair of electrons is shifted to more positive potential. In the case of N-rac-
between the Co and the C atom, it represents the transfer of one CoIIL62+, CoIIL82+, and CoIIL92+, their CVs are irreversible.
net electron, in support of the assignment of CoII in the XANES This is a result of sufficiently rapid electron transfer and
experiments. If CO2 remains bound to the catalyst, then one forward reactions, and a slow reverse reaction with large
electron processes are not necessarily high in energy, especially KCO2. However, the cathodic peak current does not change,
when they correspond to the movement of a pair of electrons. therefore a catalytic reaction can be ruled out. Interestingly,
In the six-coordinate complex (Fig. 7), this s-bonding CoIL6+ was the only complex found to have thermochromic
orbital between the Co and the C atom of CO2 becomes behaviour (i.e., purple at room temperature and yellow at low
HOMO-3, and the new HOMO shows the donation of a second temperature), but the UV-Vis spectra of CoL7(CO2)+,
(dxz) pair of electrons from the metal centre, this time to make a CoL8(CO2)+ and CoL9(CO2)+ in MeCN indicate that these
p-bond between the Co and the C. This ‘‘carbene-like’’ bonding species are likely solvated six-coordinate CoIII carboxylate
transfers a second net electron from the Co to the C, resulting in species at room temperature, i.e., (MeCN)CoIIIL(CO22)+.
the observed CoIII assignment in the XANES experiments. Binding constants increase as the CoII/I reduction potentials

2044 Chem. Soc. Rev., 2012, 41, 2036–2051 This journal is c The Royal Society of Chemistry 2012
Published on 14 December 2011. Downloaded by INDIAN INSTITUTE OF TECHNOLOGY BOMBAY on 10/10/2016 10:04:09. View Article Online

Fig. 7 Calculated orbitals binding a CO2 adduct to six-coordinate (MeCN)CoL6(CO2)+: (left) the doubly-occupied HOMO-3, top, and the
calculated C2 symmetry structure of the same view of the complex, bottom, showing formation of a s-bond between the Co centre and the C atom
of the CO2 adduct analogous to that shown in Fig. 6; and (right) the doubly-occupied HOMO, top, and the corresponding view of the structure,
bottom, showing formation of a p-bond between those two centres to form a ‘‘carbene-like’’ species. The two O–H bond distances decrease to
1.810 Å from 1.978 Å, the Co–C bond distance increases slightly to 2.003 Å from 1.981 Å, and the O–C–O bond angle decreases to 127.371 from
132.431 upon coordination by the solvent. Co (light blue), C (gray), H (white), N (blue), O (red).

Table 1 CO2 binding constants and CO2 binding rate constants for cobalt complexes measured at 25 1C, unless otherwise noted

Ligand Solvent E1/2 vs. NHEa/V KCO2 (elec)/M1 KCO2 (spec)/M1 kCO2/M1 s1 Ref.
L2 MeCN 0.10 o0.5 17
L2 DMSO 0.07 o4 16
L3 MeCN 0.65 o0.5 17
L3 DMSO 0.62 o4 16
L4 MeCN 1.04 1.7  0.5 4.0  1.3 17
L4 DMSO 1.01 75 16
L4 MeOH/MeCN 7 1.1  106 20
L5 MeCN 1.10 25  6 26  8 17
N-meso-L6 MeCN 1.10 165  15 17
N-meso-L6 DMSO 1.05 (2.6  0.5)  102 16
N-meso-L6 H2O 6.0  106c 1.6  107 36
N-rac-L6 MeCN 1.10 (6  2)  104 (1.2  0.5)  104 17
N-rac-L6 DMSO 1.05 (3.0  0.7)  104 16
N-rac-L6 DMF 1.06 (1.8  0.6)  104 16
N-rac-L6 THF 1.00 (3.0  0.8)  103 16
N-rac-L6 Propylene carbonate 0.96 (4.0  0.7)  104 16
N-rac-L6 MeOH/MeCN 1.1  104 1.7  108 20
N-rac-L6 (primary) H2O 4.5  108c 1.7  108 36
L6b DMSO 1.05 7  104 12
L7 MeCN 1.17 (9  3)  104 17
L7 DMSO 1.11 (1.0  0.3)  105 16
L8 MeOH/MeCN >5  104 3.7  108 20
L8 MeCN 1.27 (7  3)  105 17
L9 MeCN 1.41 (3  2)  106 17
L9 DMSO 1.31 Irreversible 16
L11 DMSO 0.73 o4 16
a
Measured in the absence of CO2; converted to vs. NHE from vs. SCE (+0.24 V); converted to vs. SCE from vs. ferrocenium/ferrocene when
necessary, using formal potentials from ref. 40 in a specific solvent and electrolyte. b N-meso and N-rac mixtures. c Calculated from kf and kr in H2O.

become more negative. Thus charge transfer from CoI to CO2 is MeCN : MeOH (5 : 1 v/v) or by pulse radiolysis in water,
an important factor in stabilising these adducts. It should be noted increase as the binding constants increase (Table 1). Over a
that the binding rate constants, measured by flash photolysis in period of a few weeks, CoL6(CO2)+ and (Sol)CoIIIL6(CO22)+

This journal is c The Royal Society of Chemistry 2012 Chem. Soc. Rev., 2012, 41, 2036–2051 2045
View Article Online

decomposed to form CoL6(CO)+, CoL62+, and carbonate in one of the L6 ligands, namely N(1) and N(8), and one oxygen
dry MeCN under 1 atm CO2. However, when the solution from the ClO4 and one oxygen from the bridging CO2, namely
Published on 14 December 2011. Downloaded by INDIAN INSTITUTE OF TECHNOLOGY BOMBAY on 10/10/2016 10:04:09.

contains proton sources (such as an amine sacrificial donor, O(21) and O(1), respectively. There is also a hydrogen bond
water, or methanol), the CO2 adducts decompose rapidly to between the hydrogen of the bridging CO2H and the ClO4
form free CO. In fact, CoL6+ and CoL1+ act as catalysts for counterion, namely between O(2) and O(21). The isolation of
photochemical and electrochemical CO2 reduction.37 The such a complex has implications about the involvement of two
spectroscopic evidence for the sequential formation of the cobalt centres in the mechanism of CO2 reduction. Meanwhile
photosensitiser radical anion, CoIL6+, CoL6(CO2)+ and (Sol)CoIIIL6(CO22)+ has been postulated as a key intermediate,
(Sol)CoIIIL6(CO22)+, with the photosensitiser p-terphenyl and the protonation of the bound CO22 has been considered as
and triethylamine as the electron donor, was investigated in the next step for CO2 reduction. Nevertheless, it has been rather
catalytic CO2 reduction systems using flash photolysis in complicated to elucidate the pathway of decomposition of
MeCN : MeOH (5 : 1 v/v).20 The electron-transfer rate constant (Sol)CoIIIL6(CO22)+ or CoIIL6(CO2)+.20 The isolation of
for the reaction of the p-terphenyl radical anion with CoIIL62+ is this dimer could be simply due to the low solubility of the species.
diffusion controlled because of the large driving force (+1.1 V or
DG = 1.1 eV). The CO2 binding rate constant 1.7  l08 M1 s1
in MeCN : MeOH (5 : 1 v/v) is similar to that in H2O obtained by 3.2. Binding of CO
pulse radiolysis. As mentioned in Section 2.2.2, the equilibrium Carbon monoxide is a much better ligand than CO2, and this fact
constant 1.1  104 M1 for the binding of CO2 to CoIL6+ becomes obvious when comparing the equilibrium constants of
obtained by flash photolysis is consistent with that previously the CO adducts of the MI macrocycles discussed here. Electro-
obtained by spectroscopic methods in MeCN. Flash photolysis20 chemistry experiments showed that CO binds reversibly to
and continuous wavelength photolysis30,37 investigations of CoIL2+ and CoIL3+.17 Conversely, CoIIL2+ (L = L4–L9)
photocatalytic systems with varying CoI macrocycles identified display well separated, irreversible waves under a CO atmosphere,
the factors controlling the thermodynamics and kinetics of CO2 a process attributed to an ErCi mechanism: rapid reduction
binding, as well as the photoreduction of CO2 to CO. Steric followed by rapid CO binding, and then rapid CoI–CO oxidation
hindrance and reduction potentials are important factors in the that results in loss of CO. All these CoIL(CO)+ complexes are
catalytic activity for photochemical CO2 reduction. Further- yellow in colour. Crystals of [CoIL6(CO)]ClO4 were isolated and
more, temperature-, solvent-, and CoII/I potential-dependent the structure was solved as shown in Fig. 9.
conversion from CoIL(CO2)+ to (Sol)CoIIIL(CO22)+ via two- A large KCO and a low nCO are consistent with strong MI to
electron transfer to the bound CO2 could be of fundamental CO back donation; nCO decreases and KCO increases as CoI
importance for the proton-coupled reduction of CO2. becomes a more powerful reductant (Table 2), i.e., CoII/I has a
Through Na(Hg) reduction of CoIL6+ in MeCN, followed more negative reduction potential.17 The CoI–macrocycle
by introduction of CO2, a dinuclear, CO2H–bridged adduct, affinity for CO exceeds its affinity for CO2 in MeCN, suggesting
Co–C(OH)–O–Co, was isolated and its structure was solved that reduction of bound CO2 to bound CO is thermodynamically
(Fig. 8).19,42 The crystal structure in Fig. 8 provided evidence favourable compared to the reduction of the free substrates.
for H-bonding stabilisation between one of the CoL6 moieties, However, in water, KCO2 and KCO are almost the same, indicating
a perchlorate counter anion, and the bridging CO2H. More water may be a better solvent for CO2 reduction, its low solubility
specifically, H-bonds exist between the amine hydrogens of notwithstanding. Backbonding with CO and charge transfer to
CO2, as well as CO2 stabilisation by N–H bonds of the ligand
(Fig. 6–9) are all important factors governing CO and CO2
binding properties of CoIL macrocycles, though these are further

Fig. 9 The solid state structure of the CoIL6(CO)+ cation, hydrogen


2+
Fig. 8 ORTEP drawing of the dication, [(CoL6)2(CO2H)(ClO4)] , atoms omitted for clarity. O1, C1, and Co lie on a crystallographic
isolated from a MeCN/diethyl ether solution of [CoIL6](ClO4) satu- 2-fold axis, which relates N1 to N8, N4 to N11, etc. The hydrogens
rated with CO2. Ellipsoids drawn at the 50% probability level, (not shown) on N1 and N8 are both on the same side of the ligand,
H-atoms omitted for clarity. The dashed lines indicate H-bonds that facing the carbon monoxide. The Co–C distance is 1.797 (10) Å. The
are thought to stablise the formation and isolation of the dinuclear figure has been reprinted with permission from ref. 43. Copyright 1989
complex.42 American Chemical Society.

2046 Chem. Soc. Rev., 2012, 41, 2036–2051 This journal is c The Royal Society of Chemistry 2012
View Article Online

Table 2 CO binding constants and stretching frequencies (nCO) for and later by Sauvage et al.45 Sauvage et al. discovered the
cobalt complexes measured in MeCN at 25 1C, unless otherwise noted. selectivity and efficiency of NiIIL12+ as an electrocatalyst to
Published on 14 December 2011. Downloaded by INDIAN INSTITUTE OF TECHNOLOGY BOMBAY on 10/10/2016 10:04:09.

Data taken from ref. 17


produce CO (as opposed to H2) in aqueous solutions. NiIL1+ is
Ligand KCO/M1 nCO/cm1 adsorbed on the surface of a Hg working electrode,46 a process
that occurs at potentials more positive than the NiII/I observed
L2 5  104 2007
L3 1.4  105 1959 in the absence of CO2, a truly remarkable phenomenon that
L4 1.1  108 1912 suggests a strong chemical interaction between the molecular
L5 1.9  108 1918 complex and the electrode surface.47,48 However, the process is
N-rac-L6 2.3  108 1916
L7 Z 3  108 1910 not so trivial, and catalytic activity is affected by the structure of
L8 Z 3  108 1915 the cylcam ligand,49 as well as its corresponding complex
L9 Z 3  108 1912, 1895 isomers,47,49 e.g., the conformational isomers of NiIIL12+ in
Chart 3. To date, NiIIL12+ is still the benchmark for other
complicated by the existence of conformational, positional, nickel ‘cyclam-like’ catalysts, and since Sauvage’s initial reports,
and coordination isomers. there have been only two other complexes that exhibit greater
electrocatalytic activity for the selective reduction of CO2 to CO
3.3. Binding of H+ under similar conditions, namely NiIIL132+ and NiIIL152+,
reported by Fujita et al.50
Cobalt hydride species with cyclam-related macrocycles are
difficult to isolate due to their facile reactivity toward oxygen,
4.1. Binding of CO2
H+, etc. A pulse radiolysis study in H2O by Creutz et al.
examined the equilibration and kinetics of protonation of Typically the value of KCO2 for Ni ‘cyclam-like’ complexes is
N-rac and N-meso isomers of CoIL6+ by acid (HA).22,36 The quite low, on the order of 10 M1 (Table 3). Kelly et al. have
kinetics and thermodynamics for formation of CoL6(H)2+ carried out a number of experimental studies with NiIL1+ in
can be determined using two kinds of reaction pathways with aqueous solution.51 It was found through pulse radiolysis
the pulse radiolysis technique (eqn (28), or eqn (29) and (30)). studies that the yield of NiIL1+ (lmax abs = 380 nm in UV-Vis
spectrum) was dependent on [CO2], with a non-visible-light-
CoIIL62+ +  H - CoL6(H)2+ (28)
absorbing NiL1(CO2)+ adduct being formed via the equili-
CoIIL62+ + eaq - CoIL6+ (29) brium in eqn (35). Some of the binding parameters were
determined to be KCO2 = 16 M1, kf = 3.2  107 M1 s1,
CoIL6+ + HA - CoL6(H)2+ + A (30) and kr = 2.0  106 s1 with a standard binding free energy
Three protonated cobalt(I) species, sec-N-rac-CoL(H)2+, (DGoaq) for eqn (35) of 1.6 kcal mol1.51 Whereas for
N-meso-CoL(H)2+, and primary-N-rac-CoL(H)2+ (whose CoL6(CO2)+ there was evidence of charge transfer from CoI
structures are similar to the CO2 analogues upon replacement to CO2, this was not the case for NiL1(CO2)+, i.e., there was
of CO2 by H+ in Chart 2), were characterised in a pulse no clear evidence of a NiII or NiIII metal centre. A later study
radiolysis study using the following reactions: revealed that the yield of NiIL1+ decreased by a second-order
decay over a range of [CO2], and this led to the conclusion that
N-rac-CoIIL62+ +  H - sec-N-rac-CoL(H)2+ (31) in order for the reduction of CO2 to take place, NiL1(CO2)+
had to receive a second electron from another NiIL1+ species.52
N-meso-CoIIL62+ +  H - N-meso-CoL(H)2+ (32)
The rate constant for this electron transfer was determined to be
N-rac-Co L6 I +
+ HA - primary-N-rac-CoL(H) 2+
+ A 1.6  108 M1 s1 in aqueous solution.52 As with CoL6(CO2)+,
(33) the binding of CO2 does not proceed by CO2 insertion into a
Ni–H bond. Although both Gangi and Fujita attempted to
N-meso-CoIL6+ + HA - N-meso-CoL(H)2+ + A (34) measure KCO2 for NiIL6+, the data could not be obtained by
While hydrogen atom addition to both isomers of CoIIL62+
is fast (B109 M1 s1), the rate constants (105–108 M1 s1)
for the reaction of N-rac-CoIL6+ with proton donors decrease
with increasing pKa of the donor acid, consistent with a
reaction occurring via proton transfer.44 The pKa values for
N-meso-CoL(H)2+ and primary-N-rac-CoL(H)2+ are reported
to be Z 13.9 and 11.4, respectively.36 Interestingly, no insertion
of CO2 into the Co–H bond was detected. Rather it was
concluded that equilibration among CO2 and hydride complexes
of a given isomer proceeds exclusively via formation of the free
CoI complex as shown in Scheme 2.

4. Nickel tetraazacyclam macrocycles


Nickel ‘cyclam-like’ complexes were studied as CO2 reduction
catalysts over three decades ago, first by Fisher and Eisenberg32 Chart 3 Conformational isomers of NiIIL12+.

This journal is c The Royal Society of Chemistry 2012 Chem. Soc. Rev., 2012, 41, 2036–2051 2047
View Article Online

Table 3 NiII/I reduction potentials and CO2 binding constants for


nickel complexes measured at 25 1C, unless otherwise noted
Published on 14 December 2011. Downloaded by INDIAN INSTITUTE OF TECHNOLOGY BOMBAY on 10/10/2016 10:04:09.

Ligand Solvent E1/2 vs. NHEa/V KCO2 (elec)/M1 Ref.


L1 MeCN 1.14 42 50
L1 H2O 1.18 16 50, 51
L1 DMSO 1.20 Irreversible 16
L2 MeCN 0.26 59
L2 DMF 0.53 58
L3 MeCN 0.58 59
L5 MeCN 0.99 59
L5 DMF 1.25 58 Fig. 10 Side view (left) and top view (right) of the RSRS-
L6b MeCN 0.98 59 NiL1(CO2)+ adduct in CPCM of water solvent; Ni (yellow), C
L6b DMF 1.24 58 (orange), H (white), N (cyan), O (red).
L10 DMSO 1.00 o4 16
DMF 1.30 58
L11 DMF 1.08 58 between the Ni and C of CO2; the electron in the singly occupied
L12 MeCN 1.08 This work molecular orbital also contributes to this s-bond. This is in contrast
L13 MeCN 1.17 42 50
L14 MeCN 1.19 42 50
to the behaviour modeled for CoL6(CO2)+ in Section 3.1.
L15 MeCN 0.60 61
a
Measured in the absence of CO2; converted to vs. NHE from vs. SCE 4.2. Binding of CO
(+0.24 V); converted to vs. SCE from vs. ferrocenium/ferrocene when
necessary, using formal potentials from ref. 40 in a specific solvent and Nickel ‘cyclam-like’ complexes react with CO to give paramagnetic
electrolyte. b N-meso and N-rac mixtures. d9 five-coordinate species with square-pyramidal geometry, as was
evidenced by electron paramagnetic resonance (EPR)56–58 and
extended X-ray absorption fine structure (EXAFS)59 spectro-
electrochemical methods due to the low binding affinity of
scopies. EPR studies by Gagné and Ingle showed that for NiII
CO2 with NiIL6+.
complexes with a-conjugated macrocyclic ligands (e.g., L2), a
kf ligand radical was formed upon one electron reduction of
NiI L1þ þ CO2 Ð NiL1ðCO2 Þþ ð35Þ the complex, giving NiII(L )+, with NiII playing a role in
kr
stabilisation.58 In contrast, complexes of ligands without
Sakaki investigated the likelihood of CO2 binding to a NiI a-conjugation (e.g., L11) showed reduction at the NiII centre
complex using ab initio calculations.53,54 Modeling NiIIL12+ to give NiIL+. In either case, NiII(L )+ or NiIL+, both
with NiII(NH3)4F2, he attempted to observe different species react with CO to give NiIL(CO)+.57–59 While in its
‘NiL(CO2)+’ intermediates, where the CO2 was C-bound in reduced state, NiII(L )+ gave isotropic spectra consistent
an Z1 fashion. He found that while the complex NiI(NH3)4(F)0 with a ligand radical species, CO addition resulted in the
could coordinate to CO2, complexes of NiI(NH3)4+, conversion to an anisotropic species. The conversion was
NiI(NH3)5+, or NiII(NH3)4(F)+ could not coordinate CO2. described as being an intramolecular electron transfer process,
It was said that for NiI(NH3)4(F)(CO2)0, the F pushes up by which the electron migrated from a ligand orbital to a
dz2 and neutralises the charge of NiI, thus allowing for strong predominately metal orbital.58 Results from Gagné and Fujita
NiI to CO2 charge transfer and stabilisation. In addition, the suggested that CO predominately coordinates Ni by a
F coordination results in an overall neutral charge of the s-interaction with the metal 4s and 4p orbitals,58,59 an assign-
complex, which decreases charge–dipole repulsion between ment based on the fact that nCO was independent of KCO for
the nickel complex and CO2. Furthermore, the charge transfer the complexes reported.
character increases the negative charge on the oxygen atom(s) Fig. 11 illustrates the determination of KCO for NiIIL122+ in
of CO2, facilitating protonation. The analogous NiI(NH3)4- MeCN using the method described in Section 2.1. In the left
(CO2)+ structure was also proposed, but there is no charge diagram, the CVs show that as the [CO] increases while [Ni] is
transfer to CO2 in the absence of F. Sakaki also modeled kept constant, there is a shift in the E1/2red to more positive
NiII(NH3)4(F)(CO2)+ as a CO2 adduct, however this complex potentials for the NiII/I couple. The inset shows that the shift
was thought to be unlikely due to the overall NiII charge in potentials is reversible, i.e., purging the 10% CO/Ar
causing repulsion of CO2.53 solution with pure Ar results in the original CV for the
With four asymmetric N-donors, cyclam (L1) has five NiII/IL122+/+ couple. The nature of the difference in the waves
possible conformational isomers when coordinated to a metal with different [CO] could be due to isomers of NiIL12(CO)+,
centre (Chart 3).55 Calculations done for this review showed or changes in electron transfer and forward and backward
that CO2 binds to RSRS-NiIL1+ in water because of the four reaction kinetics of the complex with CO (eqn (8)–(10)). The
N–H protons on the same side of the ligand (Fig. 10), right diagram shows a plot of the E1/2red vs. ln[CO] at two
permitting H-bonding stabilisation of the CO2 adduct. A different scan rates, 20 and 50 mV s1, and KCO has a value of
molecule of the solvent water does not coordinate with the 1.3  105 M1, obtained from the least-squares line-fit. The CO
metal centre of the CO2 adduct, even when two explicit solvent binding constants and stretching frequencies are summarized in
water molecules are hydrogen bonded to the CO2 in the Table 4.
adduct. Analysis of frontier orbitals shows that a doubly- Mulazzani, Blinn, and coworkers studied the CO binding
occupied molecular orbital shares its electron pair in a s-bond properties of NiIIL12+ in aqueous solution using pulse radiolysis

2048 Chem. Soc. Rev., 2012, 41, 2036–2051 This journal is c The Royal Society of Chemistry 2012
Published on 14 December 2011. Downloaded by INDIAN INSTITUTE OF TECHNOLOGY BOMBAY on 10/10/2016 10:04:09. View Article Online

Fig. 11 (left) CVs of the NiII/IL122+/+ reduction couple in the absence and presence of increasing [CO] in MeCN; inset: regeneration of NiII/I
under Ar. (right) A plot of DE/(RT/F) vs. ln[CO], the exponential of the intercept obtained from the linear fit corresponds to KCO.

Table 4 CO binding constants and stretching frequencies (nCO) for 350 nm is quite stable in the presence of O2 or H+. An initial
NiIL(CO)+ adducts, measured at 25 1C, unless otherwise noted CO addition product of NiIL1+ at 470 nm decays slowly to a
Ligand Solvent KCO (elec)/M1 nCO/cm1 Ref. product at 350 nm (1.8 s1). The rather negative DSz (14.4 
5
0.1 cal K1 mol1) and positive DHz (12.7  0.1 kcal mol1)
L1 MeCN (2.8  0.6)  10 1955 59 suggest significant solvation and structural reorganisation
L1 H2O 7.5  105 48
L2 MeCN (1.3  0.3)  102 2012 59 within the transition state. In other words, the species at
L2 DMF (1.7  0.4)  102 58 350 nm is not a new CO-adduct, but merely the thermo-
L3 MeCN (3.7  0.8)  102 1977 59 dynamically favored geometric isomer for the five-coordinate
L5 MeCN (1.8  0.4)  104 1961 59
L5 DMF (1.8  0.2)  104 58 NiI–CO adduct.
L6 MeCN (5.6  1.5)  104 1962 59 There is a difference when comparing NiIL6(CO)+ to
L6 DMF (4.7  0.5)  104 58 NiIL1(CO)+, as the former is ‘locked’ in its kinetic isomer
L10 MeCN (4.0  1.5)  104 1956 59 while the latter is free to isomerise to the more thermo-
L10 DMF 7.8  104 58
L11 DMF (4.5  0.5)  104 58 dynamically stable 350 nm-absorbing isomer. The NiIL6(CO)+
L12 MeCN 1.3  105 This work has an absorption at ca. 470 nm but does not isomerise further.
L13 DMF (9.0  2.0)  104 1939 61 The shift in absorption maximum from 470 to 350 nm in
L14 MeCN (1.8  0.4)  105 1956 61
L15 MeCN (1.2  0.4)  105 1967 61
NiIL1+ could be due to isomerisation of L1, followed by
L15 H2O 1.1  106 62 coordination of CO. The kinetic isomer (470 nm) is presumed
to be in the RRSS-configuration (Chart 3), while the
thermodynamic isomer (350 nm) is thought to be in the
and laser flash photolysis techniques.60 The NiIL1+ species RSRS-configuration (Chart 3). This observation is in support
absorbs at 380 nm in the UV-Vis spectrum, and the rate of of the calculations that show that CO2 will only bind to the
decay of this species was found to increase with increasing RSRS-configuration of NiIL1+ (Fig. 10).
[CO] in the range of 0.1–1.0 mM. The decay is followed by the Other methods have been used to characterise the NiIL(CO)+
growth of a species having a maximum at 470 nm. The species adducts. Fig. 12 exemplifies reduction of NiIIL132+ by Na(Hg)
at 470 nm decays by a first-order process with the concomitant in MeCN solution as characterised by UV-Vis spectroscopy,
formation of a new species at 350 nm with weak absorption ultimately forming NiIL13+. When CO is introduced into the
at 630 nm, and is not dependent on [CO]. The new species at bulk solution containing NiI(L13)+, a CO adduct is observed.

Fig. 12 UV-Vis spectra of (left) the systematic reduction of NiIIL132+ (black line) to NiIL13+ (red line) by Na(Hg) in MeCN; and of (right)
NiIL13+ (black) and NiIL13(CO)+ (red).

This journal is c The Royal Society of Chemistry 2012 Chem. Soc. Rev., 2012, 41, 2036–2051 2049
View Article Online

XANES and EXAFS are useful techniques to establish [(L1)Ni–H–Ni(L1)]3+ intermediate that undergoes simple
unambiguously the ligation geometry around Ni. The absorption protonation/decomposition to give H2 and NiIIL12+.
Published on 14 December 2011. Downloaded by INDIAN INSTITUTE OF TECHNOLOGY BOMBAY on 10/10/2016 10:04:09.

intensity correlates directly with the coordination number when Ni


is four-, five-, or six-coordinate. A study of NiII, NiI, and NiI–CO
5. Comparison of Co and Ni macrocycles
complexes of L2, L5, and L6 by Fujita and coworkers examined
such coordination characteristics.59 They found that for NiIIL22+ In general terms, it is true that complexes of CoIL+ form more
the coordination number was solvent dependent—six-coordinate stable adducts with CO2 and H+ than do the NiIL+ analogues.
in MeCN and four-coordinate in nitromethane. Upon reduction In fact, the standard binding free energies of CoIL6+ and
of NiIIL22+ the ligand radical species was four-coordinate, NiIL1+ for binding with CO2 and H+ differ by a factor of
and when CO was introduced, a complex mixture developed ten.51 While they cannot be compared directly because of two
that included the NiII(L )+ monomer and dimer, as well as different ligands, the charge transfer stabilisation observed in
NiIL2(CO)+. Complexes NiIIL52+ and NiIIL62+ were four- the CoL(S)+ (S = substrate) complexes seems to be absent in
coordinate square planar, as were their NiIL5+ and NiIL6+ the analogous nickel complexes. XANES data unambiguously
reduced complexes. When CO was introduced, a five-coordinate show evidence for charge transfer in CoIL(CO2)+ complexes, to
species was characterised, with the Ni–C distances in NiIL1(CO)+ give CoIIL(CO2)+ and, upon cooling, (Sol)CoIIIL(CO22)+
and NiIL2(CO)+ being short, ca. 1.8 Å. (Sol = solvent), one- and two-electron transfer products from
CoI to bound CO2, respectively. This experimental evidence is
4.3. Binding of H+ further supported through DFT calculations which showed s- and
It should be noted that in general, the nickel ‘cyclam-like’ p-interactions between Co and CO2 in the HOMOs of the one- and
complexes are not efficient catalysts for the reduction of H+ to H2, two-electron transfer products, respectively. On the other hand,
even in the absence of dissolved CO2. Thus, it is thought that a NiIL(CO2)+ complexes do not seem to enter into the same sort of
CO2 mechanism may not proceed via a Ni–H intermediate. In any charge transfer interaction of NiI bound to CO2. Rather, only a
case, the binding of H+ to NiI may be determined experimentally. s-bond interaction is calculated to be present in the Ni–CO2
The kinetics and thermodynamics of H+ binding to NiIL1+ in adduct. The difference in the stability between the two types of
aqueous solution was studied by the pulse radiolysis technique.51 It metal macrocycles, cobalt and nickel, has been attributed to
was found that the reduction of NiIIL12+ by  H goes through an differences in the MIII/II couple, as opposed to the MII/I couple,51
inner-sphere mechanism to give NiL1(H)2+, with a rate constant for which the potentials differ significantly between the two metals.
on the order of 109 M1 s1. The transient ‘hydride’ species What has also become clear is that geometric isomers play a
decomposes to NiIL1+ and H+ at a much slower rate, with crucial role in substrate binding for both cobalt and nickel
kf = 3  107 M1 s1 and kr = 5  105 s1. The binding free macrocycles. Although not discussed in great detail, it turns
energy for eqn (36) was determined to be 2.4 kcal mol1. out that different configurations of the macrocycle ligand
Interestingly, there is a pH dependent yield of NiIL1+, where the affect catalytic activity as well.50 Even with the simplest
absorbance at 380 nm in the UV-Vis spectrum corresponding to NiIL1+ complex, the RSRS-configuration is the active isomer,
NiIL1+ decreases with decreasing pH. The data give an isotherm presenting a certain non-trivial aspect to these sort of ‘cyclam-like’
suggesting an equilibrium reaction as shown in eqn (36), and the macrocyclic complexes, an aspect that is not immediately apparent
acid dissociation equilibrium constant for NiL1(H)2+ was deter- when one looks at a two-dimensional drawing of the complexes.
mined to be 0.017 M (pKa 1.8).51 When applied to different Ni Finally, what remains interesting (and not well understood) is
‘cyclam-like’ derivatives, similar isotherms are obtained. Pulse that the nickel ‘cyclam-like’ complexes are better, more selective
radiolysis methods result in a hydrated electron being converted CO2 reduction catalysts in aqueous solution. They operate at a
to  H at low pH and therefore an inner-sphere mechanism results wide pH range and maintain selectivity, producing CO as opposed
before the equilibrium reaction. However, addition of  H ap- to H2 even at low pH. This phenomenon has been attributed to the
proaches a diffusion-controlled limit and can become quite com- difference in KS between the two substrates.51 For example, the
plicated. NiIL1+ species only reacts with H+ at very low pH (pKa of
NiL1(H)2+ o 2) so that addition of a proton cannot compete with
kf CO2 addition in a CO2 saturated solution, thus allowing for the
NiI L1þ þ Hþ Ð NiL1ðHÞ2þ ð36Þ
kr proton-assisted reduction of CO2 to CO. Conversely, CoL(H)2+
complexes are extremely sensitive to oxygen and H+, thereby
Kelly et al. also studied H+ reduction by NiIL1+ and found making them more suited for proton reduction to H2.
that NiL1(H)2+ is reduced by another NiIL1+ to form
NiL1(H)+ and NiIIL12+.52 This evidence suggests that binding
6. Future prospects
of H+ to NiIL1+ is not sufficient to drive the two-electron
reduction of H+ to H2 and NiIIIL13+. The mechanism by There has been resurgence in the many subfields of artificial
which the second electron is provided to NiL1(H)2+ by NiIL1+ photosynthesis in recent years, and at the heart of that is the
was proposed to occur either by an outer- or an inner-sphere catalytic conversion of sunlight into a sustainable energy source
reduction, following a pathway similar to that of CO2 reduction such as solar fuels. Molecular catalysts like the cobalt and nickel
by NiIL1+. For the former mechanism it was suggested that ‘cyclam-like’ complexes that are the subject of this review will
NiL1(H)+ undergoes protonation followed by decomposition receive a great deal of attention in the coming years. They contain
to yield H2 and NiIIL12+. Proton reduction via an inner-sphere inexpensive, earth-abundant metals that can activate CO2 in
mechanism was proposed to occur through a proton-bridged electro- and photocatalytic systems.

2050 Chem. Soc. Rev., 2012, 41, 2036–2051 This journal is c The Royal Society of Chemistry 2012
View Article Online

Acknowledgements V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg,


S. Dapprich, A. D. Daniels, O. Farkas, J. B. Foresman, J. V. Ortiz,
Published on 14 December 2011. Downloaded by INDIAN INSTITUTE OF TECHNOLOGY BOMBAY on 10/10/2016 10:04:09.

We thank Dr Carol Creutz in the Chemistry Department at J. Cioslowski and D. J. Fox, Gaussian 09, Revision A.02, Gaussian, Inc.,
Wallingford, CT, 2009.
Brookhaven National Laboratory (BNL) for her careful reading
24 J. E. Bartmess, J. Phys. Chem., 1994, 98, 6420–6424.
of the manuscript and suggestions. We thank Dr David J. Szalda at 25 C. P. Kelly, C. J. Cramer and D. G. Truhler, J. Phys. Chem. B,
Baruch College, CUNY, for making the ORTEP diagram in Fig. 8. 2007, 111, 408–422.
The work at BNL is funded under contract DE-AC02-98CH10886 26 M. D. Tissandier, K. A. Cowen, W. Y. Feng, E. Gundlach,
M. H. Cohen, A. D. Earhart and J. V. Coe, J. Phys. Chem. A,
with the U.S. Department of Energy and supported by its Division 1998, 102, 7787–7794.
of Chemical Sciences, Geosciences, & Biosciences, Office of Basic 27 D. M. Camaioni and C. A. Schwerdtfeger, J. Phys. Chem. A, 2005,
Energy Sciences under its Solar Energy Utilization initiative. We 109, 10795–10797.
also thank Toyota Motor Engineering & Manufacturing of North 28 C. P. Kelly, C. J. Cramer and D. G. Truhler, J. Phys. Chem. B,
2006, 110, 16066–16081.
America, Inc., for funding for the CO2 utilization research via a 29 A. H. A. Tinnemans, T. P. M. Koster, D. H. M. W. Thewissen and
Cooperative Research and Development Agreement (CRADA). A. Mackor, Recl. Trav. Chim. Pays-Bas, 1984, 103, 288–295.
30 S. Matsuoka, K. Yamamoto, T. Ogata, M. Kusaba,
N. Nakashima, E. Fujita and S. Yanagida, J. Am. Chem. Soc.,
References 1993, 115, 601–609.
31 T. Ogata, Y. Yamamoto, Y. Wada, K. Murakoshi, M. Kusaba,
1 P. Atkins, T. Overton, J. Rourke, M. Weller and F. Armstrong, N. Nakashima, A. Ishida, S. Takamuku and S. Yanagida, J. Phys.
Inorganic Chemistry, Oxford University Press, Oxford, UK, 4th edn, Chem., 1995, 99, 11916–11922.
2006. 32 B. Fisher and R. Eisenberg, J. Am. Chem. Soc., 1980, 102, 7361–7363.
2 W. H. Koppenol and J. D. Rush, J. Phys. Chem., 1987, 91, 4429–4430. 33 D. J. Szalda, C. L. Schwarz, J. F. Endicott, E. Fujita and
3 J. N. Butler, Carbon Dioxide Equilibria and Their Applications, C. Creutz, Inorg. Chem., 1989, 28, 3214–3219.
Addison-Wesley Publishing Company, Inc., Reading, Massachusetts, 34 L. G. Warner, N. J. Rose and D. H. Busch, J. Am. Chem. Soc.,
1982. 1967, 89, 703–704.
4 E. E. Benson, C. P. Kubiak, A. J. Sathrum and J. M. Smieja, 35 L. G. Warner, N. J. Rose and D. H. Busch, J. Am. Chem. Soc.,
Chem. Soc. Rev., 2009, 38, 89–99. 1968, 90, 6938–6946.
5 T. Yui, Y. Tamaki, K. Sekizawa and O. Ishitani, Top. Curr. Chem., 36 C. Creutz, H. A. Schwarz, J. F. Wishart, E. Fujita and N. Sutin,
2011, 303, 151–184. J. Am. Chem. Soc., 1991, 113, 3361–3371.
6 J.-M. Savéant, Chem. Rev., 2008, 108, 2348–2378. 37 E. Fujita, C. Creutz, N. Sutin and B. S. Brunschwig, Inorg. Chem.,
7 A. J. Morris, G. J. Meyer and E. Fujita, Acc. Chem. Res., 2009, 42, 1993, 32, 2657–2662.
1983–1994. 38 E. Fujita, L. R. Furenlid and M. W. Renner, J. Am. Chem. Soc.,
8 M. D. Doherty, D. C. Grills, J. T. Muckerman, D. E. Polyansky 1997, 119, 4549–4550.
and E. Fujita, Coord. Chem. Rev., 2010, 254, 2472–2482. 39 E. Fujita and R. van Eldik, Inorg. Chem., 1998, 37, 360–362.
9 M. R. DuBois and D. L. DuBois, Acc. Chem. Res., 2009, 42, 1974–1982. 40 N. G. Connelly and W. E. Geiger, Chem. Rev., 1996, 96, 877–910.
10 H. Takeda and O. Ishitani, Coord. Chem. Rev., 2010, 254, 346–354. 41 R. S. Nicholson and I. Shain, Anal. Chem., 1964, 36, 706–723.
11 A. J. Bard and L. R. Faulkner, Electrochemical methods: fundamentals 42 E. Fujita and D. J. Szalda, Inorg. Chim. Acta, 2000, 297, 139–144.
and applications, John Wiley & Sons, Inc., New York, 2nd edn, 2001. 43 D. J. Szalda, E. Fujita and C. Creutz, Inorg. Chem., 1989, 28,
12 D. A. Gangi and R. R. Durand Jr., J. Chem. Soc., Chem. 1446–1450.
Commun., 1986, 697–699. 44 E. Fujita, J. F. Wishart and R. van Eldik, Inorg. Chem., 2002, 41,
13 R. Battino and H. L. Clever, Chem. Rev., 1966, 66, 395–463. 1579–1583.
14 E. Wilhelm and R. Battino, Chem. Rev., 1973, 73, 1–9. 45 M. Beley, J.-P. Collin, R. Ruppert and J.-P. Sauvage, J. Chem.
15 P. G. T. Fogg and W. Gerrard, Solubility of Gases in Liquids: A Soc., Chem. Commun., 1984, 1315–1316.
Critical Evaluation of Gas/Liquid Systems in Theory and Practice, 46 M. Beley, J.-P. Collin, R. Ruppert and J.-P. Sauvage, J. Am. Chem.
John Wiley and Sons, Chichester, New York, Brisbane, Toronto, Soc., 1986, 108, 7461–7467.
Singapore, 1991. 47 G. B. Balazs and F. C. Anson, J. Electroanal. Chem., 1992, 322,
16 M. H. Schmidt, G. M. Miskelly and N. S. Lewis, J. Am. Chem. 325–345.
Soc., 1990, 112, 3420–3426. 48 M. Fujihira, Y. Hirata and K. Suga, J. Electroanal. Chem., 1990,
17 E. Fujita, C. Creutz, N. Sutin and D. J. Szalda, J. Am. Chem. Soc., 292, 199–215.
1991, 113, 343–353. 49 K. Bujno, R. Bilewicz, L. Siegfried and T. A. Kaden, J. Electroanal.
18 M. Rudolph, S. Dautz and E.-G. Jäger, J. Am. Chem. Soc., 2000, Chem., 1998, 445, 47–53.
122, 10821–10830. 50 E. Fujita, J. Haff, R. Sanzenbacher and H. Elias, Inorg. Chem.,
19 E. Fujita, D. J. Szalda, C. Creutz and N. Sutin, J. Am. Chem. Soc., 1994, 33, 4627–4628.
1988, 110, 4870–4871. 51 C. A. Kelly, Q. G. Mulazzani, M. Venturi, E. L. Blinn and M. A.
20 T. Ogata, S. Yanagida, B. S. Brunschwig and E. Fujita, J. Am. J. Rodgers, J. Am. Chem. Soc., 1995, 117, 4911–4919.
Chem. Soc., 1995, 117, 6708–6716. 52 C. A. Kelly, E. L. Blinn, N. Camaioni, M. D’Angelantonio and
21 G. V. Buxton, C. L. Greenstock, W. P. Helman and A. B. Ross, Q. G. Mulazzani, Inorg. Chem., 1999, 38, 1579–1584.
J. Phys. Chem. Ref. Data, 1988, 17, 513–886. 53 S. Sakaki, J. Am. Chem. Soc., 1990, 112, 7813–7814.
22 C. Creutz, H. A. Schwarz, J. F. Wishart, E. Fujita and N. Sutin, 54 S. Sakaki, J. Am. Chem. Soc., 1992, 114, 2055–2062.
J. Am. Chem. Soc., 1989, 111, 1153–1154. 55 B. Bosnich, C. K. Poon and M. L. Tobe, Inorg. Chem., 1965, 4,
23 M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, 1102–1108.
J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, 56 F. V. Lovecchio, E. S. Gore and D. H. Busch, J. Am. Chem. Soc.,
G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, 1974, 96, 3109–3118.
A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, 57 R. R. Gagné and D. M. Ingle, J. Am. Chem. Soc., 1980, 102, 1444–1446.
M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. 58 R. R. Gagné and D. M. Ingle, Inorg. Chem., 1981, 20, 420–425.
Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. 59 L. R. Furenlid, M. W. Renner, D. J. Szalda and E. Fujita, J. Am.
Montgomery, Jr., J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, Chem. Soc., 1991, 113, 883–892.
E. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi, 60 C. A. Kelly, Q. G. Mulazzani, E. L. Blinn and M. A. J. Rodgers,
J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, Inorg. Chem., 1996, 35, 5122–5126.
S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, J. M. Millam, 61 D. J. Szalda, E. Fujita, R. Sanzenbacher, H. Paulus and H. Elias,
M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, Inorg. Chem., 1994, 33, 5855–5863.
J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, 62 G. B. Balazs and F. C. Anson, J. Electroanal. Chem., 1993, 361,
R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, 149–157.

This journal is c The Royal Society of Chemistry 2012 Chem. Soc. Rev., 2012, 41, 2036–2051 2051

Das könnte Ihnen auch gefallen