This is the peer reviewed version of the following article: “Sustainable knowledge-driven approaches in transition metalcatalyzed transformations” which has been published in final form at:
https://onlinelibrary.wiley.com/doi/abs/10.1002/cssc.201900914
This article may be used for non-commercial purposes in accordance with Wiley Terms and Conditions for Self-Archiving."

Sustainable knowledge-driven approaches in transition metalcatalyzed transformations
J. Sanjosé-Orduna,‡ a,b Á. L. Mudarra,‡ a,b Dr. S. Martínez de Salinas,‡ a Dr. M. H. Pérez-Temprano* a
Dedicated to Prof. Pablo Espinet on the occasion of his 70th birthday

Abstract: The sustainable synthesis of relevant scaffolds for their use in the pharmaceutical, agrochemical and material sectors
constitutes one of the most urgent challenges that the chemical community needs to overcome. In this context, the development of
innovative and more efficient catalytic processes based on fundamental understanding of the underlying reaction mechanisms remains
a largely unresolved challenge for academic and industrial chemists. Here, we cover selected examples of computational and
experimental knowledge-driven approaches for the rational design of transition metal catalyzed transformations.

1. Introduction
In 1987, the World Commission on Environment and
Development (WCED) published the report “Our Common Future”,
also known as the Brundtland report, to alert on the most critical
global issues: environmental protection, economic growth and
social equity.[1] This report also introduced the key concept of
sustainable development as “development that meets the needs
of the present without compromising the ability of future
generations to meet their own needs.” It is undeniable that
Chemistry is crucial for achieving a transition to a circular
economy, by providing sustainable strategies for the use of raw
materials (or building blocks) and energy sources.[2] Over the past
two decades, transition metal (TM) catalysis has played a key role
in tackling some of the major problems that our world faces such
as climate change, energy conversion or human health by

[a]

[b]

[‡]

J. Sanjosé-Orduna, Á. L. Mudarra, Dr. S. Martínez de Salinas, Dr.
M. H. Pérez-Temprano
Institute of Chemical Research of Catalonia (ICIQ)
Avgda. Països Catalans 16, 43007 Tarragona (Spain)
E-mail: mperez@iciq.es
J. Sanjosé-Orduna, Á. L. Mudarra
Departament de Química Analítica i Química Orgànica, Universitat
Rovira i Virgili, C/ Marcel·lí Domingo s/n, 43007 Tarragona (Spain)
These authors contributed equally to this work.

developing more efficient processes and/or designing new
catalysts. In this regard, most advances have been accomplished
based on trial-and-error approaches or serendipitous discoveries.
Surprisingly, one strategy that has received very limited attention,
despite its potential to provoke a paradigm shift in terms of
sustainability, is the rational design of chemical transformations
based
on
knowledge-driven
approaches.
Mechanistic
investigations are typically applied as a posteriori tool, focused on
understanding the reaction mechanism of successful
transformations. However, this concept is beginning to evolve due
to the employment of fundamental knowledge as reaction design
resource. Over the past few years, computational methods have
showed their tremendous capability, not only for predicting
catalyst performance but also for improving their efficiency.[3]
Experimental mechanistic strategies can also facilitate
tremendously the sustainable development of transition metalcatalyzed transformations. From our point of view, this paradigmshift has to rely on the use of key reaction intermediates as
“knowledge building blocks” (KBBs). The foundation of this
approach is based on this conception: the success or failure of a
whole chemical process relies on the performance of the reaction
intermediates involved in each elementary step that constitutes
the catalytic cycle. This may seem very obvious, but surprisingly,
it is often overlooked and it can be a completely game-changer
for non-efficient transformations, enabling gain in terms of money,
time and human resources. Therefore, we envision to design
efficient transformations by trapping these reaction intermediates
and using them to overcome limitations and explore innovative
reactivities.

In this minireview, we discuss different computational and/or
experimental knowledge-driven approaches, including the
synergistic cooperation between them, for the rational design of
TM-catalyzed transformations. In particular, to demonstrate the
advantages of these approaches, herein we review their
application in different relevant organic processes, including two
cornerstones in modern organic chemistry: (i) TM-catalyzed
cross-couplings, that emerged in the late 70s revolutionizing the
way in which molecules are built,[4] and (ii) C–H functionalization
reactions, one of the “holy grails” in synthetic organic chemistry.[5]

Ángel L. Mudarra was born in Escacena
del Campo (Huelva, Spain) in 1992. He
received his B.S.c. degree in Chemistry
from University of Huelva (Huelva, Spain)
in 2014. In 2015, he got his M.S.c. degree
from the University of Sevilla (Spain), and
next he started his PhD studies at the
Institute

of

Chemical

Research

of

Catalonia (ICIQ) under the supervision of
Dr. Mónica H. Pérez-Temprano and Prof.
Feliu Maseras. His Ph.D. is focused on
mechanistic investigations, combining computational and experimental

Knowledge Building
Blocks

Sustainable
Chemistry

context of homogenous catalysis.

Sara Martínez de Salinas received her B.S.c.
degree in Chemistry from the University of

DFT
analysis

Descriptors

approaches, to understand and develop bimetallic processes in the

Oviedo in 2010. Then, she obtained her
M.S.c. degree in Synthesis and Chemical
Reactivity at the same university in 2011
under the supervision of Prof. Elena Lastra.
In 2015, she got her Ph. D. in organometallic
chemistry under the supervision of Prof.

Figure 1. Rational design of chemical transformations based on knowledgedriven approaches.

Elena Lastra working in rhodium systems. In
December 2015, she joined the group of Dr.
Mónica

H.

Pérez-Temprano

as

a

postdoctoral researcher and she is currently working to obtain fundamental
understanding of relevant organometallic processes in the context of
bimetallic catalysis and trifluoromethylation.

Jesús Sanjosé-Orduna was born in
Barcelona, Spain, in 1991. He received
both his B.S.c. and M.S.c. from the
University of Barcelona (UB) in 2014 and
2015, respectively. Shortly afterwards, he
joined the Pérez-Temprano group at the
Institute

of

Chemical

Research

of

Catalonia (ICIQ) in Tarragona, Spain, to
pursue his Ph.D. degree. Since then, he
investigates the intricacies of Cp*Cocatalyzed

C–H

functionalization

reactions, not only to provide a comprehensive mechanistic picture of
these transformations but also to improve their efficiency based on
fundamental knowledge.

Mónica H. Pérez-Temprano was born in
Valladolid, Spain, in 1982. She recevied her
PhD degree at the University of Valladolid,
under the supervision of Prof. Espinet and
Prof. Casares, in 2011. Next, she moved to
the University of Michigan to work with Prof.
Sanford on the synthesis and reactivity of
high-valent palladium (IV) complexes. In
2015, she began her independent career as
Junior Group Leader at the Institute of
Chemical Research of Catalonia (ICIQ). Her
research program is focused on the rational design of more sustainable
approaches for the synthesis of organic molecules using fundamental
organometallic chemistry.

2. Computational tools for rational design in
homogeneous catalysis
Over the last two decades, computational chemistry has evolved
into a key tool for unravelling the mechanistic intricacies of
transition metal-catalyzed systems.[6,3b] Currently, due to the
major advances in the development of theoretical methods,
software and hardware, it is possible to resolve Quantum

Mechanics (QM) systems at a reasonable time-scale. This allows
the exploration and understanding at molecular level of chemical
transformations, even without simplifying the systems. In this
regard, Density Functional Theory (DFT) stands out as a powerful
posteriori tool which provides virtual access to critical mechanistic
information when experimental strategies exhibit intrinsic
limitations. Indeed, modern DFT methods have reached such
level of accuracy that can even quantitatively reproduce
experimental outcomes, turning not only into a real-time
complementary strategy for supporting mechanistic proposals but
also into a truly predictive tool.[6d-e,7] This later approach is
particularly promising since it paves the way for the rational
design
of
more
efficient
transition
metal-catalyzed
transformations and, subsequently, the development of
sustainable organic processes. The ideal work-flow for computerbased discoveries involves the synergistic cooperation between
DFT calculations and experiments. First, preliminary results and
computation are essential to make an initial prediction. Next, this
hypothesis is tested experimentally, and if the desired goal is not
achieved, iterative calculation/experimental observation feedback
loops are performed in order to fine-tune the initial guess (Figure
2). In this section, we will cover selected examples where it is
shown the capability of computation to provide meaningful
predictions and their impact on the development of innovative and
more efficient TM-catalyzed reactions. We have classified the
selected systems into two categories, depending on the
computed data treatment: (i) the DFT-design of efficient catalysts;
and (ii) the employment of statistical analysis of big data sets to
predict a desired performance.

Figure 2. Work-flow for merging the capabilities of computational chemistry and
experiments in the reaction optimization or discovery.

2.1 DFT calculations for enabling catalyst design
One of the main challenges associated to TM-catalyzed
transformations is the potential formation of catalytically inactive
species and/or off-cycle complexes, along with undesired side
reactions, that can dramatically impact on the efficiency of these
systems. Over the past few years, DFT calculations have been
used to overcome these limitations by gaining mechanistic
insights on certain intermediates and/or transition states (TSs)
and, subsequently, predicting essential modifications on the
catalyst design process.

2.1.1 Accessing virtual intermediates to understand key
processes in nucleophilic trifluoromethylation reactions
An excellent example on the employment of computational
predictions to promote challenging reactivities has been reported
by Schoenebeck and co-workers in the context of Pd-catalyzed
nucleophilic trifluoromethylation reactions (Scheme 1).[8] Only few
examples of bulky and/or wide-bite-angle phosphines have
demonstrated their capability to promote the challenging
formation of Ar–CF3 from LnPdII(Ar)(CF3) complexes.[8c,i,m]
Ar

X

LnPd

Oxidative
Addition

Ar
X
Si CF3
Si = TMS or TES

LnPd

Pd Catalytic
cycle

Reductive
Elimination
Ar

Transmetalation

Si X
CF3

LnPd

Ar
CF3

Scheme 1. Proposed catalytic cycle for Pd0/II-catalyzed trifluoromethylations.

Between 2011 and 2014, Schoenebeck et al. explored the
potential of computational chemistry in the design of an efficient
phosphine ligand for facilitating challenging C–CF3 bond-forming
reductive eliminations.[8d,f] Initially, they compared, theoretically,
the efficiency of two synthetically accessible LnPd(Ph)(CF3)
complexes bearing Xantphos and dppe as ligands. The former
was known to promote the product release while the latter
required high temperature even to afford Ph-CF3 in low yield.[8a-b]
This computational investigation revealed that, instead of the bite
angle of the phosphine, the crucial factor for facilitating the
reductive elimination was the repulsive interaction between the
ligand and the organic groups involved in the coupling. With this
information in hand, they designed an unprecedented and
experimentally efficient politrifluomethylated phosphine with a
narrow bite angle (Scheme 2).[8f] As mentioned above, the low
barrier of the reductive elimination event can be explained due to
the electrostatic repulsion between the ligand and both, the aryl
and trifluoromethyl groups, on the palladium(II) complex. Other
remarkable factor for the success of this reaction is the
stabilization of Pd0 due to the electron-withdrawing nature of CF3
substituent in the phosphine. This work stands out as a
representative case of computational-guided ligand design.

Scheme 2. Computational study of phosphine ligand scaffolds.

2.1.2 Dissecting energy components in Cu-catalyzed antiMarkovnikov hydroamination reactions

Another interesting example of computer-based ligand design has
been reported by Buchwald and Liu for the copper-catalyzed antiMarkovnikov hydroamination reactions of unactivated terminal
alkenes.[9] These transformations are particularly relevant for
medicinal chemistry since the resulting organic scaffold is a
prevalent motif in bioactive molecules.[10] Previous mechanistic
investigations demonstrated that, among the different elementary
steps involved in the catalytic cycle, the hydrocupration is the ratedetermining step (RDS) when using unactivated and/or terminal
olefinic substrates (Scheme 3).
s-bond
Metathesis

R3SiOBz

R
LCuH

R

LCuOBz

Reductive
Elimination
R

Bn
N
Bn

Hydrocupration

RDS

R3SiH

unactivated
alkene

Catalytic
cycle

R
L
Cu
Bn N OBz
Bn

CuL
Oxidative
Addition
Bn
N
Bn OBz

Scheme 3. Proposed catalytic cycle for the copper-catalyzed anti-Markovnikov
hydroamination reactions.

Combining computational and experimental approaches, they
were able to predict potential modifications on SEGPHOS-based
ligand scaffolds to accelerate the hydrocupration step. Initially,
preliminary kinetic studies revealed a promising and unexpected
rate increase when using CF3-SEGPHOS, a trifluoromethylated
ligand. Next, the authors performed a computational ligandsubstrate interaction analysis, between DTBM-SEGPHOS or CF3SEGPHOS and propene as model substrate, to rationally
introduce further modifications into the ligand that could lead to
an increase in the reaction rate (Scheme 4). At this point, they
were able to identify and evaluate the main energy contributions
to the activation barrier of the RDS, such as the distortion energy

necessary to reach the TS geometry or stabilizing through-space
and/or through-bond interactions between the olefin and the
LCuH moiety. The energy decomposition analysis disclosed a
significantly stronger through-bond interaction for CF3-SEGPHOS
than for DTBM-SEGPHOS in the TS. This is likely due to the
electron-withdrawing nature of the trifluoromethyl substituents,
which increases the Lewis acidity of the copper catalyst and
favors the binding of the alkene. For the through-space
interactions, both ligands provided comparable energies but for
different motives: TSDTBM-SEGPHOS was more stabilized by attractive
London dispersion forces whereas TSCF3-SEGPHOS exhibited more
important through-space electrostatic interactions via C–F…H–C
contacts between the fluorinated ligand and the olefin. With this
information in hand, they predicted that the substitution of the CF3
moiety by a larger perfluorinated chain, i-C3F7, could maintain the
existing favorable through-space and through-bond interactions
and increase the stabilizing London dispersion interactions. When
the reactivity of the (i-C3F7-SEGPHOS)CuH catalyst was tested,
the product formation was speed up at its initial stage, but the final
yield did not improve. This experimental result pointed out the
high activity of the catalyst along with its low stability. In order to
find the perfect balance between the different stabilizing factors,
the authors decided to design a hybrid SEGPHOS-based ligand,
merging the high stability of DTBM-SEGPHOS and the high
activity of i-C3F7-SEGPHOS. When using this new ligand, the
reaction rate of the hydroamination product was 62 times faster
than the one for DTBM-SEGPHOS but maintaining its stability. In
addition, this catalyst system was successfully used for the
transformation of a wide substrate scope under preparative
conditions. This work showed the powerful combination of kinetic
investigations, computer-driven catalyst design and experimental
verification for the rational design and development of more
efficient catalytic systems.

Scheme 4. (a) Model system. (b) Ligand design.

2.1.3 Engineered iridacycles for the efficient formation of
lactams
In 2018, Chang and co-workers reported a very elegant example
of the synergistic cooperation between experimental and
computational mechanistic investigations for the selective
synthesis of -lactams, a prevalent scaffold in medicinal
chemistry, via a key metal-nitrenoid intermediate.[11a] Despite
these cyclic amides are prevalent scaffolds in medicinal
chemistry, this TM-catalyzed C–H amidation strategy had been
hampered due to the inherit instability of the proposed
carbonylnitrene intermediate. This highly reactive species tends
to decompose affording an undesired isocyanate through a
Curtius-type rearrangement (Scheme 5).[11b,c,d]
Scheme 6. Stoichiometric experiments using 1Ir-ArFCN.

Scheme 5. Reactivity of metal-nitrenoid intermediates.

The authors tackled this challenge by engineering Cp*IrIII
complexes that could block the decomposition route and facilitate
the C–H insertion. Using the well-defined iridacycle 1Ir-ArFCN,
they performed different stoichiometric reactions using phenyl1,4,2-dioxazole derivatives as nitrogen source. They observed
that
1,4,2-dioxazol-5-one
reacts
faster
than
(oisopropyl)phenyldioxazole, affording the desired lactam in less
than 5 minutes along with CO2 extrusion. In sharp contrast, when
they tested a dioxazolone containing a more flexible alkyl chain,
instead of the corresponding lactam, they detected the formation
of a six-membered Cp*IrIII-amido complex formed via a C–N
coupling. It should be noticed that the isocyanate by-product was
not observed in these transformations (Scheme 6).

To further understand the chemoselectivity dependence of the
reaction depending on the nature of the substrate, they computed
the energy barriers for the C–N coupling, the Curtius
rearrangement and C–H insertion starting from an iridium(V)nitrene intermediate and using the dioxazolone containing the
flexible chain as substrate. DFT calculations unraveled that the
most facile transformation from this high-valent Cp*IrV species is
the C–N coupling. In order to prevent this side-reaction, the
authors proposed the substitution of the phenylpyridine ligand by
a monoanionic LX-donor ligand (X = O, N), since the N–O or N–
N couplings would be more disfavored. Next, they perfomed a
careful analysis of the computed natural bond orbitals (NBO) of
every transition state (Scheme 7). These calculations revealed
that the Curtius rearrangement transition state should be more
sensitive to changes of the partial charge of the Ir center.
Therefore, they hypothesized that the employment of electronrich ligands could increase this energy barrier, shutting down this
undesired decomposition route.

enantioselectivity when using chiral diamine scaffolds. DFT
studies demonstrated that the presence of transient
intramolecular hydrogen bonding interactions between the
substrate and NH substituent of the diamine ligand are
responsible for the excellent stereoselectivity.

Scheme 9. Rational-driven tailoring of iridium catalysts.
Scheme 7. Computational studies on the reactivity of Cp*IrV nitrenoid species.

2.2 Predictive statistical models for rational design
Following this DFT-based rational design, the authors tested a
new family of easily synthesized and air-stable Cp*IrIII catalysts
containing different LX-type ligands. The authors observed almost
quantitative formation of the corresponding lactam when using
LNR ligands, in less than 2 hours at room temperature. This
benchmark reaction demonstrated the potential of this
transformation since it allows the preparation of unprotected lactams. Using these optimal reaction conditions, Chang and coworkers explored the intramolecular amidation of different types
of C–H bonds, using a wide variety of dioxazolones derivatives
(Scheme 8). They applied their amidation protocol for the late
stage functionalization of complex molecules such as high-value
amino acids to convert them into lactams.

Scheme 8. Cp*IrIII-catalyzed synthesis of -lactams using readily available
carboxylic acids as starting materials.

Related to this work, more recently, Chang et al. described a
second generation of tailored iridacycles to enable the formation
of asymmetric -lactams (Scheme 9).[12] In this work, they
evaluated
different
ligands,
observing
an
excellent

Despite the promising results shown in Section 2.1, the rational
design of efficient catalytic systems based on computational
chemistry present several challenges from a theoretical point of
view. Among others, the necessity of time-demanding
calculations for the evaluation of different reaction mechanisms,
which can operate simultaneously, and the one-by-one analysis
of certain key species, which could present different
conformations, slows down the DFT-based rational design of TMcatalyzed organic transformations.[13] An alternative strategy to
overcome these limitations is the use of statistical modelling
methods, which handle big amounts of data at a significant lower
computational cost. These statistical approaches can predict
reaction performances by correlating both calculated and
experimental structural/physical organic molecular descriptors
with reaction outputs such as activity or selectivity. Since the
pioneering development of the Hammett parameter and its
application in Linear Free Energy Relationships (LFERs), more
complex multidimensional correlations and quantitative
parameters have been generated by the organic chemistry
community to predict experimental results.[14] It should be noticed
that not only the selection of a representative molecular descriptor
is key for the success of this approach but also its statistical
significance. Therefore, this strategy requires the treatment of
relevant and abundant experimental data, including, if existed,
negative outcomes. This is remarkable since both negative and
positive results are equally valuable in the research procedure.
Here, we will cover selected examples on the application of
statistical correlations for predicting reaction outcomes where
fundamental mechanistic information is provided as consequence
of the LFERs.
2.2.1 Parametrization in Suzuki cross-couplings reactions
The Suzuki-Miyaura reaction is one of the most general and
powerful methodologies used in the formation of CC bonds in

medicinal and organic chemistry.[15] In 2016, Sigman and coworkers described a multivariate correlation approach for
examining and predicting the outcome of a previously-reported
Pd-catalyzed cross-coupling reaction, where the nature of the
ligand determined the product formation.[16] In the literature
precedent, Fu et al. controlled the regioselectivity in the Suzuki
coupling of chloroaryl triflate derivatives by modifying the
phosphine ligand (Scheme 10).[17]

Ar

OTf Pd2dba3 (1.5 mol%)
Ligand (3 mol%)

B(OH)2 +
Cl

KF (3 equiv)
THF, 24 h, RT

OTf

Ar
+
Cl

L = PcHex3
87% yield

Ar
L = PtBu3
95% yield

In 2018, Biscoe and Sigman were able to apply predictive
statistical models to the rational design and development of
stereospecific Pd-catalyzed Suzuki cross-coupling reactions,[19]
one of the limitations associated to these transformations. These
reactions are typically hampered by a slow transmetalation step
between LnPd(Ar)X and a sterically hindered alkylboron
nucleophile and/or the tendency of the resulting alkyl-palladium(II)
complex to undergo -hydride elimination/reinsertion sequences
that entail isomerization of the alkyl stereocenter (Scheme 12).
Moreover, the transmetalation step can play a crucial role in the
stereochemical outcome of the reaction when using a singleenantiomer organoboron nucleophile, since it can proceed via
stereoretentive or stereoinvertive pathways depending on the
reaction conditions.
Ar
R

Scheme 10. Chemoselective Suzuki Cross-Couplings.

Reductive
Elimination

Given the dependence of the reaction outcome on the phosphine
nature, Sigman et al. tackled the mathematical modelling by
studying significant parameters of this ligand such as cone-angle
measurements, Sterimol values, computed infrared values or
experimental NMR measurements (Scheme 11). Based on
previous computational studies by Schoenebeck and Houk,[18]
and their own experimental cyclic voltammetry (CV)
measurements, Sigman and co-workers classified the ligands into
two groups according to the speciation of the palladium(0)
complex that undergoes the oxidative addition: large phosphines,
which promote the oxidative addition to monoligated palladium
(LPd) complexes, and “smaller” or Buchwald phosphines, which
preferentially form the bis-ligated (L2Pd) complexes that undergo
the oxidative addition event. For larger phosphines, highly
selective toward the oxidative addition of the C–Cl bond, the
reaction outcome can be predicted using the phosphine selenide
31
P NMR chemical shift as a single descriptor for the model. For
smaller and Buchwald biaryl phosphines, a more complex
multivariate predictive model, which contains four different
parameters, was found. They rationalized the relationship of
these parameters with steric and electronic characteristics of the
phosphines that explains the preferential oxidative addition of C–
OTf bond.

Scheme 11. Statistical analysis of Suzuki-Miyaura reaction selectivity.

L PdII Ar
R

R1

or R

LPd0

L PdII Ar

Ar
LPdII

R1

Ar

L
R

PdII

H

X

[B]

-Hydride
Elimination

Hampered
for large L

X
Oxidative
Addition

inversion

retention

Ar

R1

R
R1
Transmetalation
[B] X

For large L, it depends on
electronic properties of L

R1

Scheme 12. Mechanistic proposal for stereodivergent Suzuki cross-coupling.

The authors carried out initial investigations that provided
additional relevant mechanistic observations: (i) they detected a
loss on the stereofidelity of the resulting product when using
electron-deficient aryl chlorides as coupling partners; and (ii) they
did not observe a clear correlation between the stereoselectivity
of the reaction and the steric properties of the different tested
phosphines. Challenged by these results, they employed ligand
parameterization tools to unravel the factors that control the
transmetalation step in order to predict and rationally design a
new ligand-controlled enantiodivergent process. Initial studies
showed that larger ligands could suppress the undesiredhydride elimination reaction. Further statistical analysis of the
ligands that provided high enantiomeric excess (>30% ee),
indicated that electron-rich phosphines promote invertive
reactions while electron-poor phosphines favor the retentive ones
(Scheme 13a). Based on this information, they were able to
develop stereodivergent Suzuki coupling protocols using
enantioenriched alkyl boron nucleophiles, depending on the
selected phosphine ligand (Scheme 13b). The stereoretentive
pathway was facilitated by designed biaryls phosphine ligands
that combined a bulky substituent and an electron-poor group in
their scaffolds. The stereoinvertive route was promoted by PAd3,
a strongly s-donating ligand. Finally, a multivariate regression
analysis provided evidences for the origin of the enantiodivergent
transmetalation step depending on the nature of the ligand.

(a) Ph2P

P

ee = 52 %

2

MeO

Bulky moiety to
avoid -hydride
elimination

ee = -93 %

P

CF3
Electron-poor
moiety for
estereoretention

bis-CF3PhXPhos

ee = 9 %

P

2.2.2 Analyzing TM-catalyzed C–H activation by using
statistical analysis

CF3

Predicted
guess

OMe

Experimental
result
ee = 90 %

3

(b)

Ligand-controlled stereodivergence

BF3K
R

R1

+

L Pd NH2
Ar
Ar
OMs (cat.)
or
Ar X
Conditions A or B
1
R
R
R
R1
X = Cl,
Conditions A
Conditions B
Br, OTf
L = PAd3
L = bis-CF3PhXPhos
or bis-CF3PhSPhos

Scheme 13. (a) Phosphine parameterization. (b) Stereodivergent Pd-catalyzed
suzuki cross-coupling using enantioenriched alkyl boron nucleophiles.

The parameterization of phosphine ligands has also been applied
for the synthesis of benzylic ether derivatives by Ni-catalyzed
Suzuki-Miyaura cross-coupling reactions. In this work, Doyle and
Wu observed a clearly dependence between the yield of the
reaction and the steric size of the phosphine ligand.[20] However,
they revealed that presumably interchangeable descriptors for
steric parameters, such as the cone angle () and the buried
volume (%Vbur), are not correlated. The cone angle, represents a
virtual cone that encloses all the substituents on the phosphorous
atom. This means the remote sterics at the distance from the
metal center. In contrast, the buried volume only evaluates the
proximal steric effect. The analysis of 17 bulky phosphine ligands
established a positive direct correlation with the cone angle ()
and an inverse one with the buried volume (%Vbur), establishing
that the effectiveness of ligands relies on their remote steric
hindrance (Scheme 14). A multivariate linear regression,
combining both parameters, offered an excellent quantitative
model for predicting ligand reactivity in Ni-catalyzed SuzukiMiyaura couplings of benzylic acetals.
(a) Reactivity
Ar
OR
Ar

OR
Ar

O
B

B
O

Ni(cod)2 (15 mol%)
Ligand (30 mol%)

O
B

OR

Toluene, 60 ºC, 16 h

Ar

Ar

Ar

(b) Phosphine parameters relevant to prediction and designed ligand
Cone angle ()
Parameterization

Cyp2P

R

TRIP
TRIP = 2,4,6-triisopropylphenyl

P
M

High  and low % Vbur

% Vbur

TRIP

Cyp = cyclopentyl

Scheme 14. (a) Ni-catalyzed Suzuki-Miyaura cross-coupling. (b) Crucial
parameters according to the MLR model.

Multivariate linear regressions (MLR) have been also applied for
the rational design of TM-catalyzed C–H functionalization
reactions. In 2017, Paton and Rovis demonstrated how subtle
modifications in cyclopentadienyl (Cpx) ligand structures can
impact the reaction rate and/or the selectivity of RhIII-catalyzed
annulation reactions of 1-decene (Scheme 15a) or cyclopropene
(Scheme 15b) with a benzohydroxamic acid derivative.[21] They
evaluated different electronic (IR stretching frequencies, NMR
data, redox potential, charges) and steric (Tolman cone angles
and Sterimol values) parameters of 22 different CpxRhIII
complexes to develop a quantitative predictive model that could
describe the performance of these catalysts.
(a) Reaction model for study the regioselectivity
O
OPiv
O
N
cat. [Rh]
H
NH
CsOAc
+
+
MeOH, RT
n-Oct
n-Oct
A
B

tBu

tBu
Cl Rh
Cl

O

2

NH

n-Oct

Cl Rh
Cl

2

A:B
12 : 1

A:B
1.06 : 1

Ph
(b) Reaction model for study the stereoselectivity
O
Ph
Cl Rh
OPiv
O
O
N
A:B
cat. [Rh]
Cl
H
2 19.5 : 1
NH
CsOPiv
NH
+
+
H
H
MeOH, RT
Me
H
H
Me Cl Rh
A
B
Ph Me
Ph
Ph
A:B
Cl
2 0.90 : 1

Scheme 15. Selected CpXRhIII-catalyzed reactions to study the regioselectivity
and stereoselectivity through MLRs.

They found that, in the case of 1-decene, the predictive model for
the regioselectivity was composed by three electronic descriptors
(31P chemical shift, antisymmetric CO stretching frequency and
103
Rh shielding tensor) and three steric parameters (Sterimol
value, L, and Bondi cone angles). Regarding the selected
cyclopropene as model substrate, the linear regression model
that correlates the observed diastereoselectivity and the structure
of the CpxRhIII complexes involved three electronic parameters
(31P chemical shift, NBO charges at rhodium center and RhIII/II
redox potential) and a single steric descriptor (the minimum Bondi
cone angle).
More recently, Rovis et al. performed a multivariate analysis to
control the diastereoselectity of RhIII-catalyzed cyclopropanation
of alkenes with N-enoxyphthalimides.[22] Using ethyl acrylate as
coupling partner, they first evaluated the influence of 15 Cpxligated RhIII complexes and 8 N-enoxyphthalimide derivatives in
the diastereoselectivity of a model cyclopropanation reaction
(Scheme 16a). While the trans-cyclopropane diastereomer was
generally favored, sterically demanding Cpx ligands and electrondeficient phthalimide combinations showed the highest cisselectivities. When using NaOAc as base, a multivariate
correlation of 96 catalyst/substrate pairs showed the existence of

two parallel mechanistic regimes, involving the formation of
intermediate A or B (Scheme 16b). When using electron-rich
phthalimides and small Cpx ligands, the formation of the transproduct is favored via intermediate A. For more stericallydemanding CpX ligands, electron-deficient phthalimides and/or
more strong Lewis acids, cis-product is favored. Based on this
information, the authors developed a cis-cyclopropanation
protocol using a wide range of enoxyphthalimide derivatives and
alkenes, where the phthalimide ring-opening was presumably the
rate-determining step (through Intermediate B).
(a) Model system for stereoselectivity study
O
[CpxRhCl2]2
Ph
CsOAc
+
COOEt
N O
TFE, 23 ºC
R
O

Ph

of the N-acyl group (R1, Scheme 17) along with the nature of the
carboxylate additive in the reaction outcome. Interestingly, they
provided a predictive multivariate linear model which displays
exclusively dependence on the aforementioned NBOO-avg
parameter and the Sterimol B1 value of the N-directing group. The
authors carried out further experimental mechanistic
investigations and confirmed the selective C2–H activation by the
synthesis and characterization of the corresponding
cyclometalated Cp*IrIII species. They applied this mechanisticdriven approach, not only for developing a versatile and C2selective C–H amidation protocol but also for inverting the
selectivity of a previously described C7 alkenylation reaction.

COOEt
O

(b) Mechanistic picture supported by MLR and experiments

i-Pr
O
N

CpxRh(OAc)2
Ar

O

Ar
H

H

O
H
- TFE

+ TFE

= EWG

Ar

O
O

CpxRh(OAc)2

N
H
CO2R

Scheme 16. RhIII-catalyzed
enoxyphthalimides.

O

Ar

O N

H

Rh O

H

O N
Rh O

- Flexible species H
- Interconversion cis/trans
- trans-product
Cy
Intermediate B

Ar
H

i-Pr

Intermediate A
O

O O
N
H
Rh O
H
OR
- Stiff species
- cis-product

cyclopropanation

favored for NaOAc
&
electron-poor
phthalamides

Scheme 17. Regioselective Cp*IrIII-catalyzed amidation of indoles
of

alkenes

with

N-

Chang and co-workers have also used this physical organic
approach for the rational design of regioselective Cp*IrIIIcatalyzed amidation of indoles using azides as coupling
partners.[23] They were able to selectively switch the C–H bond
cleavage from the C7 to C2 position of N-acylindole derivatives
by gaining insights of the effect of stereoelectronic properties of
the substrate and the catalyst over the reaction outcome. Initially,
the authors combined experimental and computational studies to
evaluate the possible parameters that could affect the selectivity
of the C–H metalation step. They observed that the C2-product
formation is highly dependent on the size of the N-acyl group,
increasing its ratio when reducing the size of the directing group.
Moreover, they identified high selectivity towards the C2
functionalization when using silver trifluoroacetate in lieu of more
basic carboxylates such as silver pivalate. With this information in
hand, they evaluated the effect of different parameters of the
substrate/carboxylate in the reaction outcome. First, the
evaluation of different electronic (NBO charges of oxygens and
CO stretching frequencies) and steric (Sterimol values)
descriptors of 23 carboxylic acid surrogates, showed that a
univariate model using NBOO-avg as parameter reproduces the
experimental results. Then, they evaluated the effect of the size

3. Experimental tools for rational design
Mechanistic experimental information has been historically used
as a posteriori tool to understand how reactions work. However,
recently, the isolation of putative intermediates in the catalytic
cycle has been employed for boosting the rational design and
development of new protocols. In this section, we have selected
relevant examples in which the isolation of the key reaction
intermediates and their use as “knowledge building blocks” has
provided a hallmark in the improvement of the different
methodologies.
3.1 Cross-coupling reactions
As mentioned in the introduction, the formation of CC bonds
using TM-catalyzed reactions is one of the most remarkable
transformations in organic chemistry and it has received special
attention since the second half of the 20th century.[4] We highlight
the potential of KBBs in two different reactivities: (i) palladiumcatalyzed perfluoromethylation reactions using silver nucleophiles
as transmetalating reagents and (ii) nickel-catalyzed SuzukiMiyaura coupling.
3.1.1 Perfluoromethylation reactions mediated by Palladium
and Silver Bimetallic Systems

Perfluoroalkyl groups such as CF2H or CF3 are privileged
structural motifs due to their unique capability to modify the
physical, chemical and biological properties of organic molecules.
In this context, Pd/Ag bimetallic systems have demonstrated their
ability to introduce these groups into organic scaffolds.[24] The key
step in these processes is the transmetalation between welldefined silver derivatives to oxidative addition palladium
complexes. In this regard, in 2014 and later in 2017, Shen et al.
accessed putative reactive intermediates to provide fundamental
understanding of the pivotal steps of the mechanistic proposal
shown in Scheme 18a to enable the rational design of efficient
catalytic protocols. In their pioneering work, the authors initially
investigated the reductive elimination step from LnPd(Ar)(CF2H).
Remarkably, these complexes afforded the desired product in
high yields in the presence of one equivalent of the ligand. Next,
the authors investigated the viability of the transmetalation step,
from (dppf)Pd(Ph)I or (Xantphos)Pd(Pyr)Br, using a well-defined
difluoromethyl silver complex, SIPrAg(CF2H), stable enough to
promote the CF2H transference prior to its decomposition. In both
cases the corresponding difluoromethylated products (palladium
complexes or directly the organic molecules due to the fast
reductive elimination) were obtained in quantitative yields. With
this fundamental knowledge in hand, the authors developed a
catalytic difluoromethylation protocol of aryl halides (using
catalytic amounts of silver) while in the case of the heteroaryl
halides a fine tuning of the ligand (DPePhos instead of XantPhos)
was needed to obtain high yields of the corresponding organic
products (Scheme 18b).[25]
isolated
SIPrAgCF2H

(a) Proposed catalytic cycle

considered that the reductive elimination step is the only problem
associated with these processes.[8] However, in 2006, Grushin et
al. reported that, when using Xantphos as ligand, a slow
transmetalation reaction could promote undesired competitive
reactions such as: mismatched transmetalations between arylpalladium(II) complexes and/or ligand displacement by
commercially available CF3 nucleophiles (Scheme 19a).[8a] For
this reason, we decided to study the transmetalation step
independently to overcome the aforementioned challenges.
Initially, we used a model system to compare the reactivity as CF3
shuttles of the well-defined silver (I) complexes with commercially
available nucleophilic CF3 sources. Notably, all these synthesized
silver complexes, including a unique silver(I) ate-type complex,
(Cs)[Ag(CF3)2], outperformed the widely used sources (Scheme
19b). These results were translated to a productive system, using
Xantphos as ligand. Strikingly, the transmetalation reaction of
(Xantphos)Pd(Ph)I and (Cs)[Ag(CF3)2] afforded selectively
(Xantphos)Pd(Ph)(CF3) in one pot in less than 10 min. This result
was crucial to the thoughtful model of the one-pot formation of
PhCF3 using PhI as the starting material, under stoichiometric and
catalytic conditions (Scheme 19c). This work is a proof-of-concept
of the crucial role of the nucleophile in the success or failure of
the coupling process.

NaCl +
TMSOtBu

isolated
TMSCF2H
activation

L2Pd

Transmetalation

X
X

Oxidative
Addition

SIPrAgCl
Pd
cycle

L2Pd0

(i)

L2Pd
CF2H

L

(ii)

Ph

isolated
Reductive
Elimination

CF2H

TMSCF2H
+ NaOtBu

3-Py

dppf

Xantphos

(b) Pd/Ag protocols
Pd(dba)2/dppf
SIPrAgCl (cat.), TMSCF2H (2 equiv)
NaOtBu (2 equiv)
R
Dioxane or Toluene, 80 ºC, 4-6 h
X

Pd(dba)2/DPePhos
SIPrAgCF2H (1.3 equiv)
NaOtBu (2 equiv)
Toluene, 80 ºC, 4-6 h

CF2H

CF2H
Het

Scheme 18. (a) Proposed Pd/Ag synergistic cooperation in C–CF2H bondforming reactions. (b) Pd/Ag protocols.

Recently, in 2018, our group studied the reactivity of well-defined
silver
trifluoromethyl
complexes
in
Pd-catalyzed
trifluoromethylation reactions.[8l] From a long time, it had been

Scheme 19. (a) Challenges in Pd-catalyzed trifluoromethylation reactions. (b)
Model system. (c) Synthesis of PhCF3 under stoichiometric and catalytic
conditions.

3.1.2 Nickel-catalyzed decarbonylative Suzuki-Miyaura
coupling of acyl fluorides and aryl boronic acids

As mentioned in Section 2.2.1., the Suzuki-Miyaura reaction is a
powerful methodology for the formation of CC bonds.[15] A
general feature in the Suzuki catalytic cycle is the use of an
exogenous base to form a “transmetalation active species” that
facilitates the transmetalation of the boronic acid. However, the
presence of a base can promote undesired reactions that
decomposes the organoboron nucleophile. In this regard, Sanford
and co-workers envisioned the use of a combination of a nickel
catalyst, a ligand and acid fluoride electrophile to generate a
“transmetalation” active species that would react with the boronic
acid in the absence of base (Scheme 20a) [26] Initially, the authors
evaluated the efficiency of each step involved in the catalytic cycle
(Scheme 20b). Firstly, the oxidative addition and decarbonylation
were investigated by performing the stoichiometric reaction of
Ni(cod)2, PCy3 and benzoyl fluoride at room temperature. The
benzoyl nickel fluoride complex was formed immediately, and
subsequent decarbonylation gave rise the phenyl nickel fluoride
species. The later, namely as “transmetalation active species”
reacted with different boronic acids, including ones that undergo
facile decomposition pathways, affording the biaryl products in
high yields. This last experiment demonstrated the feasibility of
the transmetalation and reductive elimination step within the
catalytic cycle. When the authors tried the translation to catalytic
conditions, they observed the formation of aryl ketone as byproduct by the attack of the boronic acid directly to the acid
fluoride. Fine tuning of the ligand (PPh2Me instead of PCy3) avoid
this undesired competitive pathway. Further developments of this
new methodology permitted the use of carboxylic acids as starting
materials that are converted in the corresponding acyl fluorides
using a fluoride source (TFFH) and a proton sponge.
(a) Strategies for Suzuki-Miyaura reaction
Ar

B(OR)2

Ar

Pd catalyst, base

Ar X
Ar
Suzuki-Miyaura reaction

B(OR)2

Ni catalyst

Ar

(b) Proposed catalytic cycle for baseisolated
free Ni-catalyzed decarbonylative
PEt3
Suzuki-Miyaura reaction
O
Ni F
O
O
Ph
PEt3
NiII F
Ar
F
Ar
Decarbonylation
CO
[Ni0] Oxidative
Ar
Addition
Ar
Ar NiII F
Reductive
Transmetalation-active
Elimination
specie
Ar NiII Ar
Transmetalation
FB(OH)2
(c) Methodology
TFFH (1 equiv)
Proton sponge (1 equiv)

O
Ar

OH

THF, RT, 15-30 min

Ar CO2H

Sanford (2018)

Ar

Scheme 20. (a) State-of-the-art of TM-catalyzed Suzuki-Miyaura crosscouplings. (b) Mechanistic proposal for base-free Ni-catalyzed reactions. (c)
Synthetic protocol.

3.2 Knowledge-driven approach in TM-catalyzed C–H
activation processes
The activation of typically inert C–H bonds to construct more
complex scaffolds is one of the "holy grails" of modern synthetic
chemistry and has received special attention since the second
half of the 20th century.[5a-b] In this context, ligand-directed
transition metal-catalyzed transformations has emerged as a very
powerful strategy since they offer the advantage of allowing the
control site-selectivity in molecules that contain multitude of C–H
bonds.[5c-k]

3.2.1 Surpassing challenging elemental steps in Iridiumcatalyzed C–H activation by an induced reductive elimination
In 2018, Chang and co-workers reported a very elegant example
of the utilization of fundamental knowledge for enabling new
reactivity patterns in Cp*IrIII-catalyzed C–H functionalization
reactions (Scheme 21a).[27] In this work, they tackled a major
challenge related to the development of efficient TM-catalyzed
transformations: the high stability of some resulting reaction
intermediates that can preclude the formation of the desired
product. In their attempt to develop a Cp*IrIII-catalyzed C–H
arylation protocol using arylsilanes as coupling partners, the
authors identified the reductive elimination step as the bottleneck
of the process when involving traditional IrI/III catalytic cycles. In
order to overcome this limitation, they proposed to modulate the
electronics of the post-transmetalation Cp*IrIII intermediate, by a
selective oxidation, facilitating the challenging C–C bond-forming
reaction from a high-valent IrIV or IrV species (Scheme 21b).

isolated
PCy3
Ph Ni

F

PCy3

B(OH)2

No base needed
isolated
or in situ
O
Ar

F

Ar

B(OH)2

Ni(cod)2/PPh2Me
THF, 100 ºC, 16 h

Ar

Ar

Scheme 21. Overview of the Cp*IrIII-catalyzed C–H arylation.

For validating their working hypothesis, they used as platform
well-defined Cp*IrIII metallacycles, analogous to previously
described reactive intermediates in iridium catalysis.[28] As
expected, the authors did not observe the corresponding C–C
coupling from a post-transmetalation Cp*IrIII complex 4Ir-Ar,
confirming its high stability (Scheme 22a). Gratifyingly, they were
able to oxidatively induce the reductive elimination event in the
presence of different silver oxidants, and develop a catalytic
version under similar reaction conditions. These preliminary
results strongly supported, not only that the reductive elimination
is the rate determining step, but also the accessibility of carbon–
carbon bond-forming reductive elimination reaction from highvalent iridium intermediates. This experimental information
agreed with DFT calculations, that showed a facile product
formation from highly oxidized IrIV or IrV intermediates. Moreover,
electrochemical studies confirmed the potential access to these
high-valent iridium species using cyclic voltammetry (CV)
(Scheme 22b). Stoichiometric experiments with different oxidants
such as AgOTFA or acetylferrocenium tetrafluoroborate
(AcFcBF4), along with EPR spectroscopy measurements
suggested that instead of a IrIII/V pathway, a IrII/IV route is operative.
Finally, the authors were able to develop an innovative Cp*IrIIIcatalyzed directed C–H arylation protocol using arylsilanes as
nucleophilic coupling partners, based on fundamental
mechanistic understanding (Scheme 22). This tailored reaction
presented a wide substrate scope and a good group tolerance
under mild reaction conditions.

Scheme 22. (a) Mechanistic investigation on oxidative induced C–C bondforming reactions from a well-defined Cp*IrIII metallacycle. (b) Catalytic protocol.

Inspired by these results, Chang and co-workers have extended
their knowledge-driven approach to the development of TMcatalyzed C–H arylation reactions using arylboronic esters as
transmetalating agents. In this work, apart from iridium systems,
they also explored the potential of oxidative induced reductive
elimination pathways from well-defined Rh and Ru complexes, in
order to promote carbon–carbon bond-forming reactions.[29]

3.2.2 Ruthenium-catalyzed C–H late-stage functionalization
The employment of knowledge-driven approaches for the
synthesis of complex organic scaffolds represents one of its most
promising applications. In 2018, Larrosa and co-workers
described the rational design of a new family of ruthenium
catalysts capable of promoting the late-stage C–H arylation of
highly
decorated
N-chelating
molecules,
including
pharmaceuticals, agrochemicals or natural products, with
aryl(pseudo)halides.[30] In this work, the authors demonstrated
how fundamental mechanistic understanding can play a crucial
role in the development of more efficient transformations. A
thorough mechanistic investigation of the previously proposed
RuII/IV catalytic cycle[31] allowed them to uncover key
unprecedented mechanistic features: (i) the p-cymene typeligands, (e.g. 1Ru-MeCN, Scheme 23) widely used in these Rucatalyzed transformations,[32] inhibited the directed arylation
reaction (ii) the employment of a p-cymene free ruthenacycle 2RuMeCN afforded a faster product formation;[30,33] and (iii) they
identified the formation of a bis-cycloruthenated RuII intermediate
3Ru-MeCN, prior to the oxidative addition step, revealing that a
second C–H activation step occurs before accessing the highvalent RuIV (Scheme 23). Based on their experimental results, the
authors proposed a new mechanistic hypothesis, involving a biscycloruthenated intermediate, as an alternative to the widely
accepted catalytic cycle (Scheme 24).

“catalyst arylation” degradation derived from a mismatched
coupling with the 2-phenylpyridine (2ppy) scaffold of the
ruthenium pre-catalyst. In order to block this unproductive
pathway, the authors tailored a new pre-catalyst, modifying the
directing group, in order to differentiate the scaffold on the
ruthenium and the targeted substrates. The N,Ndimethylbenzylamine-containing cycloruthenated complex 4RuMeCN showed an exquisite selectivity, giving rise exclusively the
bis-arylated product in excellent yield (Scheme 25).

Scheme 25. Summary of the rational evolution of ruthenium-based catalysts.

Scheme 23. Precedents and mechanistic investigations of the rutheniummediated C–H arylation.

Scheme 24. Proposed catalytic cycles for the RuII-catalyzed C–H arylation.

With this information in hand, the authors investigated the
catalytic activity of different ruthenium complexes. As expected,
Ru catalysts containing p-cymene as ligand, like 1Ru-MeCN were
inactive. When using 2Ru-MeCN as pre-catalyst, the authors
observed an excellent combined yield of the mono and diarylated
products, but unfortunately they also detected an undesired

Finally, they decided to explore the synthetic utility of this
engineered
ruthenium
pre-catalyst in the
late-stage
functionalization of relevant drugs. They rationally designed a
robust protocol for the C–H arylation of ligand-directed substrates
with aryl(pseudo)halides, compatible with a wide range of
functional groups that are prevalent in medicinal chemistry. The
authors showed the potential of this methodology testing relevant
molecules for drug discovery such as atazanavir (HIV treatment),
diazepam (anxiolytic), sulfaphenazole (antibacterial) or
bromantane (anxiolytic) derivatives. As a final proof-of-concept,
they demonstrated the capability of their tailored ruthenium
catalysts to couple organic drugs between them, as shown in
Scheme 26.

DG
H

Reversible
C−H
activation

cat. [Cp*CoIII]
additives

DG
CoIII

DG

cost-effective alternative
to Rh(III)
unique reactivity

L

Knowledge building
block for rational
design

limited fundamental
understanding

Scheme 27. Cp*CoIII-catalyzed C–H activation reactions.

Scheme 26. Overview of the capability of 4Ru-MeCN in different late-stage
functionalizations.

3.2.3 Development of more efficient Cp*Co-catalyzed C–H
functionalization reactions based on knowledge-driven
approaches
As shown above, the rational design of innovative transition
metal-catalyzed transformations based on fundamental
knowledge represents one of the most powerful strategies for the
development of sustainable organic transformations. However,
this approach can be completely prevented in emerging fields,
such as cobalt catalysis, by the dramatic lack of mechanistic
understanding.
Over the past few years, the employment of Cp*CoIII complexes,
analogous to active RhIII catalysts for C–H activation, has
represented a tremendous advance in cobalt catalysis.[34] When
compared to noble metals, cobalt catalysts offer obvious
advantages, including being earth-abundant and cheaper.
Despite the significant growth in this field, these cobalt systems
are still at their infancy when compared to Rh- and Pd-based
catalysts, especially due to the limited fundamental
organometallic understanding of these systems. The investigation
of the underlying reaction mechanisms of Cp*CoIII-catalyzed C–H
functionalization reactions has been hampered by the difficulty of
trapping high reactive transient cobalt intermediates due to the
proposed reversible nature of the C–H metalation step (Scheme
27).[34]

Intrigued by the lack of fundamental understanding, our group
became interested in bringing light into the mechanistic “black
box” of Cp*Co-catalyzed directed C–H functionalizations in order
to enable more sustainable transformations. As mentioned above,
our knowledge-driven approach is based on trapping previously
elusive highly reactive cobalt species and using them as
“knowledge building blocks” (KBBs) for rational design.
Considering the proposed reversible nature of the C–H activation
step by Cp*CoIII complexes, our first challenge was to design a
synthetic route for accessing catalytically relevant cobaltacycle
complexes.
Using MeCN as stabilizing ligand, we were able to synthesize a
direct analogue of one of the most widely invoked intermediates
in Cp*CoIII-catalyzed C–H functionalization reactions, 1Co-MeCN,
via a ligand-assisted oxidative addition followed by a halide
abstraction.[35a] We used this cyclometalated cobalt(III) complex
to provide an unprecedented comprehensive picture on the most
explored Cp*CoIII-catalyzed transformation, the oxidative alkyne
annulation, including the first direct observation and full
characterization of a cobalt resting state under catalytic conditions
(Scheme 28). Our work revealed the higher catalytic activity of
1Co-MeCN compared to the widely used [Cp*CoI2(CO)] precatalyst; providing one of the lowest cobalt catalyst loading (1
mol%) reported to that date for this type of reactions in a short
reaction time of 2 hours.

Scheme 28. (a) Synthesis and characterization of a previously elusive key
cationic cobaltacycle via a ligand-assisted oxidative addition step. (b) Catalytic
activity in a model oxidative alkyne annulation.

Inspired by these results, we took advantage of the unique
stabilizing capability of MeCN to overcome the reversible nature
of the C–H activation step and access two of the most widely
invoked cationic cobaltacycles or KBBs which have been
proposed in the literature as reactive intermediates after the C–H
metalatation step.
Moreover, we unveiled the dramatic accelerating effect of the
perfluorinated alcohol 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP)
in the C–H metalation step.[35b] Indeed, the beneficial impact of
HFIP was not only limited to this elementary step. Our
stoichiometric and catalytic studies revealed that the presence of
HFIP also enhances the efficiency of two divergent reactivities,
the hydroarylation of alkynes and oxidative alkyne annulation
using as benchmark the reaction between diphenylacetylene and
N-pyrimidinylindole (Scheme 29). These examples demonstrated
the crucial role that mechanistic understanding can play for
designing more efficient processes.

Scheme 29. (a) Synthetic route for key cationic cobaltacycles via C–H
activation. (b) Beneficial effects of HFIP as additive under catalytic conditions.

4. Conclusions and Prospects
In summary, in this minireview we have covered examples that
illustrate the potential of knowledge-driven approaches for the
sustainable development of
transition metal-catalyzed
transformations. We anticipate that the selected computational
and experimental strategies, including their synergistic
cooperation, will pave the way for the employment of mechanistic
investigations as reaction design resource. In this context, we
envision that the next revolution in sustainable synthetic
chemistry will be based on getting a deeper fundamental
understanding of processes based on earth-abundant first-row
metals or applying these knowledge-driven approaches for the
rational design of innovative photoredox or electrochemical
transition metal-catalyzed reactions. Moreover, the exploration of
more statistical analysis methods such as Principal Component
Analysis (PCA),[36] can be crucial for enabling the sustainable
development of efficient transformations. We encourage the
scientific community to merge theoretical and experimental
approaches,
along
with organic
and
organometallic
understanding, to advance beyond the frontiers of knowledge and
improve the catalytic performance of processes involving easily
accessible raw materials as starting reagents.

Acknowledgements
We thank the CERCA Programme/Generalitat de Catalunya and
the Spanish Ministry of Economy, Industry and Competitiveness
(MINECO: CTQ2016-79942-P, AIE/FEDER, EU) for the financial
support. A. L. M. thanks La Caixa-Severo Ochoa programme for

a predoctoral grant. J. S.-O. thanks Severo Ochoa Excellence
Accreditation for a pre-doctoral contract.
Keywords: sustainable chemistry • reactions mechanisms •
transition metals catalysis • rational design • computational
chemistry
[1]

[2]
[$3]

[4]

[5]

[6]

[7]

[8]

Brundtland Commission [the World Commission on Environment and
Development (WCED)] Our Common Future; Oxford University: Oxford,
1987.
a) M. Beller, G. Centi, L. Sun, ChemSusChem 2017, 10, 6; b) T. Keijer,
V. Bakker, J. C. Slootweg, Nat. Chem. 2019, 11, 190.
a) M. S. Sigman, K. C. Harper, E. N. Bess, A. Milo, Acc. Chem. Res.
2016, 49, 1292; b) C. Poree, F. Schoenebeck, Acc. Chem. Res. 2017,
50, 605; c) J. P. Reid, M. S. Sigman, Nat. Rev. Chem. 2018, 2, 290.
C. C. C. Johansson Seechurn, M. O. Kitching, T. J. Colacot, V. Snieckus,
Angew. Chem. 2012, 124, 5150; Angew. Chem. Int. Ed. 2012, 51, 5062
and references therein.
For selected recent reviews on transition-metal-catalyzed C-H
functionalization, see: a) R. G. Bergman, Science 1984, 223, 902; b) J.
A. Labinger, J. E. Bercaw, Nature 2002, 417, 507; c) D. A. Colby, R. G.
Bergman, J. A. Ellman, Chem. Rev. 2010, 110, 624; d) T. W. Lyons, M.
S. Sanford, Chem. Rev. 2010, 110, 1147; e) K. M. Engle, T.-S. Mei, M.
Wasa, J.-Q. Yu, Acc. Chem. Res. 2012, 45, 788; f) J. Yamaguchi, A. D.
Yamaguchi, K. Itami, Angew. Chem. Int. Ed. 2012, 51, 8960; Angew.
Chem. 2012, 124, 9092; g) J. Wencel-Delord, F. Glorius, Nat. Chem.
2013, 5, 369; h) G. Rouquet, N. Chatani, Angew. Chem. Int. Ed. 2013,
52, 11726; Angew. Chem. 2013, 125, 11942; i) K. Shin, H. Kim, S. Chang,
Acc. Chem. Res. 2015, 48, 1040; j) O. Daugulis, J. Roane, L. D. Tran,
Acc. Chem. Res. 2015, 48, 1053; k) J. He, M. Wasa, K. S. L. Chan, Q.
Shao, J.-Q. Yu, Chem. Rev. 2017, 117, 8754.
a) W. M. C. Sameera, F. Maseras, WIREs Comput. Mol. Sci. 2012, 2,
375; b) G.-M. Max, A. A. C. Braga, A. Lledós, G. Ujaque, F. Maseras,
Acc. Chem. Res. 2013, 46, 2626; c) J. Jover, N. Fey, Chem. Asian J.
2014, 9, 1714; d) T. Sperger, I. A. Sanhueza, I. Kalvet, F. Schoenebeck,
Chem. Rev. 2015, 115, 9532; e) T. Sperger, I. A. Sanhueza, F.
Schoenebeck, Acc. Chem. Res. 2016, 49, 1311.
a) C. L. McMullin, J. Jover, J. N. Harvey, N. Fey, Dalton Trans. 2010, 39,
10833; b) Q. Peng, F. Duarte, R. S. Paton, Chem. Soc. Rev. 2016, 45,
6093.
a) V. V. Grushin, W. J. Marshall, J. Am. Chem. Soc. 2006, 128, 4632; b)
V. V. Grushin, W. J. Marshall, J. Am. Chem. Soc. 2006, 128, 12644; c)
E. J. Cho, T. D. Senecal, T. Kinzel, Y. Zhang, D. A. Watson, S. L.
Buchwald, Science 2010, 328, 1679; d) P. Anstaett, F. Schoenebeck,
Chem. Eur. J. 2011, 17, 12340; e) V. I. Bakhmutov, F. Bozoglian, K.
Gómez, G. González, V. V. Grushin, S. A. Macgregor, E. Martin, F. M.
Miloserdov, M. A. Novikov, J. A. Panetier, L. V. Romashov,
Organometallics 2012, 31, 1315; f) M. C. Nielsen, K. J. Bonney, F.
Schoenebeck, Angew. Chem. 2014, 23, 6013; Angew. Chem. Int. Ed.
2014, 53, 5903; g) K. Natte, R. V. Jagadeesh, L. He, J. Rabeah, J. Chen,
C. Taeschler, S. Ellinger, F. Zaragoza, H. Neumann, A. Brückner, M.
Beller, Angew. Chem. 2016, 128, 2832; Angew. Chem. Int. Ed. 2016, 55,
2782; h) J. del Pozo, E. Gioria, P. Espinet, Organometallics 2017, 36,
2884; i) D. M. Ferguson, J. R. Bour, A. J. Canty, J. W. Kampf, M. S.
Sanford, J. Am. Chem. Soc. 2017, 139, 11662; j) S. T. Keaveney, F.
Schoenebeck, Angew. Chem. 2018, 130, 4137; Angew. Chem. Int. Ed.
2018, 57, 4073; (k) M. Pu, I. Sanhueza, E. Senol, F. Schoenebeck,
Angew. Chem. 2018, 130, 15301; Angew. Chem. Int. Ed. 2018, 57,
15081; (l) S. Martínez de Salinas, Á. L. Mudarra, J. Benet-Buchholz, T.
Parella, F. Maseras, M. H. Pérez-Temprano, Chem. Eur. J. 2018, 24,
11895; (m) D. M. Ferguson, J. R. Bour, A. J. Canty, J. W. Kampf, M. S.
Sanford, Organometallics 2019, 38, 519.

[9]
[10]
[11]

[12]
[13]
[14]

[15]
[16]
[17]
[18]
[19]
[20]
[21]

[22]
[23]
[24]

[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]

A. A. Thomas, K. Speck, I. Kevlishvili, Z. Lu, P. Liu, S. L. Buchwald, J.
Am. Chem. Soc. 2018, 140, 13976.
P. M. Dewick, Medicinal natural products: a biosynthetic approach, 3rd
ed.; Wiley: Chichester, West Sussex, U.K.; New York, NY, USA, 2009.
a) S. Y. Hong, Y. Park, Y. Hwang, Y. B. Kim, M.-H. Baik, S. Chang,
Science 2018, 359, 1016; b) S. Bräse, C. Gil, K. Knepper, V.
Zimmermann, Angew. Chem. 2015, 117, 5320; Angew. Chem. Int. Ed.
2015, 44, 5188; c) H. Lebel, O. Leogane, Org. Lett. 2005, 7, 4107; d) D.
Li, T. Wu, K. Liang, C. Xia, Org. Lett. 2016, 18, 2228.
Y. Park, S. Chang, Nat. Catal. DOI: 10.1038/s41929-019-0230-x.
K. D. Vogiatzis, M. V. Polynski, J. K. Kirkland, J. Townsend, A. Hashemi,
C. Liu, E. A. Pidko, Chem. Rev. 2019, 119, 2453.
a) M. Urbano-Cuadrado, J. J. Carbó, A. G. Maldonado, C. Bo, J. Chem.
Inf. Model. 2007, 47, 2228; b) A. G. Maldonado, G. Rothenberg, Chem.
Soc. Rev. 2010, 39, 1891; c) C. B. Santiago, J.-Y. Gao, M. Sigman,
Chem. Sci. 2018, 9, 2398; d) D. J. Durand, N. Fey, Chem. Rev.
DOI:10.1021/acs.chemrev.8b00588.
N. Miyaura, A. Suzuki, J. Chem. Soc. Chem. Commun. 1979, 866.
Z. L. Niemeyer, A. Milo, D. P. Hickey, M. S. Sigman, Nat. Chem. 2016,
8, 610.
A. F., Littke, C. Dai, G. C. Fu, J. Am. Chem. Soc. 2000, 122, 4020.
F. Schoenebeck, K. N. Houk, J. Am. Chem. Soc. 2010, 132, 2496.
S. Zhao, T. Gensch, B. Murray, Z. L. Niemeyer, M. S. Sigman, M. R.
Biscoe, Science 2018, 362, 670.
K. Wu, A. G. Doyle, Nat. Chem. 2017, 9, 779.
T. Piou, F. Romanov-Michailidis, M.Romanova-Michaelides, K. E.
Jackson, N. Semakul, T. D. Taggart, B. S. Newell, C. D. Rithner, R. S.
Paton, T. Rovis, J. Am. Chem. Soc. 2017, 139, 1296.
T. Piou, F. Romanov-Michailidis, M. A. Ashley, M. RomanovaMichaelides, T. Rovis, J. Am. Chem. Soc. 2018, 140, 9587.
Y. Kim, Y. Park, S. Chang, ACS Cent. Sci. 2018, 4, 768.
For selected reviews on difluoromethylation, see: a) J. Rong, C. Ni, J.
Hu, Asian J. Org. Chem. 2017, 6, 139; b) Z. Feng, Y.-L. Xiao, X. Zhang,
Acc. Chem. Res. 2018, 51, 2264. For selected reviews on
trifluoromethylation, see: c) T. Furuya, A. S. Kamlet, T. Ritter, Nature
2011, 473, 470; d) O. A. Tomashenko, V. V. Grushin, Chem. Rev. 2011,
111, 4475; e) A. Studer, Angew. Chem. 2012, 124, 9082; Angew. Chem.
Int. Ed. 2012, 51, 8950; f) Y. Ye, M. S. Sanford, Synlett 2012, 23, 2005;
g) H. Wang, D. A. Vicic, Synlett 2013, 24, 1887; h) J. Charpentier, N.
Früh, A. Togni, Chem. Rev. 2015, 115, 650; i) X. Liu, C. Xu, M. Wang,
Q. Liu, Chem. Rev. 2015, 115, 683; j) X. Yang, T. Wu, R. J. Phipps, F.
D. Toste, Chem. Rev. 2015, 115, 826.
a) Y. Gu, X. Leng, Q. Shen, Nat. Commun. 2014, 5, 5405; b) C. Lu, Y.
Gu, J. Wu, Y. Gu, Q. Shen, Chem. Sci. 2017, 8, 4848.
C. A. Malapit, J. R. Bour, C. E. Brigham, M. S. Sanford, Nature 2018,
563, 100.
K. Shin, Y. Park, M.-H. Baik, S. Chang, Nat. Chem. 2018, 10, 218.
K. Shin, S.-W. S. Chang, J. Am. Chem. Soc. 2015, 137, 8584.
J. Kim, K. Shin, S. Jin, D. Kim, S. Chang, J. Am. Chem. Soc.
DOI:10.1021/jacs.9b00364.
M. Simonetti, D. M. Cannas, X. Justo-Baringo, I. J. Vitorica-Yrezabal, I.
Larrosa, Nat. Chem. 2018, 10, 724.
S. Oi, S. Fukita, N. Hirata, N. Watanuki, S. Miyano, Y. Inoue, Org. Lett.
2001, 3, 2579.
E. F. Flegeau, C. Bruneau, P. H. Dixneuf, A. Jutand, J. Am. Chem. Soc.
2011, 133, 10161.
M. Simonetti, G. J. P. Perry, X. C. Cambeiro, F. Juliá-Hernández, J. N.
Arokianathar, I. Larrosa, J. Am. Chem. Soc. 2016, 138, 3596.
a) M. Moselage, J. Li, L. Ackermann. ACS Catal. 2016, 6, 498; b) D. Wei,
X. Zhu, J.-L. Niu, M.-P. Song, ChemCatChem 2016, 8, 1242; c) P. G.
Chirila, C. J. Whiteoak, Dalton Trans. 2017, 46, 9721; d) S. Wang, S.-Y.
Chen, X.-Q. Yu, Chem. Commun. 2017, 53, 3165; e) T. Yoshino, S.
Matsunaga, Adv. Synth. Catal. 2017, 359, 1245; f) J. Ghorai, P.
Anbarasan, Asian J. Org. Chem. DOI:10.1002/ajoc.201800452.

[35] a) J. Sanjosé-Orduna, D. Gallego, A. Garcia-Roca, E. Martin, J. BenetBuchholz, M. H. Pérez-Temprano, Angew. Chem. 2017, 129, 12305;
Angew. Chem. Int. Ed. 2017, 56, 12137; b) J. Sanjosé-Orduna, J. M. S.
Toro, M. H. Pérez-Temprano, Angew. Chem. 2018, 130, 11539; Angew.
Chem. Int. Ed. 2018, 57, 11369.
[36] a) N. Fey, Chem. Cent. J. 2015, 9, 1; b) N. Fey, A. C. Tsipis, S. E. Harris,
J. N. Harvey, A. G. Orpen, R. A. Mansson, Chem. Eur. J. 2006, 12, 291;
c) O. Lakuntza, M. Besora, F. Maseras, Inorg. Chem. 2018, 57, 14660.

Entry for the Table of Contents (Please choose one layout)
Layout 1:

MINIREVIEW
The rational design of innovative and
efficient chemical transformations
based on fundamental knowledge
constitutes one of the most promising
sustainable
strategies
for
the
synthesis of organic scaffolds. Herein,
we show the potential of theoretical
and/or experimental knowledgedriven approaches for the sustainable
development of transition metalcatalyzed transformations.

J. Sanjosé-Orduna, Á. L. Mudarra, S.
Martínez de Salinas, M. H. PérezTemprano*
Page No. – Page No.
Sustainable knowledge-driven
approaches in transition metalcatalyzed transformations