Download as pdf or txt
Download as pdf or txt
You are on page 1of 65

Review

Cite This: Chem. Rev. 2019, 119, 8781−8845 pubs.acs.org/CR

Chemical Bonding and Bonding Models of Main-Group Compounds


Lili Zhao,† Sudip Pan,† Nicole Holzmann,‡ Peter Schwerdtfeger,*,§ and Gernot Frenking*,†,∥,⊥

Institute of Advanced Synthesis, School of Chemistry and Molecular Engineering, Jiangsu National Synergetic Innovation Center
for Advanced Materials, Nanjing Tech University, Nanjing 211816, China

Scientific Computing Department, STFC Rutherford Appleton Laboratory, Harwell Oxford, Didcot OX11 0QX, United Kingdom
§
The New Zealand Institute for Advanced Study, Massey University (Albany), 0632 Auckland, New Zealand

Fachbereich Chemie, Philipps-Universität Marburg, Hans-Meerwein-Strasse, D-35043 Marburg, Germany

Donostia International Physics Center (DIPC), P.K. 1072, 20080 Donostia, Euskadi, Spain
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
Downloaded via UNIV FED DO ESPIRITO SANTO on November 13, 2020 at 22:55:10 (UTC).

ABSTRACT: The focus of this review is the presentation of the most important aspects
of chemical bonding in molecules of the main group atoms according to the current state
of knowledge. Special attention is given to the difference between the physical
mechanism of covalent bond formation and its description with chemical bonding
models, which are often confused. This is partly due to historical reasons, since until the
development of quantum theory there was no physical basis for understanding the
chemical bond. In the absence of such a basis, chemists developed heuristic models that
proved extremely valuable for understanding and predicting experimental studies. The
great success of these simple models and the associated rules led to the fact that the
model conceptions were regarded as real images of physical reality. The complicated
world of quantum theory, which eludes human imagination, made it difficult to link
heuristic models of chemical bonding with quantum chemical knowledge. In the early
days of quantum chemistry, some suggestions were made which have since proved untenable. In recent decades, there has been
a stormy development of quantum chemical methods, which are not limited to the quantitative accuracy of the calculated
properties. Also, methods have been developed where the experimentally developed models can be quantitatively expressed and
visually represented using mathematically well-defined terms that are derived from quantum chemical calculations. The
calculated numbers may however not be measurable values. Nevertheless, as orientation data for the interpretation and
classification of experimental findings as well as a guideline for new experiments, they form a coordinate system that defines the
multidimensional world of chemistry, which corresponds to the Hilbert space formalism of physics. The nonmeasurability of
model values is not a weakness of chemistry but a characteristic by which the infinite complexity of the material world becomes
scientifically accessible and very useful for chemical research. This review examines the basis of the commonly used quantum
chemical methods for calculating molecules and for analyzing their electronic structure. The bonding situation in selected
representative molecules of main-group atoms is discussed. The results are compared with textbook knowledge of common
chemistry.

CONTENTS 6. Physical Reality and Chemical Bonding Models 8794


7. Selected Examples of Chemical Bonds in Main-
1. Introduction 8782 Group Compounds 8795
2. The Physical Nature of the Chemical Bond 8783 7.1. H2+, H2, Li2+, Li2 8795
3. Historical Development and Present Situation of 7.2. H2, N2, CO, and BF 8797
Bonding Models for Main-Group Compounds: 7.3. First Octal-Row Sweep Li2−F2 and Related
The Lewis Paradigm 8786 Molecules 8799
4. Quantum Chemical Methods for Calculating 7.3.1. Li2, Be2 and Related Molecules 8801
Molecular Structures and Properties 8787 7.3.2. B2 and Related Molecules 8801
4.1. Molecular Orbital (MO) Theory 8788 7.3.3. C2 and Related Molecules 8803
4.2. Density Functional Theory (DFT) 8789 7.3.4. N2 and Related Molecules 8806
4.3. Valence-Bond (VB) Theory 8790 7.3.5. O2, F2 and Related Molecules 8808
5. Quantum Chemical Methods for Analyzing the 7.4. Chemical Bonding in Heavier Main-Group
Chemical Bond in Molecules 8790 Compounds 8810
5.1. Natural Bond Orbital Method (NBO) 8790
5.2. Quantum Theory of Atoms in Molecules
(QTAIM) 8791 Special Issue: Frontiers in Main Group Chemistry
5.3. Energy Decomposition Analysis and Natural Received: December 2, 2018
Orbitals for Chemical Valence (EDA-NOCV) 8792 Published: June 28, 2019

© 2019 American Chemical Society 8781 DOI: 10.1021/acs.chemrev.8b00722


Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

7.4.1. Hypervalent Molecules: Valence, Orbital the term “covalent bonding” at a time when modern quantum
Symmetry, and 3-Center, 4-Electron theory was not developed yet. Over the course of time it was
Bonding 8811 modified and refined like a neural network, using the increase of
7.4.2. Hypervalent Molecules: SF6 and PF5 8812 experimental observations as feedstock for improvement.
7.4.3. Multiple Bonds of Heavy Diatomic The electron-pair model of a covalent bond may be
Molecules Na2−Cl2 and N2−Bi2 and considered as a stroke of genius, where the yet unknown
Related Molecules 8813 quantum chemical nature of the chemical bond is miraculously
7.4.4. Multiple Bonds of Heavy Main-Group converted to and hidden in a very simple representation. It is
Atoms: N2/N4 vs P2/P4 8815 unsurpassed in its convenience to depict and represent
7.4.5. Multiple Bonds of Heavy Main-Group molecular structures and reactivities. The model has been
Atoms: X2CCX2 vs X2EEX2 (E = Si − Pb; X = somewhat modified over the years and the rules for using Lewis
H, F) 8817 structures slightly changed in the course of time, but the
7.4.6. Multiple Bonds of Heavy Main-Group essential features remain the same until today. This is
Atoms: HCCH vs HEEH (E = Si − Pb) 8819 astonishing, because the origin of covalent bonding is a quantum
8. Dative Bonding in Main-Group Compounds 8822 theoretical effect. This was shown for the first time by Heitler
9. The Octet Rule and the Atomic Valence Space of and London in 1927,8 when they analyzed the chemical bond in
Main-Group Atoms 8827 H2 using the newly developed quantum theory by Heisenberg9
10. Relativistic Effects in Main-Group Compounds 8829 and Schrödinger.10 Linus Pauling was the first who presented a
11. Concluding Remarks 8834 bridge between the heuristic electron-pair bonding model of
Author Information 8835 Lewis and quantum theory in his epochal book “The Nature of
Corresponding Authors 8835 the Chemical Bond”, which quickly became the reference
ORCID 8835 standard in the community for describing chemical bonding.11
Notes 8835 This was done at a time when powerful computers were not yet
Biographies 8835 available and quantum chemistry was in its infant stage. Later
Acknowledgments 8835 studies showed that the nature of the chemical bond is far more
References 8835 complicated than initially thought and that the connection
between the Lewis model and the physical nature of chemical
bonding is quite intricate. This does not come as a surprise.
1. INTRODUCTION Considering the enormous complexity of the physical universe
In the year 2000, one of the present authors wrote a review for and the rather small numbers of chemical elements and
the centennial issue of this journal entitled “The Nature of the particles12 from which it is constructed, it requires a sheer
Bonding in Transition-Metal Compounds”.1 The introduction infinite variety of chemical bonding to create such a richness of
describes the peculiar dichotomy of using heuristic bonding the material phenomena in our universe including the complex
models along with quantum chemical calculations for the biochemical nature of animate beings.
description of molecular structures, which is characteristic for A fundamental approach of chemistry deals with the
present-day chemical research. This situation holds even more classification and arrangement of the molecular structures and
for main-group compounds, where the depiction of the bonding reactivities using simple models, rules, and guidelines, which
situation in terms of “resonating” Lewis structures and molecular serve as classification schemes for the vast amount of
orbitals (MOs) in combination with quantum chemical compounds and reactions. Chemistry as it is currently applied
calculations along with charge- and energy-partitioning methods and understood can be considered as the science of fuzzy
is frequently presented side-by-side. The connection between concepts, because many models are poorly defined and
the experimental observations and numerical results of such somewhat arbitrarily used and, yet, they have proven to be
calculations is usually guided by the attempt to grasp the very helpful as an orientation for experimental studies. Chemical
complicate numerical information, which is provided by models were termed as “unicorns”,13 because every chemist
quantum chemical calculations, with a simple bonding model. seems to know what they mean although a precise definition is
This can clearly be considered a Hercules (and sometimes often not available. For a scientist who is trained with exact
Sisyphus) task, which tries to connect two extremes: The physical quantities, chemical models may appear as unscientific
quantum theoretical description of the electronic structure, chimera. But if we only would focus on measurable quantities
which is governed by the Schrö dinger equation in the (so-called observables in quantum mechanical terms), dis-
nonrelativistic case and the Dirac equation in the relativistic regarding the many beautiful concepts developed over the past
case, is given in mathematical terms that are elusive objects for centuries, chemistry would be deprived of its inner beauty and
the human imagination trained with classical objects. It is a many major discoveries would not have been made. The term
genuine chemical approach to illustrate this with symbols and “chemical bond” is not an observable in a strict quantum
figures that are as simple as possible, appealing, and useful to mechanical sense but refers to the extensive symbolism used in
chemists. chemistry to represent the attractive force between atoms or
The most important model for chemical bonding is clearly the molecules. A puristic viewpoint ignores their enormous
heuristic assignment of an electron pair to a chemical bond first usefulness for breaking down the ever increasing complexity
suggested by Gilbert Lewis in 1916.2 It was developed by and richness of chemical findings into well-ordered classification
inspection of chemical structures and stabilities and by schemes. Chemical bonding models are not right or wrong, they are
evaluating similarities mainly of main-group compounds of the more or less usef ul. The usefulness of models comes from their
first octal row of the periodic system. The model was further help as an ordering scheme of experimental observations and
elaborated by Langmuir in 1919−1921,3−6 who introduced the their use as guidance for future experiments and interpretations.
octet rule7 along with the 18- and 32-electron rules and coined But it is a grave mistake to confuse a bonding model with the
8782 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

physical nature of the interatomic interactions that give rise to wave functions, which may either have a positive (+) or a
what we call a chemical bond. This has been the source of negative (−) sign. The alternative signs yield two possible
numerous controversies and misunderstandings in the past. options for the new wave function, which describe the bonding
This review article has two goals. One goal is to present the and antibonding combinations of the interacting electrons. The
current understanding of the physical nature of covalent positive interference leads to an accumulation of electronic
bonding in main-group compounds. This shall be done without charge in the interatomic region whereas the negative
digging deeply into the mathematical details of the quantum interference induces charge depletion. It is only at the level of
chemical description of the interatomic interactions. This is not the wave function that the information, which is provided by the
an easy task, because the quantum theoretical description of the leading sign, becomes operative. The electronic charge is derived
chemical bond is quite complicated and has many facets, which from squaring the wave function, which annihilates the sign
cannot all be accounted for in the present review article. We information given by the wave function. Mulliken,17 Hund,18
strive for a concise presentation of the most important results Lennard-Jones,19 and Hü ckel20 recognized the gain in
and conclusions that came out of the numerous studies, which information that is provided by the sign of the wave functions.
were done since the ground-breaking work by Heitler and Many years later Fukui21 developed the frontier molecular
London. For a more detailed discussion of the quantum orbital (FMO) theory and Woodward and Hoffmann22
theoretical nature of the chemical bond, we refer to pertinent introduced the orbital symmetry rules for pericyclic reactions,
review articles and monographs.14−16 which are based on the information that is provided by the
The second goal of this account concerns the use of modern orbitals. Nevertheless, chemists still have difficulties in accepting
chemical bonding models and the Lewis electron-pair the more fundamental nature of the wave function as the origin
approximation for describing molecular structures and reac- of the chemical bond compared to a simple particle character of
tivities. The central aspect of this work aims at bridging bonding the electrons.
models with the quantum theoretical nature of the chemical The most important conclusion that stems from the Heitler−
bond. The focus is the connection between the results of London work is the finding that the physical origin of the
standard quantum chemical calculations, which are mostly based covalent bond is not due to the formation of an electron pair.
on molecular orbital (MO) theory or density functional theory Covalent bonding comes from the interference of the wave
(DFT), and the presentation of molecules with a bonding functions of the interacting species.23 Chemical bonding occurs
model. There are various methods available that convert the already when only one electron occupies the new wave function,
results of theoretical calculations into models that aim at such as in H2+. The second electron usually strengthens the
describing the essential chemical features of the investigated chemical bond further such as in H2, but there are molecules
species. The most important methods will be critically discussed, where the second electron even weakens the bond. This is the
and the results for archetypical molecules will be presented. It is case in Li2, which has lower bond dissociation energy than
not possible to review all kinds of chemical bonds of main-group Li2+.24 The frequent occurrence of molecules with an even
compounds in this work. The present account is restricted to number of electrons is due to the Pauli principle, which allows a
selected classes of archetypical molecules and bonds, which maximum of two electrons occupying the same spatial orbital.
represent not all but the most important types of chemical bonds This is a quantum theoretical postulate originating from the
in molecules of the sp-block elements of the periodic table. more fundamental spin statistics theorem, which requests that
electrons are different by at least one quantum number, but it is
2. THE PHYSICAL NATURE OF THE CHEMICAL BOND not the physical origin per se of the chemical bond. The finding
The physical origin of the covalent bond was for a long time a that electron pairing is not the physical origin of the chemical
mystery not only for chemists but for science in general. It was bond becomes obvious by the stability of numerous molecules,
clear that from the four elementary forces in physics only which have unpaired electrons such as dioxygen O2 and the large
electrostatic forces could account for the strong interatomic number of paramagnetic transition metal complexes. The
interactions that yield a chemical bond. But the attraction takes specific e−e pairing interaction is repulsive in most cases, and
place between neutral species, and the application of the classical it can be energetically more favorable that the two electrons
Coulomb law does not give results that agree with experimental occupy spatially different orbitals, as for example in the ground
findings. Gilbert Lewis knew about the problem when he state of O2. The stabilization of the second electron in a doubly
suggested in 1916 that the chemical bond should be identified occupied MO stems mainly from the interference of the wave
with an electron pair. He speculated, “Electric forces between function.
particles which are very close together do not obey the simple law of We wish to emphasize that Pauli repulsion does not introduce
inverse squares which holds at greater distances”.2 When the puzzle a new type of physical force, which may be overcome by other
about the physical origin of the chemical bond was finally solved forces. All interatomic forces, which are relevant for chemical
by Heitler and London in 1927,8 who employed the newly bonding, come from Coulombic interactions. However, the
developed quantum theory by Heisenberg9 and Schrödinger10 mathematical description of Coulomb forces between atoms
to the interactions between two hydrogen atoms as well as requires quantum theoretical postulates. The Pauli (exchange)
between two helium atoms, it came as a shock not only to forces are due to the requirement that the wave function must be
chemists but to natural sciences altogether. The solution was in antisymmetric, which requests that two electron may not have
conflict with all ideas that had been envisaged by scientists for the same quantum numbers.
explaining the chemical bond. The large difference between the classical and the quantum
According to the quantum theoretical approach by Heitler theoretical description of an electron comes to the fore when
and London, the electrostatic interactions between atoms have calculating the electron−electron repulsion as a function of the
to be described in terms of mathematical functions describing e−e distance either way. Figure 1 shows the 1/r12 curve for the
quantum objects rather than classical particles. Chemical repulsion between two point charges q calculated following the
bonding comes from the interference of spatially extended Coulomb law for classical particles (eq 1) and the repulsion
8783 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Figure 1. Calculated interaction energies between two charged particles Figure 2. Copy of the figure by Heitler and London from their 1927
as a function of their distance r12. Repulsive interactions between two paper which shows the potential energy curves of dihydrogen calculated
electrons calculated classically ΔEelstat(classical) = q1q2/r12 (dashed classically (E11) and using the wave function Ψ (Eα and Eβ). Reprinted
line); quasiclassical repulsion between two electrons in 1s orbitals with permission from ref 8. Copyright 1927 Springer Nature.
ΔEelstat = ∫ ρ1ρ2/r12 dτ1 dτ2 (dotted line); exchange repulsion between
two electrons with the same spin in 1s orbitals ΔEPauli (solid line).
Reprinted with permission from ref 28. Copyright 2006 John Wiley and Figure 2 shows the original curves by Heitler and London8 for
Sons. the interaction between two hydrogen atoms. The energy curve
E11, which is based on the quasi-classical Coulomb interactions
between the electronic charges ρ of two electrons with opposite between the superimposed atomic charge distributions, has only
spin in H2 obtained by the square of the wave function Ψ (eqs a weak energy minimum of ∼0.5 eV at a rather long distance of
2a): ∼1.7 atomic units, while the curve Eα for the attractive
interactions that are calculated using the wave functions has a
ΔEelstat(classical) = q1q2 /r12 (1) much deeper potential of ∼2.5 eV at a more reasonable bond
distance of 1.5 atomic units. The curve Eβ refers to the repulsive
ρ = |Ψ|2 (2a) interactions between two hydrogen atoms where the electrons
have the same spin. The occurrence of two energy curves for the
ΔEelstat(qc) = ∫ dτ1dτ2ρ1ρ2/r12 (2b)
interactions between the hydrogen atoms is a quantum
theoretical phenomenon that cannot be explained by classical
The two curves are nearly indistinguishable at distances >2 Å physics. This must be kept in mind when bonding models like
where the wave function overlap is negligible. At shorter the Lewis electron pair, which are the results of human
distances the electrostatic repulsion ΔEelstat(qc) (qc = quasi interpretation of experiment and thus based on classical objects,
classical) given by eq 2b is much less than the classical repulsion are used to explain chemical findings.
between two point charges and approaches a finite value at r = 0 The essential finding of the Heitler−London work, which
whereas the classical eq 1 leads to an infinite Coulomb repulsion. features the very core of covalent bonding, might be presented in
The third curve in Figure 1 depicts the Pauli (exchange) a simplified way. We skip some mathematical details, in order to
repulsion between two electrons having the same spin. It extract the central message. Scheme 1 shows the quasiclassical
becomes obvious that the onset of the Pauli repulsion, approach (eqs 3−5) for chemical bonding in H2, which employs
depending on the overlap of the wave functions, takes place at the electronic charge distribution ρ(H) of the hydrogen atoms
shorter distances but has a much steeper slope than the as starting point for the interatomic interactions. The calculation
Coulomb repulsion. The Pauli repulsion is the physical origin of using Coulomb’s law results in E11, which has only a shallow
the loosely defined concept of steric repulsion between bulky energy minimum (Figure 2). The bottom part of Scheme 1
substituents. It also provides a quantum theoretical basis for the delineates the quantum theoretical approach, where the wave
VSEPR (Valence Shell Electron-Pair Repulsion) model of function Ψ(H) is used as elementary physical quantity rather
Nyholm and Gillespie.25−27 The total Coulombic interactions than ρ(H) for the description of the electron. Since ρ(H) is the
between neutral species are usually attractive, because of the square29 of Ψ(H) (eq 6), both ±Ψ(H) may be used for the
electrostatic attraction between atomic nuclei and electrons.28 generation of the wave function Ψ(H2) (eq 7) in a simple one-
This is discussed in more detail below. particle model. The electronic charge distribution ρ(H2) is then
8784 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Scheme 1. Schematic Description of the Interatomic alteration of E, V, and T for H2 as a function of the H−H
Interactions between Two Hydrogen Atoms Using (a) a distance.
Classical Approach and (b) a Quantum Theoretical
Approach

Figure 3. Energy curves of the total energy E, potential energy V, and


kinetic energy T as a function of the H−H distance. Reprinted with
permission from ref 42. Copyright 2019 Springer Nature.

When the hydrogen atoms approach each other, the kinetic


energy along with the total energy is at first lowered whereas the
potential energy rises. This is because the electronic charge
experiences a larger volume in the overlapping valence space.
The potential energy rises, because the “electron distance” to the
nuclei is longer than in the atoms. This is the region where the
actual bond formation mainly takes place. At shorter distances,
the effect of electronic charge depletion at the nuclei dominates
leading to shrinkage of the effective atomic radii. As a result, the
given by squaring |Ψ(H2)|2 (eq 8). The associated binomial in eq depleted charge in the core region experiences a stronger
9 gives the term ±2[c1c2 Ψ(Ha)Ψ(Hb)], which describes the attraction by the nucleus, which lowers V and increases T. The
interference of the atomic wave functions that leads to attraction latter process eventually overcompensates the reverse change of
(+ sign) or repulsion (− sign). The respective attractive and T and V in the bonding region and yields an overall increase of
repulsive energy terms are shown in eqs 10a and 10b. This result kinetic energy and lowering of potential energy at the
comes from a quantum theoretical description of the interatomic equilibrium distance. The conclusion is that the driving force
interactions leading to the curves Eα and Eβ in Figure 2. The for the accumulation of electronic charge in the bonding region
interference term Ψ(Ha)Ψ(Hb) is a pure quantum theoretical associated with the formation of the chemical bond is the
expression, which does not have a classical analogue. lowering of the kinetic energy. This was shown for the first time
Chemical bonding is therefore a quantum chemical effect and by Hellmann.30 After some controversy,14 it was finally
can be seen as caused by the interference of atomic/fragmental confirmed by Ruedenberg.31−36
wave functions. Chemical bonding is due to electron-sharing but not The finding that the origin of covalent bonding is due to the
due to electron pairing. There is a second aspect of covalent interference of the wave functions of the bonding atoms is not
bonding that is often misstated in chemistry textbooks and refers restricted to dihydrogen, and it is not confined to nonpolar
to the contribution of kinetic and potential energies to the bonds. It will be shown below that H2 is in many ways untypical
energy lowering due to the formation of a chemical bond. for covalent bonding, but the physical origin for covalence is the
According to the virial theorem at equilibrium distances E = 0.5 same for all atoms. Ruedenberg and others have demonstrated
V = −T, it follows that the change in the total energy E has the this in detailed theoretical studies.31−34 Electrostatic attraction,
same sign as the potential energy V, whereas the kinetic energy T exchange (Pauli) repulsion, and further factors contribute to the
has the reverse sign. For the bond energy ΔEb thus holds ΔEb = intricate combination that yields a chemical bond, but the
0.5ΔVb = −ΔTb. This agrees with the popular statement that the interference of the wave function remains at the origin of
bonding electrons, due to charge accumulation in the covalent bonding.
interatomic region, are energetically stabilized, because they The following conclusions arise from this section:
are attracted by two nuclei. A close examination of the change in • The physical origin of the covalent bond between atoms is
the electronic structure of two atoms along the bond path reveals the interference of the atomic wave functions but not the
that the formation of the chemical bond is rather complex and formation of an electron pair. Chemical bonding comes
exhibits a paradoxical behavior concerning the contribution of f rom electron-sharing but not f rom electron-pairing between
the kinetic energy T and potential energy V. Figure 3 shows the atoms.
8785 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

• The driving force for the formation of a chemical bond is electrostatic interaction between charged particles providing the
the lowering of the kinetic energy of the accumulated mathematical framework for the puzzling behavior of matter at
charge in the bonding region. an atomic scale were introduced ten years later by Heisenberg9
• The strength of the e−e repulsion as a function of the and Schrödinger.10
distance exhibits a drastically different behavior when The suggestions of Lewis were picked up by Langmuir, who
electrons are considered as point charges compared to elaborated on the electron-pair model in a series of four papers
orbitals. The Pauli repulsion between electrons that from 1919−1921.3−6 Langmuir coined the term “covalence” for
possess the same spin is much stronger than Coulombic the chemical bond, and he formulated the octet rule for main-
repulsion in regions where the orbital overlap becomes group atoms along with the related 18-electron rule for
important. transition metals and the 32-electron rule for lanthanides. The
latter rules were later rationalized by quantum theory with the
• The frequent appearance of an electron pair in chemical occupation of the valence shells of the atoms. In 1923, Lewis
bonds is due to the Pauli principle. It may be reasonably published the book Valence and the Structure of Atoms and
used as a model for depicting molecular structures and Molecules, where he extended his viewpoint on the electron-pair
chemical bonds (and particularly useful in explaining bonding model.39 The book was partly written as a response to
organic reaction mechanisms), but it must not be the Langmuir papers, which received much attention and led
confused with the physical origin of the chemical bond. some chemists to articulate the “Langmuir-Lewis model”.40
Lewis stressed the importance of the electron pairing for
3. HISTORICAL DEVELOPMENT AND PRESENT chemical bonding, saying “Whether we are dealing with organic or
SITUATION OF BONDING MODELS FOR inorganic compounds, the chemical bond is always such a pair of
MAIN-GROUP COMPOUNDS: THE LEWIS chemical bonds.”39 The dogmatic adherence to the electron pair
PARADIGM as the origin of the covalent bond and a deeply rooted aversion
The development of theoretical concepts and models to against quantum theory, which was at that time in an embryonic
understand the physical world is a fascinating chapter in the state, may have contributed to his unwillingness to open his
history of mankind.37 Chemistry played a prominent role in the mind for the upcoming important developments. In his book, he
endeavor to explain observations in nature and in experiments called quantum theory “the entering wedge of scientific bolshevism”
and to formulate conclusions that may be used for future but then he concedes “Quantum theory has been criticized for
predictions. Based on the corpuscular theory of matter, which f urnishing no adequate mechanism, but presumably the root of our
was not undisputed until the early 20th century, and following present problem lies deeper than this, and it is hardly likely that any
the introduction of the periodic system of elements by mechanism based on our existing modes of thought will suf f ice for
Mendeleev and Meyer, chemists designed symbols and formulas the explanation of the many new phenomena which the study of the
as synonyms for characterizing substances that are composed of atom is disclosing”.39 It appears that Lewis was neither willing nor
atoms. The Scottish chemist Archibald Couper was the first who prepared to accept the findings of quantum theory, which finally
in 1858 suggested a dash connecting two atoms as a symbol for a came in 1927. His latest study solely devoted to the chemical
chemical bond without any physical meaning assigned to it.38 bond appeared in 1933,41 six years after the Heitler−London
The 19th century witnessed numerous suggestions for paper about the quantum chemical nature of covalent bonding.
explaining the ever-increasing number of experimental observa- The very long paper, which does not have a single reference,
tions in terms of bonding models. The wealth of information reads like a defense of the electron-pair model in spite of its
resulting from chemical research was used as source for a neural apparent deficiencies for describing some molecules. It seems,
network to produce helpful concepts that were improved or however, that Gilbert Lewis himself recognized the limitations of
dismissed by experimental results. Chemical research was his approach. In the final statement of the 1923 book, he pointed
additionally fuelled by the fact that its products became the out that his model may have to be modified in the future and that
foundation of the chemical industry that arose as an important he wants “...to emphasize the necessity of maintaining an opening of
new commercial branch. But the concepts and models were mind; so that, when the solution of these problems, which now seem
considered to have mere symbolic meaning rather than physical so baff ling, is ultimately of fered, its acceptance will not be retarded
relevance. Three-dimensional structures were only later assigned by the conventions and the inadequate mental abstractions of the
to the chemical formula, most prominently by van’t Hoff, who past”.39 His request, which was perhaps inspired by recognizing
established together with Le Bel the tetrahedral model for his own limitations, has not lost its timeliness until today.
carbon compounds to explain the occurrence of enantiomers An important contribution to the manifestation of the
and the phenomenon of optical activity. However, the question electron-pair bond as fundamental model for chemical bonding
about the physical foundation behind these models could not be is the work of Linus Pauling. Unlike Gilbert Lewis, Pauling had
answered. an open mind for quantum theory, which he studied during his
In 1916 Gilbert Lewis published his seminal work in which he research stays in Europe in 1926 in the laboratories of Niels
suggested that the symbolic dash between two atoms should be Bohr, Arnold Sommerfeld, and Erwin Schrödinger. Like Gilbert
identified with an electron pair.2 He also recognized that stable Lewis, he had an enormous knowledge of chemistry and he was
molecules have mostly atoms that possess eight electrons in their also acquainted with the newly emerging technique of X-ray
valence shell. Lewis introduced the cubic model for an atom crystallography. Pauling blended the information from exper-
where the electrons are located at the corners of the cube. He imental chemistry, quantum theory, and structural chemistry
realized that the cubic atom may describe single and double into a far-sighted but also biased vision of molecular structure
bonds but not triple bonds, for which the tetrahedron was a and bonding. He published the results in a series of papers that
better model. As mentioned above, Lewis was aware of the fact finally culminated in the epochal book The Nature of the
that his electron-pair model violates the laws of classical physics Chemical Bond, which was first published in 1939.11 The book
for electrostatic forces. The quantum theoretical laws for the clearly shaped the vision of the chemical bond for generations of
8786 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

chemists until today. Pauling succeeded in the difficult task to suggested an arrow as symbol for a dative bond A→B, as it is
combine the classical viewpoint of the structure of molecules called now.45 Sidgwick also showed that the peculiar dipole
familiar to experimental chemists with the information that was moments of some molecules such as carbon monoxide, which
provided by the rather exotic perspective of quantum theory. He has its negative end at carbon, can be explained with the
recognized a resemblance of the electron-pair model of Lewis to direction of the donor−acceptor bond, which he sketched with
the Heitler−London ansatz for the quantum chemical treatment the formula .46 This explanation was later supported by
of H2, which uses an electron-pair function for the H−H bond, quantum chemical calculations.47 Pauling dismissed the concept
and he developed his theory of resonating structures for the of dative bonds, because he found it inconvenient,11 which may
description of molecules. The localized picture of valence bond be the reason that the relevance of dative bonding in main-group
(VB) theory, which was introduced by Heitler and London, compounds has not been recognized for a very long time.
appeared to him as far more suitable for chemistry as the Haaland stressed the model of dative bonding in main-group
delocalized results of molecular orbital (MO) theory, which he compounds in a review article that appeared in 1989, which was,
strongly opposed. however, restricted to classical donor−acceptor complexes
The opposition of Pauling against MO theory is in retrospect mainly of group 13−15 adducts.48 The full potential of the
difficult to understand for rational reasons, because the model of dative bonding in main-group chemistry came to the
advantages of the MO approach over the VB method for fore in systematic theoretical studies of low-valent compounds
explaining chemical phenomena soon became obvious. The in the recent decade49 that led to the discovery of divalent
electronic triplet ground state of O219 and the peculiar stability carbon(0) compounds (“carbones”)50−54 and related systems,
of aromatic compounds20 could easily be explained with the and contributed to the amazing development of subvalent main-
symmetry of the MO wave function. Pauling did not recognize group chemistry.55,56 This is discussed in a dedicated section
the information that is provided by the symmetry of the wave below. The renaissance of the model of dative bonding was not
function, which later became an important reason for the undisputed,57−59 but recent developments show that it is now
breakthrough of MO theory explaining molecular structure and generally accepted as a valid description for the bonding
reactivity. It was mainly due to the pioneering FMO work by situation in main-group compounds that was already envisaged
Fukui21 and to the orbital symmetry rules by Woodward and by Lewis and Sidgwick.49,60−62
Hoffmann22 that MO theory became the leading source for Finally, we want to comment on molecules that are
bonding models. A second reason for the present dominance of considered as species, which exhibit “unusual bonds”. The
MO theory is the development of computers and efficient origin of the unusual bonding situation is not always the peculiar
algorithms for solving the Schrödinger equation, where the electronic structure but the difficulty to describe it with the
orthogonality of the orbitals make MO calculations substantially Lewis electron-pair model. A striking example is dioxygen O2,
faster than VB calculations. which has an electronic triplet (X3Σg−) ground state. The
More important for the present topic is the fact that MO bonding situation can easily be understood with an orbital
theory is neither in conflict with nor does it invalidate the interaction diagram. “Unusual bonding situations” may some-
electron-pair model of Lewis. Quite on the contrary, the MO times only indicate that the Lewis model is not suited for the
method is even better suited for explaining molecular structures electronic structure, which may easily be understood by another
in terms of Lewis structures than VB calculations, and it offers a model such as the MO correlation diagram.63
quantum theoretical foundation for writing molecules possess- The following conclusions arise from this section:
ing unusual bonds with the electron-pair model.42 At the same • The shared electron-pair bonding model is unsurpassed in
time, it clearly shows the limitations of simple Lewis structures its simplicity for describing molecular structures, but it
for describing chemical bonding. The diatomic molecules Be2 must not be confused with the physical origin of chemical
and O2, which are discussed below, are prominent examples for bonding. The Lewis bonding model lacks the quantum
where the straightforward application of the Lewis paradigm fails theoretical information about the symmetry of the wave
without using the symmetry of the underlying wave function. functions, which leads to problems particularly for
Without knowledge of the orbital symmetry, there is no way to molecules with delocalized bonds and hypervalent
understand why Be2 is essentially unbound with a rather tiny species.
dissociation energy of 2.3 kcal/mol and why O2 has a triplet
• The choice of the best Lewis structure for a given
ground state. Dioxygen is also an example for a molecule where
molecule with unusual bonds should be made in
the electronic structure may not reasonably be expressed by
conjunction with quantum chemical calculations. The
Lewis structures. There is no Lewis formula that satisfies the
evaluation of the symmetry of the wave function, which is
octet rule and the triplet state along with the double bond in O2.
provided by MO calculations, is very helpful for finding
According to the Lewis bonding model and the octet rule,
the best description of the bonding situation.
dioxygen should be a singlet with a classical double bond OO.
Lewis introduced a particular type of electron-pair bond, • The distinction between electron-sharing bonding and
which he used for a general definition of acids and bases that now dative bonding, which was already introduced by Lewis,
carry his name. He wrote that “a basic substance is one which has a then recognized by Sidgwick and later by Haaland, is a
lone pair of electrons which may be used to complete the stable group very useful tool to characterize the nature of the chemical
of other atoms and (...) an acid substance is one which can employ a bond.
lone pair f rom another”.39 He elaborated this model in a later
publication in 1938, which focuses on acid and bases.43 Sidgwick 4. QUANTUM CHEMICAL METHODS FOR
recognized the value of the dative bond for the description of CALCULATING MOLECULAR STRUCTURES AND
coordinative compounds, which he described in 1923.44 It was PROPERTIES
further elaborated in his monograph published in 1927, where It has become a standard procedure in chemical research to
he introduced the notion of donor and acceptor bonds and supplement experimental results with quantum chemical
8787 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

calculations in order to elucidate the bonding situation in E0 MO = E0 0 + E01 + E0 2 + E0 3 + ... (12)


molecules. This is often done with a mere presentation of the
calculated data without critical examination of the theoretical The associated energies are given by eq 12, where E0MO refers
method used. It is regrettable that the information which is to the energy of the electronic ground state, E00 is the Hartree−
provided by the calculations remains often obscure and Fock energy, and the remaining terms are the correlation
questionable, because it is frequently given without knowledge contributions of Nth order arising from a step-by-step variational
of the relevance of the calculated data and the approximations treatment. The first step in determining the MOs consists of a
used. In the following, we will sketch the most important linear combination of the atomic orbitals (LCAO) χa (eq 13)
methods that are presently used in chemical research. We will which gives, via Hartree−Fock calculations (eq 14), the
not discuss the details of the mathematical methods. Our focus is molecular orbitals φi that are used to construct the Slater
not on the technical features of the quite sophisticated determinant Φ0 as a product wave function (eq 15):
approaches used by the quantum chemistry community; they
can be found elsewhere in the literature.64−66 We rather φi = ∑ cα χia
α (13)
emphasize the fundamental aspects of the three most important
methods, where we highlight the information that is given by the Fiφi = εiφi (14)
calculations with respect to the nature of the chemical bond and
the relationship to the bonding models. The sheer presentation Φ0 = |φ1φ2φ3 ... φN|
of calculated numbers in publications is rather useless without a (15)
critical examination of the relevance of the data obtained, which The order of eqs 13 and 15 indicates the basic difference
requires knowledge about the approximations that are inherent between MO theory and VB theory (described below): LCAO-
in the respective method. MO theory is a product of sums while VB theory rests on a sum of
The historically eldest quantum theoretical method is Valence products. MO calculations use orthogonal functions, which,
Bond (VB) theory, which was used by Heitler and London in compared to VB theory, are much easier to implement into
their pioneering study in 1927.8 It was the prevailing approach computer programs leading to significantly faster algorithms.
for some decades, which is mainly due to the strong support by This is the reason why quantum chemical calculations that strive
Pauling.11 The advent of computers and the development of for numerical accuracy are nowadays mostly based on MO
numerical algorithms and codes showed that Molecular Orbital theory; that is, much larger molecules can be calculated with
(MO) theory is far better suited for quantum chemical higher precision than with the VB method.
calculations, basically rendering VB methods obsolete for We want to point out that the molecular orbitals φi in the
numerical studies. Wave function based MO methods clearly Slater determinant Φ0 (eq 15), the so-called canonical MOs, are
became the dominant approaches for quantum chemical a particular set of orbitals that are most commonly used, as these
calculations during the 1970s. They were gradually replaced in have properties that make them well suited for chemical bonding
the past two decades by Density Functional Theory (DFT), models. The symmetry of the orbitals was soon recognized as
which is presently, in numerous different variations, the most important information that indicates the stability and reactivity
important tool for quantum chemical studies of larger of molecules. Hückel realized that the peculiar stability of
molecules.67 Examination of the basis of the Kohn−Sham aromatic compounds can be explained with the symmetry of the
(KS) DFT calculations reveals that they are essentially π MOs, which was later formulated with the famous Hückel 4n +
parametrized variants of the MO method, which is the reason 2 rule. The delocalized picture of the canonical MOs was for
that some authors call them Density Functional Approximation some time an obstacle for accepting it as a valid description of
(DFA). We will use the more common term DFT for this chemical bonding for chemists, because it did not agree with the
method. Because of the close resemblance of the fundamental localized picture of the electron-pair bond of Lewis. The
equations in wave function based MO and KS-DFT calculations, breakthrough came with the FMO model of Fukui,21 the advent
we describe the two methods first before we introduce the basis of the orbital symmetry rules by Woodward and Hoffmann,22
of VB calculations. The latter has lost its relevance not only and the publication of several textbooks on MO theory after
because of its computational drawback as already mentioned but 1960.72−74
also because the apparent advance for the interpretation of MO theory does not automatically yield delocalized MOs.
chemical bonding in terms of the electron-pair model of Lewis The canonical set of orbitals in the Slater determinant is only a
has been lost with the development of the Natural Bond Orbital particular choice, because they provide the information that is
(NBO) method by Weinhold introduced below.68−71 given by the symmetry of the orbitals. The orbitals can be
4.1. Molecular Orbital (MO) Theory transformed via unitary transformations into equivalent MOs
that are subject to the chosen conditions. Thus, the delocalized
MO theory expresses the electronic wave function for the canonical MOs can be transformed to localized orbitals, which
electronic ground state of a molecule Ψ0MO as a sum of Slater closely resemble the electron-pair model of Lewis. It is not
determinants ΦN (eq 11) where the first term Φ0 is usually correct to say that MO calculations necessarily yield delocalized
already a very close approximation to Ψ0MO. The sum of the orbitals, which are difficult to associate with the Lewis electron-
remaining terms runs over the singly (c1), doubly (c2), and pair model. On the contrary, the localized valence MOs exhibit
higher excited configurations of Φ0. The theoretical fundament spatial extensions of the electron pairs that may be associated
of MO theory can be grasped by understanding the elementary with electron-pair bonds and lone pairs. However, the symmetry
steps for the formation of Ψ0MO (configuration interaction information of the canonical MOs is lost and localized MOs do
procedure). not have an eigenvalue attached. Unlike localized orbitals,
canonical orbitals provide symmetry information and they
Ψ0 MO = c0 Φ0 + ∑ c1Φ1 + ∑ c2Φ2 + ∑ c3Φ3+... possess energy eigenvalues, which lead to the situation that
(11) usually only the latter are used as models.
8788 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

However, the use of canonical MOs that construct the Slater correlation through a functional. The expression for the Pauli
determinant Φ0 (eq 15) has a drawback: Only the construction repulsion in FKS is however different from that in FHF. There are
of the first component of the wave function Ψ0MO (eq 11) is further modifications in various different forms of the KS
taken into account. Although the energy contribution of Φ0 is in operator FKS that have been developed over the past decades.78
most cases >99% of the total energy, it may not be sufficient for There is now a rather large and still growing number of DFT
the bond energy, and thus, the use of Hartree−Fock orbitals φi approximations available with different KS operators FKS that
for chemical bonding may be questioned. It turned out, however, strive at the same holy grail: To calculate the KS orbitals φiKS via
that in most cases the MOs that arise from Hartree−Fock eq 16, which, after quadrature and summation, give a total
calculations (eq 14) are sufficient for a qualitative discussion of density ρ(r) that is as close as possible to the exact density. Once
the bonding situation except for van der Waals interactions, this is achieved, all physical properties such as the ground state
which are dominated by electron correlation. Nevertheless, they energy can be obtained. The computational costs of such DFT
have been replaced in recent years by the Kohn−Sham orbitals calculations are comparable to ab initio calculations at the HF
φiKS that arise from DFT calculations, which are briefly level, with the additional bonus that electron correlation is
introduced in the next section. already considered. The enormous progress which has been
4.2. Density Functional Theory (DFT) made in the past two decades in developing DFT led to the
present situation where quantum chemical calculations are now
The basis of density functional theory is the Hohenberg−Kohn
mainly performed using DFT with various functionals.79,80 One
Theorem,75 which states that the ground state properties of a
important aspect that distinguishes DFT from MO theory must
many-electron system are uniquely determined by the electron
be stressed, however. The quality and reliability of MO
density ρ(r), depending only on the three spatial coordinates
calculations can stepwise be improved toward the exact solution
and the external potential; hence, the total energy of the ground
state is a unique functional of the electron density. Unlike the by using higher correlated methods and larger basis sets. This
wave function based MO and VB theories, DFT takes the leads eventually to the correct physical properties in agreement
density as the starting point of the calculations where the total with experiment, albeit at steeply increasing computing costs.
density of an N-electron system is given by the sum of the one- DFT calculations at a certain level can, at the moment, not be
electron densities. Thus, the result of a DFT calculation is (in improved in this fashion.81
principle) not an approximate energy E0N as in MO calculations We do not wish to discuss the pros and cons of various DFT
but the correct total energy E0 (compare eq 12), provided that approximations here. We rather focus on aspects relevant for the
the correct density ρ(r) and functional are known. It must be present topic, which is the nature of the chemical bond. The
realized that DFT does not focus on the formation of the Kohn−Sham orbitals φiKS were originally considered a mere
electron density ρ(r) but rather on the connection between the technical detail that has accessory character for obtaining the
density and the associated energy. The density ρ(r) is exact electron density ρ(r). Inspection of the shape and the
considered as the starting point of DFT. But the work by energy levels of φiKS along with their atomic orbital composition
Heitler and London8 has shown that the chemical bonds in a showed that they are closely related to HF orbitals φiHF, with the
molecule can be described by the interference of wave functions. additional bonus that the φiKS are associated with the exact wave
The exact density ρ(r) can of course be obtained from the exact function Ψ0 while the φiHF are associated with the HF wave
wave function Ψ, but it can also be constructed from the Kohn− function Φ0 (eq 11). This led to the present situation that KS
Sham (KS)76 functions shown in eq 16, which are the working orbitals φiKS are routinely used for the analysis of the electronic
horse of present DFT methods. Within KS theory the total structure and bonding situation in molecules in the framework
density of an N-electron system is simply the sum of one- of qualitative molecular orbital theory, which originally
electron densities. The KS equations look similar to the employed semiempirical methods or Hartree−Fock orbitals.
Hartree−Fock equations (eq 14), which are shown again in eq Most of the works, which are discussed in this review, are
17, and we use the superscript HF in order to distinguish them therefore based on DFT calculations and the analysis of Kohn−
from the Kohn−Sham equations: Sham orbitals φiKS.78−84
There is one important aspect of DFT that is very important
F KSiφi KS = εi KSφi KS (16) for the present topic. There are molecules which possess unusual
electronic structures in the sense that they are poorly described
F HFiφi HF = εi HFφi HF with MO methods that use only one determinant as leading
(17)
configuration (static correlation). Such molecules usually pose
The crucial difference between the Kohn−Sham (KS) problems for traditional DFT methods, which so far are
equations (eq 16) and the Hartree−Fock (HF) equations (eq restricted in the wave function picture to single determinant
17) lies in the expressions for the operators FKS and FHF.77,78 The variants. Fractional occupation numbers can be, however, used
HF operator FHF contains well-defined terms for the kinetic and and, similar to wave function based methods, multideterminant
potential energy of an electron and its average electrostatic DFT methods are being developed to solve problems such as
repulsion with the other electrons as well as the exchange (Pauli) double-counting electron correlation.85−88 These are interesting
interaction with other electrons that have the same spin. This developments, but they have not yet reached the stage of routine
leads to the HF energy E00, which is the first term in eq 12. The methods. Accurate wave function based MO methods still play
individual correlation with the other electrons is neglected but an important role for checking the reliability of DFT calculations
can be considered by the higher order terms in eq 12, which, for physical properties, particularly in molecules that are not well
when included in a CI procedure, is quite computer time- described with one Lewis structure. Approximate DFT methods
consuming. This is the well-known bottleneck of correlated ab as we use them today cover electron correlation in an
initio calculations. unspecified way as a consequence of their semiempirical nature.
DFT calculations try to rectify this problem by adding a Therefore, they cannot be used as a black box, and the results
further term to the KS operator FKS that considers exchange- should always be checked for a smaller test set against high
8789 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

quality wave function results. We warn against an indiscriminate should always be checked for a smaller test set against high
use of DFT calculations for analyzing the bonding situation in a quality wave function results.
molecule. • VB calculations are more cumbersome compared to MO
4.3. Valence-Bond (VB) Theory and particularly KS-DFT calculations. The identification
of the leading terms of VB calculations with covalent and
Unlike the MO method, where the wave function Ψ0MO is ionic bonding has no physical fundament.
assembled as the product of one-electron functions φi yielding
the Slater determinant Φ0 (eqs 11 and 15), the VB wave function 5. QUANTUM CHEMICAL METHODS FOR ANALYZING
Ψ0VB is expressed as the sum of localized two-center product- THE CHEMICAL BOND IN MOLECULES
functions (λaλb).89−91 The occupation of the product-functions
λaλb by two electrons of opposite spins gives the electron-sharing It has become a standard part, also in experimental work, to
covalent term (λa−λb) (“Heitler−London” term) and the two present the results of quantum chemical calculations in
ionic terms (λa|− λb+) and (λa+ λb|−). The basic VB expansion, conjunction with a breakdown of the calculated numbers in
which considers all possible products in the molecule, is given by terms of chemically meaningful information about the bonding
eq 18, where the summation runs over all electron pairs in the situation and electronic structure in the molecules. There are
molecule:90 several methods available, which serve as a bridge between the
computed numbers of the quantum chemical calculations and
Ψ0 VB = ∑ c1(λa−λ b) + ∑ c2(λa|−λ b+) + ∑ c3(λa+λ b|−) the heuristic bonding schemes such as the Lewis model. In the
(18)
following, we present a short summary of the approaches and
basic features of the most common methods, which should be
As mentioned above, the use of orthogonal orbitals in MO understood in order to prevent misinterpretation of the results.
calculations leads nowadays to much faster algorithms in Reasonable conclusions about the bonding situation in a
computer programs than VB methods, where overlapping molecule are only possible when the most important details of
hybrid orbitals are employed. VB methods do not play a role the sophisticated algorithms and the inherent assumptions,
in theoretical research of larger molecules, which strives for which are intrinsically coded in the algorithms, are known. A
numerical accuracy. They were, however, favored over MO mere presentation of the calculated numbers without knowledge
theory by Pauling, because they appear at first sight to be more of how they were obtained is meaningless.
closely related to the electron-pair model of Lewis. The first term Numerous charge- and energy-partitioning methods have
in eq 18 could be identified with the covalent electron-pair bond, been developed over the course of time. The five conditions
whereas the second and third terms give the ionic bonding given in Scheme 2 should be met by a partitioning procedure, in
contribution. The coefficients c1 and c2/c3 appear as ideally order to provide a meaningful interpretation of the vast
suited to indicate the covalent and ionic character of a bond. complexity of chemical observations in terms of a bonding
This was suggested by Pauling in his book The Nature of the model.
Chemical Bond, which shaped the understanding of chemical
bonding of many chemists for decades.11 The development of Scheme 2. Five Conditions That Should Be Fulfilled by a
computers and algorithms for carrying out VB calculations, Model for a Chemical Bond
however, made it possible to check the validity of this qualitative
model. It turned out that there are molecules such as F2 where
neither the covalent term (λa−λb) nor the ionic terms (λa|− λb+)
and (λa+ λb|−) alone yield the correct bonding energy. It is the
mixing of the two terms that leads to a quantitative account of
the bond energy. This makes the interpretation of the VB terms
in eq 18 for indicating covalent and ionic bonding questionable.
The Lewis electron-pair bonding model refers to the total bond 5.1. Natural Bond Orbital Method (NBO)
while the covalent term (λa−λb) covers only part of the total The presently most popular tool for a bonding analysis of
energy of a VB calculation.92 The nature of the chemical bond in molecules is the NBO (Natural Bond Orbital) method by
F2 and other diatomic molecules is discussed further below. The Weinhold and co-workers.68−71 It was introduced in the 1980s
conclusion is that VB methods are not necessarily better suited and has been further developed in the course of time during
for the interpretation of the chemical bond in terms of the Lewis which it underwent significant changes.93 Thus, the results of an
bonding model. early variant of the NBO program may be quite different from
The following conclusions arise from this section: those of a newer version. The goal of the NBO method is to
• Molecular orbital calculations provide a reliable frame- provide the most reasonable Lewis structure for a calculated
work for the identification of the electronic structure of a molecule along with information about the polarity of the bond
molecule, but they may become quite expensive when a as well as the hybridization of the atoms and their partial charges.
single determinant calculation is not adequate for the The problem of determining the best Lewis structure is that
situation. subjective criteria are taken into account when selecting the
mathematical steps and computer algorithms. This is true for
• The results of MO calculations can easily by expressed in any chemical bond model. It is adamantly important that the
terms of Lewis structures by using localized MOs. subjective criteria must be known when the results are
• Present DFT calculations can be considered as para- interpreted and compared with the values of other methods.
metrized MO calculations. They are much more efficient, The NBO method takes the one-electron density matrix,
and they often give very good results at lower costs than which is given by a quantum chemical calculation, as starting
Hartree−Fock based MO calculations. However, they point for the partitioning procedure. Because the NBO method
should not be used as black box methods, and the results uses the density ρ instead of the wave function Ψ as initial point,
8790 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

it may be used in conjunction with any correlated ab initio but not to provide information about the electronic structure of
method or DFT calculation. This is a strong point of the a molecule.
method. But there are two little known aspects of the NBO The NBO method was further developed toward the NRT
method, which may lead to debatable results. One aspect (Natural Resonance Theory) method, where energy expressions
concerns the choice of the atomic valence orbitals, which are for the intramolecular orbital interaction were suggested.104−106
considered for the formation of the molecular orbitals. The The NRT was met with immediate scepticism,107 and the energy
NBO algorithm makes a preselection of those atomic functions values that are provided should be taken with some caution. The
that are considered as genuine valence orbitals and those who NBO method focuses exclusively on orbital interactions in a
are termed Rydberg functions. The two sets are differently molecule. Electrostatic interactions and Pauli repulsion are only
treated in the subsequent algorithms, where the Rydberg orbitals indirectly considered. This may lead to conflicting interpreta-
are technically treated with lower priority, which leads to biased tions about chemical bonding when other methods are
results. The division of the atomic orbitals into the two sets of employed that include all physical forces.108−110 A strong
atomic functions is the subjective choice of the authors. Thus, point of the NBO method is the finding that the numerical
the outcome of NBO calculations is not “natural”, it is rather the results are usually rather insensitive to the level of theory, which
consequence of what the authors consider to be “reasonable”. is of great advantage compared with the rather outdated
The NBO method considers only those outermost atomic Mulliken population analysis whose results are plagued by large
functions as genuine valence orbitals, which are occupied in the variations when different basis sets are used.
electronic ground state of the atom. Thus, the NBO algorithm An extension of the NBO method to molecules that exhibit
considers the alkali and alkaline earth atoms having a (n)s strongly delocalized bonds, such as electron deficient boron
valence shell and the group 13−18 atoms to possess a (n)s(n)p compounds, has been developed by Boldyrev and co-workers.111
valence shell. The question whether the octet rule is valid for The Adaptive Natural Density Partitioning (AdNDP) method
main group atoms cannot be addressed in NBO calculations, describes N-center-two-electron bonds where N gives the
because the answer is already coded into the algorithms.94 The number of atoms to which the two-electron bond should be
above criterion leads to a critical situation for transition metals, assigned. The AdNDP method has been found to be particularly
which have only an (n)s(n − 1)d valence shell. The latter choice useful for the description of organic and inorganic molecules and
leads to a 12-electron rule rather than 18-electron rule for clusters where the standard NBO method is less suitable. A
transition metal complexes, because the (n)p AOs are only reformulation of the AdNDP approach has recently been
treated as Rydberg orbitals.95−97 This has been criticized by introduced by Pendás and Francesco, which provides Lewis
several authors, because the 18-electron rule is well established structures from real space analyses of general wave functions.112
in transition metal chemistry.98−100 Very recently, the transition A warning shall be addressed when the results of a bonding
metal compounds [(CO)3Ni-E]− (E = Li − Cs) were reported, analysis are judged by their agreement with “chemical intuition”.
where the covalent Ni−E bond at the nickel end comes mainly The vaguely defined concept of chemical intuition, which refers
from the Ni(4p) AO.101 The NBO calculations using version 6.0 to the wealth of experimental experience in chemistry, may be
do not give a faithful description of the Ni−E bonds. But even helpful as a guideline for future experiments and for the design of
for main group atoms, the preselection of valence orbitals a specific bonding model, but it is very questionable when it
according to the NBO method can sometimes become comes to a physical interpretation of chemical bonding. As
questionable. It was lately shown that the heavier earth alkaline discussed above, chemical bonding can be regarded as a
atoms M = Ca, Sr, Ba form strongly bonded octa-carbonyls quantum theoretical phenomenon due to the interference of
M(CO)8 using their (n − 1)d orbitals, which challenge the wave functions. Quantum theory and chemical intuition often
restriction of the valence orbitals to (n)s functions.102 do not fit together well, since human sensory perception, like
The second critical aspect concerns the algorithms of the classical physics, is limited in space and time. Both quantum
current eight orthogonalization steps, which have changed theory and relativity are often at odds with the human mind and
during the different versions of NBO and sometimes lead to intuition and may perhaps only be accessible through
inconsistent results when different versions are used. Early mathematics. The difficulty of getting along with quantum
versions of the program showed that the 3d participation of theory was suitably expressed by Niels Bohr: “Those who are not
sulfur in the bonding in SF6 is rather small, which agrees with the shocked when they f irst come across quantum theory cannot possibly
general notion that d-AOs of the heavier main-group atoms are have understood it”.113
polarization functions and that the model of hypervalence in
5.2. Quantum Theory of Atoms in Molecules (QTAIM)
such compounds should be discarded.103 However, the recent
version NBO 6.0 surprisingly gives Lewis structures for SF6 with The NBO method is a charge-partitioning procedure that aims
perfect sp3d2 hybridization. A similar result is obtained for PF5, at the division of the wave function Ψ into atomic regions. In
which, according to NBO 6.0, has sp3d hybridized P−F bonds. contrast, the QTAIM (Quantum Theory of Atoms in Molecules)
This is likely due to the projection operator in the NBO version developed by Bader is based on a topological analysis of the
6.0, which yields an inappropriate hybridization for SF6 and PF5. electron density ρ(r) into atomic basins.114 The mathematically
This has been fixed in the latest version NBO 7.0, which was well-defined QTAIM method avoids some shortcomings and
distributed while this review was written. The problem remains problems of wave function based partitioning schemes. Using
that the biased treatment of the outermost atomic orbitals and the first derivatives (gradient field) ∇ρ(r) and second
the technical details of the orthogonalization steps may lead to derivatives (Laplacian) ∇2ρ(r) of the electron density, physi-
results which should be checked in comparison with other cally well-defined atomic regions and interatomic bond paths
methods that are based on different partitioning schemes. may be identified, which typify a molecular structure by the zero-
Whatever method is used, it holds that one should know the flux surfaces that separate the atomic basins. The zero-flux
details of a method in order to make a reasonable statement. The surfaces are defined as the gradient vector field whose
aim of the NBO method is to propose a suitable Lewis structure trajectories of ∇ρ(r) do not vanish at the atomic nuclei but at
8791 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

the bond critical point rb, which is a signature of a bond path. A 5.3. Energy Decomposition Analysis and Natural Orbitals
bond critical point rb has two negative and one positive for Chemical Valence (EDA-NOCV)
eigenvalues of the second derivatives of ρ(rb). Other critical The NBO and QTAIM methods are charge-partitioning
points at which the gradient of the electron density ∇ρ(r) techniques, which divide the electronic structure of molecules
vanishes define rings and cages. The bond path is the trajectory into atomic regions. There are complementary methods that
which belongs to the positive eigenvalue connecting the bond first break up the total energy of a molecule into fragments,
critical point rb and the bonded atomic nuclei. Thus, the which are then reassembled in a stepwise fashion where each
QTAIM method gives a physically sound description of the step is assigned to a particular type of interaction. The choice of
skeletal structure of the molecule in terms of atomic nuclei and the fragments is determined by the question of interest. Trivially,
chemical bonds, which has been proven useful for a variety of in a diatomic molecule X2 the two atoms are chosen as
chemical systems.115 interacting fragments, but already when a heteroatomic species
The QTAIM method is possibly the physically best founded such as LiF shall be analyzed, it may be more useful to choose the
theory for chemical bonding, providing a rigorous quantum ionic fragments Li+ and F− for the bonding analysis. This seems
theoretical proof that the model of discussing the chemical to make the method somewhat arbitrary and therefore
behavior of a molecule in terms of atomic properties is justified. questionable, but the possibility of using different fragments
Problems arise when the QTAIM results are connected to the for the bonding analysis provides more flexibility that may be
manifold of chemical observations, which are often more used to address different questions about the bonding nature.
meaningfully described with heuristic orbital models. As Thus, the choice of neutral Li and F as interacting fragments
mentioned above, the wave function and the orbitals exhibit includes all changes along the bond formation between the
important information through their symmetry, which is lost isolated atoms toward LiF, whereas the choice of the ions Li+
and F− addresses the question about the nature of the eventually
when only the electron density ρ is used instead of Ψ. This
formed bond. These are two different questions, which have two
makes the QTAIM method perhaps less powerful for making
different answers that can be addressed by choosing different
predictions of chemical reactions. Another problem is the use of fragments.
the bond critical point rb as an indicator of a genuine chemical The EDA-NOCV method123,124 is a combination of the EDA
bond. Diatomic He2 is a weakly bonded van der Waals complex (Energy Decomposition Analysis) method, which goes back to
with negligible interference of the atomic wave functions, and initial ideas by Morokuma and Kitaura125 and later by Ziegler
thus, it has no covalent bond. QTAIM calculations indicate a and Rauk,126 and the original charge-partitioning method
bond critical point and a bond path just like in H2. This was NOCV (Natural Orbitals for Chemical Valence) by Mitoraj
recognized by Bader, who clearly stated that “bond paths are not and Michalak.127,128 It has been proven as a very powerful tool
chemical bonds”.116 On the other hand, there are molecules for bonding analyses, because it considers all types of physical
where by visual inspection of the electronic structure one would interactions, which eventually sum up to the experimentally
ascribe a covalent bond between two atoms but there is no bond observable bond dissociation energy (BDE). In particular, the
path.117,118 further partitioning of the orbital (covalent) term into pairwise
There has been theoretical work to distinguish different types orbital interactions has been proven very helpful, because it
of chemical bonds that possess a bond critical point using provides a quantitative expression of the FMO model of Fukui21
QTAIM parameters. The most successful one goes back to the and the orbital symmetry rules of Woodward and Hoffman.22 A
investigations by Cremer and Kraka, who suggested that the nice feature of the EDA-NOCV model is that the orbital
energy density at the bond critical point H(rb) is a useful interactions are not only given in terms of numbers, which
criterion for defining the nature of a chemical bond.119 indicate their strength, but the associated charge migration can
According to the Cremer−Kraka criterion, a covalent bond visually be shown. In the following, we present the most
between two atoms A and B is defined by the existence of a zero- important aspects of the EDA-NOCV method. A more detailed
flux surface and bond critical point rb between the atoms discussion of the method and its application has been given in
(necessary condition) and a negative and thereby stabilizing local recent review articles.129,130 At the present time, EDA-NOCV
energy density H(rb) (suf f icient condition). H(rb) will be close to analyses can only be carried out using DFT or Hartree−Fock
zero or positive if the interaction between A and B comes from calculations. Since correlation energy is missing in the latter
the electrostatic or dispersion forces. This approach has been method, more meaningful results are obtained with DFT
approaches. The results do not change very much when different
found very helpful for the bonding analysis between electro-
functionals are employed.
negative atoms.120,121 But the numerical absolute values may
The focus of the EDA approach is the instantaneous
become very small when bonds between heavy atoms are interaction energy ΔEint of a bond A−B between two (or
involved and the degree of covalent bonding is not directly more) fragments A and B in the particular electronic reference
available from the values of H(rb). state and in the frozen geometry of AB.131,132 It is very important
In spite of the caveats, it may be stated that the QTAIM to recognize the correct electronic reference states of the
method has been established as an important method to fragments in order to provide a meaningful analysis of the
elucidate the electronic structure not only in molecules but also interactions A−B. In doubtful cases, one may carry out EDA
in solids.122 Inspection of the Laplacian distribution of the calculations with different electronic states of the fragments and
electron density ∇2ρ(r) provides valuable information about the compare the results. Those fragments, which give the smallest
areas of charge depletion and charge accumulation in a energy change during the bond formation step, are the most
molecule, which come from the interatomic interaction. The useful species for the bonding analysis.
combination of QTAIM calculations with the results of an MO The interaction energy ΔEint is divided into three main
analysis is a very powerful tool, which helps to express the components, which are sometimes augmented by a fourth
bonding situation in a molecule with an appropriate model. expression for dispersion (van der Waals) interactions:
8792 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

ΔEint = ΔEelstat + ΔEpauli + ΔEorb + (ΔEdisp) (19) There is one important aspect of the ΔEelstat term in the above
EDA method that deserves to be discussed. The partitioning
In the first step of the EDA procedure, the fragments A and B scheme uses overlapping spheres of the atomic charge
are superimposed with their frozen densities at the geometry of distributions, which means that some electronic charge in the
the molecule AB, which gives the quasi-classical electrostatic molecule that is close to the nucleus of atom α is assigned as
interaction between the unperturbed charge distributions of the coming from atom β. This may appear odd for human
prepared fragments ΔEelstat. The term is usually attractive, imagination that is shaped by senses, which are trained by
because the frozen charge densities of the fragments overlap and classical hard objects. If one assumes the atoms are spheres with
thus penetrate each other, which induces attraction with the impenetrable surfaces it seems more “reasonable” to cut a
atomic nuclei of the other fragment.28 The overlapping densities molecule into atomic pieces by sharp boundaries as for instance
may have electrons that have the same spin, which according to in the IQA (Interacting Quantum Atoms) model,147−150 which
the Pauli principle is not allowed. This is rectified in the second is based on Bader’s QTAIM definition of separated atomic
step of the EDA procedure, where the superposition of the basins.114 However, in quantum theory, an electron is described
unperturbed electron densities of the isolated atoms is subject to by a wave function, which has an exponentially decreasing but
antisymmetrization (operator  ) and renormalization (constant finite value even at larger distances and so has its square, which
N), which gives the product wave function Ψ0 = N Â [ΨA(α) gives the charge density of the electron (see below). Thus, for
ΨB(β)] still without interference of the fragment wave functions example in N2, an electron of Nα has a finite density also close to
ΨA(α) and ΨB(β). The energy term ΔEPauli, which is calculated in the nucleus of Nβ. The estimate of Coulombic interactions using
the second step, accounts for the electron−electron repulsion interpenetrating charge distributions in the EDA method thus
due to the orthogonality requirement of the orbitals. The final appears as a physically reasonable approach consistent with
term ΔEorb accounts for the formation of covalent bonding via quantum theory, which is commonly used in many theoretical
interfragment mixing of the orbitals but also for the polarization approaches, such as Mulliken analysis, Natural Bond Orbital
within the fragments via intrafragment orbital mixing. Polar- analysis, Hirshfeld partitioning,151 etc. On the other hand, the
ization and charge transfer between the fragments are not wave function of an electron of Nα in N2 is subject to
separated in the EDA approach. interference with the wave functions of the electrons of Nβ. This
If the two fragments of the chemical bond are in an leads to some ambiguity in assigning the charge in a molecule
electronically excited state or if they have more than one that comes from the interfered wave functions to a particular
atom, there is an electronic and possibly a geometric relaxation atom. In conclusion, the many models of interpenetrating
of the fragments during bond rupture into the equilibrium charges as well as strictly separated charges may all appear more
geometry of the fragments, which corresponds to the or less physically reasonable.
preparation energy ΔEprep. If the energy value is added to The NOCV scheme127 divides the orbital interaction term
ΔEint, one obtains the bond dissociation energy De, which is by ΔEorb in the EDA-NOCV approach into pairwise contributions
definition the negative value of the total bond energy ΔE: of interacting orbitals of the two fragments. The starting point is
the deformation density Δρ(r), which is the difference between
ΔE ( = −De) = ΔEint + ΔEprep (20) the densities of the fragments before and after bond formation.
The deformation density Δρ(r) can be expressed in terms of
Although the individual terms in eqs 19 and 20 are not pairs of complementary eigenfunctions (ψk ψ−k) with the
observable quantities, the sum of them gives the experimentally eigenvalues υk and υ−k that possess the same absolute value but
measurable bond dissociation energy. The absolute value of opposite sign. The absolute numbers of the eigenvalues υk are a
ΔEint is a faithful expression for the intrinsic bond strength, measure for the magnitude of the charge transfer:
whereas the De values are affected by the geometrical and
possibly the electronic relaxation of the fragments. There are
Δρ(r) = ∑ υk[−ψ 2−k(r) + ψ 2k(r)] = ∑ Δρk(r)
k k (21)
molecules where the fragments A and B are lower in energy than
the molecule AB, which makes the De values useless for The NOCVs ψk and the associated eigenvalues υk are obtained
estimating the bond strength A−B. This holds in particular for through diagonalization of the difference density matrix ΔPμν of
high-energy materials and explosives where bond breaking is a the system. Equation 21 facilitates the expression of the total
highly exothermic process. It is thus advisible to use the charge deformation Δρ(r) that goes along with the bond
interaction energy ΔEint for comparing the bond strength of formation in terms of pairwise charge contributions Δρk(r)
molecules. which come from particular pairs of (NOCV) orbitals. The total
There are cases where the electronic state of the fragments orbital interaction ΔEorb may likewise be derived from pairwise
used in the bonding analysis is not obvious. For example, N- orbital interaction energies ΔEkorb which are associated with
oxides R3NO might be described with a dative bond R3N→O or Δρk(r):
with an electron-sharing single bond R3N+−O− or even a double
bond R3NO.133 The EDA method can be used to suggest the ΔEorb = ∑ ΔEk orb = ∑ υk[−FTS−k ,−k + FTSk ,k ]
k k (22)
best description according to the electronic structure, by using
the fragments in the respective electronic state in the actual Experience has shown that the ΔEorb term of the EDA-NOCV
calculation. Those fragments, whose energy changes the least approach has usually only a very small number of significant
during bond formation, appear the best choice for the contributions of ΔEkorb, which makes it possible to identify
representation of the bonding situation. The smallest absolute specific orbital interactions that lead to a chemical bond. This
value of the orbital term ΔEorb may be used as a measure for the connects the results of the EDA-NOCV method with classical
best Lewis description of a molecule. This procedure has been models of orbital interactions63 and with the Lewis electron-pair
proven very helpful in a variety of compounds where the model.152 Numerous examples where the EDA-NOCV method
bonding situation is not directly clear.49,134−146 proved helpful for a variety of main-group compounds134−146
8793 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

and transition metal complexes153−163 have been reported in each other. It would be much better to rely only on measurable
recent years. For a fruitful and enlightening discussion about quantities and on physical observables.
different results of the NBO method and EDA-NOCV We think that such criticism stems from a misunderstanding
calculations on the nature of bonding in transition metal of the role and the relevance of bonding models and physical
compounds, we refer to two recent papers by Weinhold and co- theory in chemistry. It is also based on a questionable definition
workers.164,165 of observables and physical reality, whose understanding has
The following conclusions arise from this section: profoundly changed with the advent of quantum theory. The
• There are several powerful analytical tools available which charge and mass of an electron are fundamental constants which
extract information from quantum chemical methods enter the Schrödinger (or Dirac) equation, which then are used
aiming at an understanding of the numerical results in to calculate properties to very high accuracy in perfect
terms of a bonding model. But the user must be aware of agreement with experiment. It is often said that the wave
the details of the methods in order to provide a function does not correspond to a quantity which can be
meaningful interpretation of the results of the charge- measured; it is a purely mathematical construct to describe the
and energy-partitioning schemes. electronic charge distribution. This led to a debate between
Heisenberg and Einstein: “One cannot observe the electron orbits
• The NBO method has been designed to extract the best inside the atom. [...] but since it is reasonable to consider only those
Lewis electron-pair description of the molecular elec- quantities in a theory that can be measured, it seemed natural to me
tronic structure. The atomic orbitals are not identically to introduce them only as entities, as representatives of electron
treated; the results are biased toward a localized electron- orbits, so to speak” (Heisenberg). “But you do not seriously believe
pair bonding model favoring preselected atomic orbitals, that only observable quantities should be considered in a physical
which is considered by the program designers as theory?” (Einstein).166 But the one-particle density ρ obtained
reasonable. The NBO method explicitely focuses only from the wave function is also a mathematical construct, which is
on orbital interactions; other interatomic forces such as used to describe experimental results. In this case, however, the
Coulomb forces are not directly taken into account. charge density can be probed experimentally as it is done for
• The QTAIM method takes the electron density as starting example in X-ray diffraction. Since the outcome of a Diels−Alder
point for the analysis of the electronic structure. The reaction can only be explained with the help of the symmetry of
information that comes from the symmetry of the orbitals the wave function Ψ, which uniquely identifies the interacting
is not considered. It provides a direct connection between species, it acquires the quality and the status of a physically
the physical description of interatomic interactions and relevant entity that becomes evident by the reaction product.
the electronic structure. Note that a complete wave function of a particle is a valid
• The EDA-NOCV method is a combination of charge- and expression containing all information about its physical
energy-partitioning scheme. It provides a quantitative properties.
interpretation of the chemical bond in terms of physically We want to make a reference to a recent perspective on
meaningful contributions. The NOCV method is a bridge quantum mechanics and chemical concepts by Clark, Murray,
between quantum chemical calculations and the FMO and Politzer (CMP).167 Although the focus of their work is on
model of Fukui and the orbital symmetry rules by noncovalent bonding, the central topic of their work has much in
Woodward and Hoffmann. The strength of the orbital common with the present article. The authors cite in the
interactions can be expressed numerically, and the beginning a statement by Ivanic, Atchity, and Ruedenberg
associated charge deformation can be visualized with whose final words nicely sum up the central theme “...the
the help of the deformation densities. resolution of molecules in terms of atoms is not f undamental to
• All analytical methods have their strengths and weak- rigorous physical theory”.168 CMP rightfully point out that “the
nesses. It is advisible to compare the results of several distinction between what is mathematical modeling and what is
methods in order to give a reasonable interpretation of the physical reality may become blurred; the model may be taken to
bonding situation. The results of the bonding analysis represent reality. This can be misleading and conf using.”167 We
suggest a model description, which helps to understand agree, but the situation becomes a bit complicated when the
the experimental results in the framework of an ordering status of the wave function is concerned. The authors argue that
scheme that is based on quantum chemical calculations. the electron density represents physical reality but not the wave
But bonding models are not right or wrong; they are more function. They quote Schrödinger, who warned against using the
or less useful. wave function for the direct interpretation of physical
observations.169 The pioneers of quantum theory are known
6. PHYSICAL REALITY AND CHEMICAL BONDING not only for their ground breaking contributions to physics but
MODELS also for their limited acceptance of the outcome of their theories.
A frequent topic of controversial discussions among chemists is Einstein never accepted quantum mechanics as a valid physical
the relationship between physical reality and chemical bonding theory, and Dirac wrongly suggested that relativistic effects are
models. This holds in particular for models that are derived from negligible for chemistry. The direct relevance of the symmetry of
quantum chemical calculations such as the NBO and EDA- the wave functions for chemical reactions was not obvious when
NOCV methods, which quantitatively define expressions related Schrödinger made his statement in 1926.169 The consideration
to the chemical bond that are given in terms of calculated of the wave function or electron density as the basis of the
numbers. It is sometimes criticized that these numbers cannot physical reality of matter can be seen as representing the
be measured by experiment and, therefore, they would not have continuous versus atomistic theory of the universe. They are
any physical meaning. Moreover, different definitions of rather complementary and not mutually exclusive. It is often said
equivalent terms such as atomic partial charges or bond orders that the present generation of scientists is standing on the
may lead to very different values and sometimes even contradict shoulders of giants. But this enables us to see further than they
8794 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

could, and we should not be afraid of challenging traditional to faciliate the comparison of molecules and reactions.
viewpoints. If one only repeats earlier findings, this means that Different tools can be applied to achieve a better
nothing new has been learned. understanding; their choice may depend on personal
The numbers, which come from chemical bonding models preferences and training. Different results and different
such as NBO or EDA-NOCV, arise from a breakdown of Ψ into views from different methods should be analyzed in the
fragments, which are defined via the particular charge- and light of the particular decomposition scheme, whose
energy-partitioning method. They are clearly not observables in suppositions and technical implementations must be
a quantum mechanical sense. Their relevance comes exclusively known for a meaningful interpretation of the results.
from their ability to construct an ordering scheme, which helps
to divide the enormous complexity of chemical reactions and 7. SELECTED EXAMPLES OF CHEMICAL BONDS IN
molecular structures into specific classes and groups. Histor- MAIN-GROUP COMPOUNDS
ically, this was done by the finding that certain molecules exhibit In the following, we discuss the chemical bond in selected main-
similar behavior. Heuristic notions were suggested such as group compounds, which serve as examples for the wide variety
aromaticity, conjugation, nucleophilicity, formal charge, etc. of molecules that possess different types of bonds. We try to
Although their definition is vague, these models are crucial for cover a wide range of bonding, which may be used as templates
bringing order into the pandemonium of experimental results. for other species. The goal of this section is to show how modern
Bonding models are vital for chemical research, because methods of bonding analysis may be used to (a) elucidate the
otherwise there would not be any systematic ordering of the nature of the interatomic interactions and (b) express the bond
infinite number of molecules and reactions. If one would explain in terms of a model that agrees with the physical mechanism of
chemical results only in terms of physical observables (including the bond. The latter part may lead to conflicting results with
the wave function) without bonding models, it would result in a earlier work presented in textbooks, because previous studies
huge table with data that do not provide any intuitive were often based on ad-hoc assumptions that are not justified.
understanding. Chemistry is by its very nature a fuzzy science, We present results of quantum chemical calculations mostly
because the jungle of boundless number of molecular structures using DFT methods, which are well suited for a bonding
and reactions request methodical schemes that serve as guidance analysis. The focus is not on high numerical accuracy but on the
for future directions of chemical research. The quantum nature of the interactions. On the other hand, the calculated
theoretical postulates are the physical framework for chemistry, observables of the molecules given in this work such as the
but the inner substance of chemistry is guided by models. equilibrium geometry and the bond dissociation energy are
Unlike the earlier heuristic models, which were based on assured to be in reasonable agreement with experiment, in
assumptions that are rooted in classical physics, modern accordance with condition 1 in Scheme 2.
bonding models come from well-defined though chemical- 7.1. H2+, H2, Li2+, Li2
intuition guided partitioning schemes of accurate quantum
chemical methods. They should fulfill the conditions that are Table 1 shows the calculated energy components for the
shown in Scheme 2. The advantage is that the comparison and interactions in H2+, H2, Li2+, and Li2 according to the EDA-
trend of different molecules can now be made with the help of
numbers rather than hand-waving arguments. The finding that Table 1. EDA-NOCV Results of H2, H2+, Li2, and Li2+ Species
different partitioning schemes may lead to different and at the RPBE/TZ2P//BP86/def2-TZVPP Levela
sometimes conflicting views should not be taken as argument Molecule H2 H2+ Li2 Li2+
against their use. Instead, the assumptions of a model and the +
Fragments [H] + [H] [H] + [H] [Li] + [Li] [Li]+ + [Li]
technical details of their implementation should be considered
before a conclusion is made. This requests the knowledge about ΔEint −106.6 −68.7 −20.8 −28.5
ΔEPauli 0.0 0.0 1.9 2.1
the most important aspects of the theoretical methods that are
ΔEelstatb 5.5 17.6 −8.3 (36.7%) 1.6
employed, just as is done with experimental spectroscopic
ΔEorbb −112.1 −86.3 −14.3 −32.2
methods. (100%) (100%) (63.3%) (100%)
The following conclusions arise from this section: ΔEorb(σ)c −111.5 −84.9 −12.9 −32.2
• In the quantum world, the physical reality of an electron is (99.5%) (98.4%) (90.2%) (100%)
a
not entirely assigned to its electronic charge distribution Energies are in kcal/mol. bThe values within the parentheses show
ρ, which represents only a projection onto a space of lower the contribution to the total attractive interactions ΔEelstat + ΔEorb.
information content; its completeness is only provided by For those cases where ΔEelstat is positive, it is excluded. cThe
remaining small contributions come from polarization functions.
its wave function Ψ. The wave function Ψ contains more
information about the behavior of the electron than ρ. In
chemistry, this comes to the fore for example in the NOCV method. The intrinsic interaction energy ΔEint is in these
outcome of pericyclic reactions, or in any spectroscopic cases identical to the bond dissociation energy De, because there
investigation, which can only be explained when the is no electronic or geometric relaxation of the fragments. The
symmetry and sign pattern of Ψ is considered. one-electron bond in H2+ has a bond energy of De = 68.5 kcal/
mol, which is nearly two-thirds of the bond strength in H2. The
• Models, in particular bonding models, are an integral part bonding in H2+ comes exclusively from the interference of the
of chemical research. Chemical research depends on the wave function, and the orbital interactions ΔEorb provide 100%
existence of bonding models, which provide an ordering of the attractive contribution to ΔEint. The electrostatic
scheme for the infinite number of molecules and interactions ΔEelstat at the equilibrium distance are repulsive. A
reactions. naı̈ve view might anticipate some Coulombic attraction of the
• The numerical outcome of modern partitioning schemes electron of the hydrogen atom by H+. This is misleading,
is not measurable quantities, but they are very useful tools because the electronic charge would be withdrawn from the
8795 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

vicinity of the nucleus toward the internuclear region. It can in H2+ originating from the hydrogen atom ecompasses the
easily be shown that the electrostatic attraction in the proton and the bonding area.
internuclear region is weaker than in the core region, because The bond formation of the valence isoelectronic species Li2+
the distance to the nuclei becomes larger. The stabilization and Li2 may superficially be expected to exhibit similar features,
comes from the lowering of the kinetic energy due to the but there are important differences between the valence
interference of the wave function.170,171 This is the physical electrons of hydrogen and lithium. Below the 1s shell of
origin of ΔEorb.172 hydrogen there is the bare nucleus, whereas below the 2s shell of
The second electron in H2 enhances the bond energy to a lithium there is a polarizable 1s core. Charge flow of the valence
smaller extent than the first electron in H2+. The electrostatic electrons of H during bond formation of H2 leaves the remaining
interactions ΔEelstat at the equilibrium geometry of H2 are electronic charge under the Coulombic attraction of the nucleus,
repulsive, but only to a minor extent that does not significantly which induces a rather large contraction of the 1s AO.173 In
influence the overall internal energy. They become weakly contrast, the remaining 2s valence electrons of lithium in Li2
attractive only at longer distances. The deformation densities, encounter Pauli repulsion by the 1s shell, which prevents
which are associated with the bond formation in H2+ and H2, are penetration toward the nucleus. This leads to significant
shown in Figure 4. The direction of the charge flow from the differences of the physical mechanisms of the bond formation
between the lithium and the hydrogen dimers. This becomes
obvious by inspection of the EDA-NOCV results in Table 1 and
the associated deformation densities in Figure 4.
The bond in Li2 is much weaker (De = 20.5 kcal/mol) than in
H2 (De = 112.9 kcal/mol). This can be mainly attributed to the
more diffuse 2s AO of lithium and the less weakly bonded 2s
electron in Li compared with the 1s electron in hydrogen. The
ionization energy (IE) of hydrogen atom is 13.60 eV whereas the
IE of Li is only 5.39 eV. This induces a significantly smaller
energy lowering through the interference of the wave
functions.173 The most surprising feature is the stronger one-
electron bond in Li2+ compared with the two-electron bond in
Li2. The rather soft singly occupied 2s AO of lithium becomes
strongly polarized by the positive charge and it mixes with the
2p(σ) orbital, which leads to enhanced interference of the
atomic wave functions compared with Li2. The deformation
density in Li2+ shown in Figure 4 nicely vizualizes the
polarization of the electronic charge upon bond formation.
Figure 4. Plot of deformation densities Δρ of H2, H2+, Li2, and Li2+ This effect is so strong that the final energetic stabilization in Li2+
species at the RPBE/TZP//BP86/def2-TZVPP level. Energy values are is stronger than in Li2. The second electron weakens particularly
in kcal/mol. The direction of the charge flow is red → blue. the interference of the wave functions, because the interelec-
tronic interactions reduce the polarization, which effectively
weakens the bond. The results for H2+, H2, Li2+, and Li2 are
striking evidence that chemical bonding does not originate from
donating to the accepting areas has the color code red → blue. It spin coupling of the electrons, which is often wrongly
becomes obvious that there is a charge accumulation in H2 from stated.174−176
the hydrogen atoms to the interatomic region. The charge flow The following conclusions arise from this section:

Table 2. Energy Partitioning Analysis of the H−H, N−N, C−O, and B−F Bonds of Diatomics A−B at BP86/TZ2P+a
A−B H2 N2 CO BF
ΔEint −113.0 −240.2 −267.6 −186.0
ΔEPauli 0.0 802.2 582.7 484.4
ΔEelstatb +5.8 −312.8 (30.0%) −240.1 (28.2%) −214.0 (31.9%)
ΔEorbb −118.7 (100%) −729.6 (70.0%) −610.1 (71.8%) −456.4 (68.1%)
ΔEσc −118.7 (100%) −478.7 (65.6%) −310.6 (50.9%) −404.8 (88.7%)
ΔEπc 0.0 −250.9 (34.4%) −299.4 (49.1%) −51.8 (11.3%)
R(A−B)e 0.750 (0.741) 1.102 (1.098) 1.136 (1.128) 1.275 (1.262)
P(A−B) 1.0 3.03 2.30 0.86
μ⃗b,d 0 0 −0.19 (−0.11) −1.03
q(A) 0 0 +0.46 +0.54
De −111.6 −237.6 −265.6 −184.3
D0e −105.0 (−103.3) −234.2 (−225.0) −262.6 (−255.7 ± 1) −182.4 (179.9 ± 3)

a
The interacting fragments of N2, CO, and BF are shown in Figure 6. Energy values in kcal/mol, bond lengths R in Å, dipole moments μ⃗ in Debye.
NBO atomic partial charge q(A), and Wiberg bond order P(A−B). Bond dissociation energies De and ZPE corrected values D0. bThe values in
parentheses give the percentage contribution to the total attractive interactions ΔEelstat + ΔEorb. cThe values in parentheses give the percentage
contribution to the total orbital interactions ΔEorb. dA negative value indicates that the first atom of A−B is the negative end of the dipole moment.
e
Experimental values from ref 24 are given in parentheses.

8796 DOI: 10.1021/acs.chemrev.8b00722


Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

• The covalent bond in Li2 is much weaker than in H2,


because the less strongly bonded 2s electron of Li,
shielded by the 1s core electrons, is much more diffuse
than the 1s electron of H and encounters weaker
stabilization through interference.
• The one-electron bond in Li2+ is stronger than the two-
electron bond in Li2, because the induction enhances the
interference of the wave functions. This is striking
evidence that covalent bonding does not come from
electron pairing but from electron-sharing.
• The charge flow associated with the covalent bond
formation can nicely be visualized using the deformation
density from EDA-NOCV calculations.
7.2. H2, N2, CO, and BF
The isoelectronic molecules N2, CO, and BF are well suited to
discuss the results of a quantum chemical bonding analysis in
comparison with traditional bonding models. Table 2 shows the
numerical NBO and EDA-NOCV results of the three diatomic
species along with H2, which demonstrates the exceptional
features of dihydrogen. The chemical bond of H2 is the standard
model for introducing covalent bonding, because the occurrence
of bonding in terms of interfering wave functions can best be
studied by choosing the simplest example. The data in Table 2
also show the calculated and experimental bond lengths and
BDEs, which are in good agreement.
There are three major qualitative differences when one goes
from H2 to N2. One difference is the occurrence of Pauli
repulsion between electrons with the same spin. The very large
value of ΔEPauli = 802.2 kcal/mol in N2, which is the largest
absolute value of all energy components, demonstrates the
relevance of the Pauli repulsion for chemical bonding. The
second difference is the large attractive Coulombic attraction of
ΔEelstat = −312.8 kcal/mol, which contributes 30% to the total
interatomic attraction in N2. This comes from the overlap of the
electron density of one nitrogen atom with the nucleus of the
other nitrogen atom, which carries a much larger positive charge
than hydrogen. A systematic study of 224 homo- and
heterodiatomic molecules by Spackman and Maslen in 1986
showed that the quasiclassical Coulomb attraction in nearly all Figure 5. (a) Overlap integrals of the atomic 2s and 2p orbitals of N2 as
of them is quite strong.177 The rather weak electrostatic a function of the interatomic distance. (b) Calculated EDA values for
attraction in H2 at longer bond lengths, which becomes repulsive N2 as a function of the interatomic distance. The reference value re = 0.0
is the calculated equilibrium bond length 1.102 Å. Reprinted with
at equilibrium distances, comes from the small nuclear charge of permission from ref 28. Copyright 2006 John Wiley and Sons.
hydrogen. A mathematical explanation for this finding was given
by Kutzelnigg178 and later by Bickelhaupt and Baerends.131 The
third major difference concerns the occurrence of π bonding in distance and is rather found at a bond length that is ∼0.2 Å
N2, which is not surprising. The breakdown of the orbital shorter than the equilibrium value. The additional σ bonds
interactions ΔEorb suggests that two-thirds of the covalent should further shorten the bond. The explanation for the
bonding comes from σ orbitals and the remaining one-third equilibrium bond length, which is clearly longer than what one
from π orbitals. This agrees with the general understanding that would expect using the maximum overlap guideline, is given in
π bonds are weaker than σ bonds.179 Figure 5b. It shows the variation of the different energy
The relevance of Pauli repulsion for chemical bonding comes components as a function of the interatomic distance. All
clearly to the fore when the factors determining the equilibrium attractive components of ΔEint still become stronger when the
length of a chemical bond are considered. A popular explanation interatomic distance becomes shorter than the equilibrium
from the early days of quantum chemistry suggests that the bond value. It is the Pauli repulsion ΔEPauli which prevents further
length of a σ bond, re, is mainly determined by the maximum shortening of re below the equilibrium distance. It is only in H2 that
overlap of the orbitals. The bond becomes further shortened the equilibrium bond length is determined by the maximum
when there is an additional π bond, because the maximum overlap, because there is no Pauli repulsion in dihydrogen.180
overlap of a π orbital is at r = 0. This explanation is, however, not The electronic structure of CO exhibits peculiar features,
valid as the strong effect of Pauli repulsion is ignored. Figure 5a which can be explained with the help of the EDA-NOCV results.
shows the overlap of the σ and π orbitals of N2 at different CO has a stronger bond than N2; the BDE of CO (De = 265.6
interatomic distances r. It becomes obvious that the maximum kcal/mol) is larger than that for N2 (De = 237.6 kcal/mol). The
overlap of the σ orbitals is not yet reached at equilibrium inspection of the energy terms shows that the attractive
8797 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Figure 6. Schematic representation of the most important Lewis structures of N2, CO, and BF and the electron configurations of the atoms which lead
to electron-sharing bonds A−B or dative bonds A→B. Reprinted with permission from ref 152. Copyright 2019 Springer Nature.

Figure 7. Contour line diagram showing the Laplacian ∇2ρ(r) of (a) CO and (b) BF at BP86/def2-TZVPP. Blue solid lines indicate areas of charge
depletion (∇2ρ(r) > 0), and red solid lines indicate areas of charge accumulation (∇2ρ(r) < 0). The bond critical point is shown in black. Solid lines
connecting the atomic nuclei are the bond paths, while the solid lines crossing the bond paths indicate the zero-flux lines which separate the donor and
acceptor moieties. The thick green arrow indicates the atomic dipole component at C in CO and B in BF. Reprinted with permission from ref 47.
Copyright 2006 John Wiley and Sons. .

components in CO, i.e., electrostatic attraction ΔEelstat and which one has in mind; this may be the overlap population or the
orbital interaction ΔEorb, are both weaker than in N2. The energetic stabilization of the σ and π bonds.
stronger bond in CO comes from the significantly weaker Pauli There is a third property of CO that has puzzled chemists for
repulsion ΔEPauli than in N2. The strength of ΔEPauli strongly some time and which has led to controversies: The dipole
depends on the overlap of the orbitals, even more than ΔEorb moment of CO is rather small and has its negative end at the less
(see Figure 1).28 The overlap of the unequal σ AOs in CO is electronegative carbon atom (−0.11 D).181 A simple consid-
smaller than the overlap of the equal σ AOs in N2, which leads to eration in terms of electronegativities would suggest a sizable
weaker Pauli repulsion that even compensates for the weaker dipole moment with the negative end at oxygen. The calculation
Coulomb attraction in the former molecule. of the atomic partial charges with the NBO method gives indeed
The breakdown of ΔEorb of CO into the σ and π components a negative charge at oxygen and a positive charge at carbon
suggests that the latter contribution provides 49% of the orbital (Table 2). The apparent contradiction is easily resolved when
interactions; that is, σ and π bonding have nearly the same one recalls that dipole moments are vector properties whereas
strength. This is in contrast to N2, where π bonding contributes partial charges are scalar properties. The crucial quantity here is
only 34% of ΔEorb. This can be explained with the different types the topography of the charge distribution. This becomes obvious
of covalent interactions in CO and N2. Figure 6 shows by inspection of the Laplacian distribution ∇2ρ(r) of CO, which
schematically the interacting orbitals in the two molecules. is shown in Figure 7a. The electronic charge in the internuclear
The σ and π bonds in N2 come from electron-sharing region is clearly polarized toward the more electronegative
interactions between the nitrogen atoms in the 4S ground oxygen atom. This gives a bond charge component to the dipole
state. In contrast, the interactions between carbon and oxygen in moment that is directed toward oxygen. The key components
the 3P ground states lead to a dative σ bond C←O and two are the charges around the nuclei. The charge concentration
electron-sharing π bonds CO, which agrees with the original around the oxygen nucleus is spherically symmetric, but the
description suggested by Sidgwick.46 Since dative bonds charge concentration around the carbon atom has a local center
are always weaker than electron-sharing bonds between the of charge concentration, which is rather far away from the
same atoms (see the section below), the percentage contribution nucleus. This is indicated by the green arrow. This local charge
of π bonding in CO is higher than in N2. At the same time, the can be associated with the lone-pair electrons at carbon, which
bond order of CO becomes smaller (2.30) than that in N2 strongly influence the chemical reactivity of CO. The aspherical
(3.03), because the overlap of the uneven orbitals is smaller than charge distribution at carbon introduces a large dipole
those of even orbitals. The answer to the question concerning component toward the carbon end, which overcompensates
the multiple bond character of a bond depends on the criterion the effect of the polarity on the dipole moment. A breakdown of
8798 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

the total dipole moment into individual orbital components


showed that all valence orbitals have a dipole component C → O
except the HOMO (Highest Occupied Molecular Orbital),
which has a large component in the opposite direction C←O
that overcompensates the former (Table 3).47 The unexpected
direction of the dipole moment of CO comes from the
topography of the charge distribution.

Table 3. Orbital Components to the Total Dipole Moment of


CO at BP86/6-311++G(3df, 3pd)a
Orbital Bond Moment
MO 1 σ 4. 67
MO 2 σ −6. 23
MO 3 σ 1. 57
MO 4 σ 4. 52
MO 5 π 1. 75
MO 6 π 1. 75
MO 7 σ (HOMO) −8. 21
∑ −0.18 (exp. −0.11)
a
Values in Debye. Negative values indicate that the negative end is at
the carbon atom. The origin of the CO12+ ion is chosen at the position
where the dipole moment of the nuclei becomes zero. The orbital
components were calculated with nuclear charges which are scaled by
6/7 (carbon) and 8/7 (oxygen). This leads to orbital components of
the dipole which after summation give the total dipole moment of the
neutral molecule, which is origin independent. The data are taken
from ref 47. Figure 8. MO correlation diagram of the 2s and 2p AOs of first octal-
row atoms. Reprinted with permission from ref 28. Copyright 2006
Table 2 shows that the BDE of BF is much smaller (De = 184.3 John Wiley and Sons.
kcal/mol) than that of isoelectronic N2 and CO. All three energy
terms of the EDA-NOCV are weaker in BF than in N2 and CO, and Pauli repulsion, but not by the maximum orbital
and therefore, they do not explain the overall reason for the bond overlap.
weakening. A possible rationalization of the bond weakening • CO has a triple bond, which consists of a dative σ bond
comes from the strength of the σ and π bonds in BF. The and two electron-sharing π bonds . The stronger
absolute and relative contribution of π bonding in BF is much bond in CO than in N2 can be explained with the weaker
smaller than in the other isoelectronic dimers. Inspection of the Pauli repulsion in carbon monoxide.
interacting orbitals explains the finding. Figure 6 shows that the
π interactions in BF come from dative bonding from the • BF has an electron-sharing B−F σ bond, which is
very electronegative fluorine atom to the electropositive boron, supported by two relatively weak dative π bonds .
whereas the σ bond is due to electron-sharing interactions in B− The weak π bonds explain why BF is more weakly bonded
F. Since the former provide only 11% to the covalent bond, it than N2 and CO.
seems reasonable to describe the molecule with a Lewis • The dipole moments of CO (−0.11 D) and BF (−1.03
structure that has only a single bond. The nature of the σ and D), which have their negative ends at the respectively
π bonding components in BF are the opposite of CO. more electropositive atoms C and B, are due to the
The electronic structure of BF nicely supports the above anisotropic charge distribution in the molecules. The σ
explanation for the dipole moment in CO. The Laplacian lone-pair electrons induce a local vector component,
distribution ∇2ρ(r) of BF shown in Figure 7b shows that the which leads to the surprising direction of the dipole
charge distribution in the bonding region is even more polarized moments.
toward fluorine due to the larger difference in electronegativities 7.3. First Octal-Row Sweep Li2−F2 and Related Molecules
compared to CO. However, the local charge concentration at
The physical nature of the chemical bond can be best
boron in BF has a larger distance from the nucleus than the
understood by analyzing the interatomic interactions in
charge concentration at carbon in CO, because the lone-pair
diatomic molecules. Polyatomic molecules introduce a plethora
orbital at boron is more diffuse than the carbon lone-pair in CO.
of variations, which are the infinite playground of chemistry, but
As a result, the total dipole moment of BF is even larger (−1.03
they can significantly change the essential features of chemical
D) than that of CO with the negative end at boron.
bonding. Particular classes of compounds, which exhibit specific
The following conclusions arise from this section:
types of chemical bonds such as clusters and aromatic systems,
• The appearance of valence electrons that possess the same are discussed in other articles of this thematic issue. We focus
spin introduces strong Pauli repulsion in N2, which is mainly on diatomic molecules and we extend our above
missing in H2. presented studies to a systematic first-row sweep of Li2−F2,
• The equilibrium bond lengths in molecules that have where we try to connect classical bonding models with a
more than two electrons is determined by the equilibrium quantum chemical analysis. We also discuss recent theoretical
between Coulombic forces, attractive orbital interactions, studies of low-valent main-group compounds, which possess
8799 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Table 4. Energy Partitioning Analysis of the First Row Dimers E2 (E = Li−F) in C2v at BP86/TZ2Pa
E Li Be B C N O F
el. State 1
Σg+ 1
Σg+ 3
Σg − 1
Σg+ 1
Σg+ 3
Σg− 1
Σg+
ΔEint −20.7 −7.9 −74.7 −140.8 −240.2 −141.9 −52.9
ΔEPauli 1.8 41.6 135.0 252.2 802.4 464.9 146.1
ΔEelstatb −8.3 (36.9%) −17.9 (36.1%) −33.1 (15.8%) −3.2 (0.8%) −312.9 (30.0%) −159.7 (26.3%) −41.0 (20.7%)
ΔEorbb −14.2 (63.1%) −31.6 (63.9%) −176.5 (84.2%) −389.8 (99.2%) −729.8 (70.0%) −447.1 (73.7%) −157.8 (79.3%)
ΔΕ(σ)c −14.2 −31.6 −104.5 (59.2%) −201.7 (51.8%) −478.8 (65.6%) −319.5 (71.5%) −151.5 (96.0%)
ΔΕ(π)c 0.00 0.00 −72.0 (40.8%) −188.0 (48.2%) −251.0 (34.4%) −127.6 (28.5%) −6.2 (4.0%)
ΔEcorr.f 0.3 0.00 1.9 3.3 4.2 4.8 2.7
De d 20.4 (24.6) 7.9 (2.2; 2.7[calc.])e 72.8 (71.2) 137.5 (145.9) 236.1 (228.4) 137.1 (120.2) 50.2 (38.3)
E−Ed 2.731 (2.673) 2.442 (2.45)e 1.617 (1.590) 1.253 (1.243) 1.102 (1.098) 1.224 (1.208) 1.420 (1.412)
a
Energies in kcal/mol, distances E−E in Å. Values are taken from ref 28. bValues in parentheses give the percentage contribution to the total
attractive interactions ΔEelstat + ΔEorb. cValues in parentheses give the percentage contribution to the total orbital interactions ΔEorb. dExperimental
values in parentheses from ref 24, unless otherwise specified. eExperimental value for De and E−E; Bondybey, V. E. Chem. Phys. Lett. 1984, 109,
436; calculated value for De; Martin, J. M. L. Chem. Phys. Lett. 1999, 303, 399. fCorrection for the spin polarization.

Figure 9. Schematic representation of the electron configuration and the orientation of the atoms with the chosen orbital populations for the EDA.
Note that for the 3Σg− states of Si2, O2, and S2 a spin change of one electron in a singly occupied p(π) orbital takes place in the EDA calculation.
Reprinted with permission from ref 28. Copyright 2006 John Wiley and Sons.

chemical bonds that can be understood by using the undistorted elements becomes slightly higher in energy than the 1πu MO.
E2 fragments as reference. This explains why B2 has a X3Σg− ground state and C2 has a
Figure 8 shows a correlation diagram of the (n)s(n)p valence X1Σg+ ground state.24 The hybridization lowers the antibonding
AOs of main-group atoms of the first octal row Li−F with the character of the 2σu+ MO and raises the antibonding strength of
MOs of the diatomic molecules. The shape and the splitting of the 3σu+ MO. Thus, the strengths of the bonding 2σg+ MO and
the MOs consider the energy levels of the AOs and the antibonding 2σu+ MO do not fully cancel. However, the energy
symmetry-allowed mixing (hybridization) of the orbitals. The difference between the 3σg+ MO and the 1πu MO is very small
hybridization lowers the energy level of the bonding 2σg+ MO and the ordering depends on the nature of the atoms. The
and raises the energy of the bonding 3σg+ MO, which for some analogous valence 4σg+ MO and 2πu MO of Si2 exhibit a reverse
8800 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

order; that is, the 4σg+ is below the 2πu. Thus, the ground state of
Si2 is the X3Σg− triplet state. On the other hand, Al2 has the same
X3Σg− ground state as B2.24
7.3.1. Li2, Be2 and Related Molecules. The MO diagram
in Figure 8 and the aufbau principle are a useful starting point for
the discussion of chemical bonds in diatomic molecules Li2−F2.
But they are not sufficient for understanding all features of the
bonds, as was shown above for the comparison of Li2 with Li2+. It
must be complemented by a quantum chemical analysis, which
considers all types of interatomic interactions. Table 4 gives the
results of EDA calculations of Li2−F2.28 The comparison of the
calculated bond lengths and bond dissociation energies shows
the same trends, which makes the data a reasonbale fundament
for the bonding analysis. It is important to consider the
geometrical arrangement of the atomic configurations for the
interatomic interactions, which are shown in Figure 9, in order
to understand the calculated numbers. The construction of the
AO arrangements corresponds to the MO correlation diagram.
The EDA results for Li2 were already given in Table 1, albeit
with a different DFT functional. A comparison with the numbers
in Table 4 shows that the results change very little when a
different functional is used. The EDA values for Be2 suggest very
weak bonding, which is due to the strong Pauli repulsion
compensating most of the electrostatic attraction and covalent
contribution. The net attration of ΔEa1(σ) = −31.6 kcal/mol in
Be2 reveals that the hybridization in the bonding 2σg+ MO and
antibonding 2σu+ MO yields a net stabilization, which is
annihilated by the Pauli repulsion. The naı̈ve view of a bond
order of zero for Be2 is only supported when the Pauli repulsion Figure 10. Shape and eigenvalues of the lowest unoccupied Hartree−
is considered. Fock MO (LUMO) and the relevant occupied valence orbitals of Be2F2
using a cc-pVTZ basis set.182 Reprinted with permission from ref 182.
Inspection of the occupied valence MOs in Be2 gives a hint Copyright 2016 John Wiley and Sons.
toward attempts to stabilize molecules that posssess a Be−Be
bond. Substituents which withdraw electronic charge from the
antibonding 2σu+ MO but donate electronic charge into the repulsion in B2 (X3Σg−) comes only from electrons in the σ
bonding 2σg+ MO should enhance the Be−Be bond. This is space, the total σ interactions, i.e., attraction due to interference
indeed achieved in Be2F2, which has a theoretically predicted of the wave functions (ΔEorb) plus repulsion due to spatial
bond dissociation energy of D0298 = 65.0 kcal/mol at the exclusion of electrons with the same spin (ΔEPauli), become
CCSD(T)/cc-VTZ level of theory.182 Figure 10 shows the destabilizing. The simple bonding model of the MO correlation
valence MOs of the molecule. The antibonding inner lobes of Be diagram in terms of only π bonding in B2 is thus supported when
in the HOMO−4 have nearly completely disappeared whereas the Pauli repulsion is considered.
the HOMO is a bonding orbital. This explains the surprisingly The analysis of the orbital interactions in B2 was the starting
large strength of the Be−Be bond in Be2F2, which is also point for the eventually successful synthesis of a molecule with a
predicted to be thermodynamically stable relative to Be + BeF2 genuine BB triple bond. Calculations of group-13 complexes
in the gas phase. The isolation of Be2F2 in a bulk is energetically NHCMe→E2←NHCMe (NHCMe = N-heterocyclic carbene with
unfavorable, because the large atomization energy of beryllium methyl groups at nitrogen) with E = B, Al, Ga, In predicted that
from the metal (ΔHf = 77.4 kcal/mol)183 and the heat of the boron species B2(NHCMe)2 has a linear coordinated
sublimation of BeF2 (ΔHsub = 55.6 kcal/mol)184 make it unlikely geometry with a very short B−B distance.185 A bonding analysis
that pure beryllium subfluoride can be prepared. But Be2F2 suggested that the B2 species in B2(NHCMe)2 is in the
should be a stable molecule in the gas phase, and it may also be electronically excited (3)1Σg+ reference state, which has two
isolated in an inert matrix.182 vacant σ MOs with suitable symmetry that may serve as acceptor
7.3.2. B2 and Related Molecules. According to the simple orbitals for donor−acceptor interactions with the NHCMe
MO correlation diagram, most of the covalent bonding in the ligands (Figure 11). The energy calculations predicted that
X3Σg− ground state of B2 comes from π bonding, because the B2(NHCMe)2 is thermodynamically stable with regard to
contributions from the bonding 2σg+ MO and antibonding 2σu+ dissociation into B2 in the electronic X3Σg− ground state and
MO cancel to some extent. The EDA calculation of B2 suggests two NHCMe ligands by >180 kcal/mol. The bond strength of the
that the overall bonding comes mainly from the ΔEorb term, intrinsic interactions NHCMe→E2←NHCMe clearly compen-
which comprises 60% σ bonding and 40% π bonding. The π sates for the excitation energy of B2 for the process X3Σg−→
interactions come from the two singly occupied degenerate π (3)1Σg+, which amounts to 106.4 kcal/mol.
orbitals. Thus, the hybridization of the bonding 2σg+ MO and The theoretically predicted stability of B2(NHCMe)2 was
antibonding 2σu+ MO in B2 is very effective and leads to a verified by the synthesis and X-ray structural characterization of
predominantly σ bonded species, if only the stabilizing orbital B 2 (NHC R ) 2 carrying bulky groups R at nitrogen by
interactions are considered. A different interpretation arises Braunschweig and co-workers.186,187 The experimental value
when the Pauli repulsion is taken into account. Since Pauli for the B−B distance of 1.45 Å was in very good agreement with
8801 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Figure 11. Schematic diagram of the valence orbitals and orbital occupation of B2 in (a) the X3Σg− ground state and (b) (3)1Σg+ excited state. (c)
Schematic representation of the charge donation from the out-of-phase (+,−) and in-phase (+,+) combinations of the ligand lone-pair σ orbitals into
the vacant orbitals of B2 in the 1Σg+ excited state. (d) Calculated energies of the excitation energy of B2 from the ground state to the reference state and
bond dissociation energy De of B2(NHCMe)2 at BP86/def2-TZVPP.189 Reprinted with permission from ref 189. Copyright 2015 Royal Society of
Chemistry.

the calculated data of 1.47 Å for the model compound. But the detailed insight into the change of the electronic structure of the
notation of a genuine BB triple bond in B2(NHCR)2 was fragments that comes from the formation of the bonds. The
challenged on the basis of experimental results, where the strongest interaction ΔEσ1 (−112.7 kcal/mol) comes from the
boron−boron distance in OBBO was taken as reference for a in-phase (+,+) NHCMe→B2← NHCMe σ-donation into the
single bond.188 This led to a detailed bonding analysis of LUMO+1 of (3)1Σg+ B2 while the contribution of the out-of-
B2(NHCMe)2 with the EDA-NOCV method, which convinc- phase (+,−) donation ΔEσ2 (−86.9 kcal/mol) into the lower-
ingly demonstrated that the molecules B2(NHCR)2 have a lying LUMO (Lowest Unoccupied Molecular Orbital) is a bit
boron−boron triple bond.189 It was shown that there is a weaker due to the better overlap of the former (+,+) donation.
significant π-bonding contribution in the boron−boron bond of The other two contributions ΔEπ1 and ΔEπ2 arise from
OBBO which makes it unsuitable as reference for a single bond. NHCMe←B2→NHCMe π-backdonation, which is smaller than
Table 5 gives the numerical EDA-NOCV results for the σ-donation. They are not degenerate, because the planes of
B2(NHCMe)2. Figure 12 displays the deformation densities of the NHC ligands are in a twisted conformation.
the four most important pairwise orbital interactions Δρ1−4 The shape of the deformation densities shows not only the
along with the interacting orbitals of the fragments and the formation of the boron−ligand bonds but also the alteration of
resulting MOs of the complex. The combined pictures provide the boron−boron bond. The shapes of Δρ1 and Δρ2 and the
8802 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Table 5. Results of the EDA-NOCV calculations for NHCMe amino carbene) ligand enhances the π backdonation to the
→ B2 ← NHCMe at BP86/TZ2P Using the Fragments ligand, which leads to a longer B−B and shorter B-L
B2[(3)1Σg+] and (NHCMe)2 as Interacting Speciesa bonds.190,191
We point out that the compound B2(NHCR)2 that was
B2(NHCMe)2
isolated by Braunschweig in 2012 was not the first L→B2←L
Interacting fragments B2[(3) Σg+] and (NHCMe)2
1
species experimentally observed. Zhou and co-workers reported
ΔEint −307.5 in 2002 the synthesis of the dicarbonyl complex B2(CO)2 in a
ΔEPauli 259.0 low-temperature argon matrix.192 The experimental and
ΔEelstatb −252.3 (44.5%) theoretical vibrational spectra suggest that the molecule has a
ΔEorbb −314.2 (55.5%) linear structure. The authors proposed that B2(CO)2 exhibits
ΔEσ1 L→(B2)←L (+,+) donationc −112.7 (35.9%) some boron−boron triple bond character, and they suggested
ΔEσ2 L→(B2)←L (+,-) donationc −86.9 (27.7%) that the molecule should be drawn as OCBBCO. A
ΔEπ1 L←(B2)→L π backdonationc −48.0 (15.3%) similar conclusion was drawn by Li and co-workers, who
ΔEπ2 L←(B2)→L π′ backdonationc −42.4 (13.5%)
observed the boron compound [B2(BO)2]− in the gas phase
ΔErest −24.2 (7.7%)
using photoelectron spectroscopy.193 The authors reported
q(B2) −0.36
a
quantum chemical calculations of the series [B2(BO)2]0,−,2−,
All energy values in kcal/mol. Calculated NBO partial charge of B2 in with the dianion being isoelectronic with B2(CO)2. Subsequent
e. bThe value in parentheses gives the percentage contribution to the
theoretical studies of B2L2 with various ligands supported the
total attractive interactions ΔEelstat + ΔEorb. cThe value in parentheses
gives the percentage contribution to the total orbital interactions description of the bonding situation in terms of dative bonds
ΔEorb. L→B2←L. EDA calculations of B2(CO)2 showed substantial π
backdonation OC←B2→CO, which are nearly as strong as the σ
associated σ orbitals HOMO−6 and HOMO−13 of the donation OC→B2← CO.194
complex (Figure 12a, b) suggest that the NHCMe →B Table 6 shows the EDA results of B2L2 species with
interactions concurrently exert an opposite effect on the isoelectronic ligands L = CO, N2, BO−, not only for the dative
boron−boron bonding. The HOMO−6 is bonding and bonds between the ligands and B2 but also for the boron−boron
HOMO−13 is antibonding with respect to the B−B bonds. bonds. It becomes obvious that all bonds have a significant
Thus, the effect of the σ interactions NHCMe→B2←NHCMe on triple-bond character, which let the authors propose that the
the B−B σ bond partially cancel each other. In contrast, the LBBL systems are molecules with all-triple bonds.
effect of the π backdonation NHCMe←B2→NHCMe clearly 7.3.3. C2 and Related Molecules. The EDA results for the
weakens the B−B bond, which becomes visible by inspection of X1Σg+ ground state of C2 suggest that the attractive σ and π
Δρ3 and Δρ4 (Figure 12c, d). This has been utilized to fine-tune orbital interactions have nearly the same strength (Table 4). The
the B−B bond by variation of the π-acceptor strength of ligands attraction by the σ orbitals is compensated by the Pauli repulsion
L in L→B2← L. Replacement of the unsaturated NHC ligand by of the doubly occupied σ MOs, which leaves the π contribution
the saturated homologue NHCsat or the CAAC (cyclic alkyl as the sole component of the stabilizing orbital interactions. But

Figure 12. Plot of the interacting donor and acceptor orbitals and calculated eigenvalues ε of (NHCMe)2 and (1Σg+) B2 (right two columns) and
matching MOs of the complex NHCMe→BB←NHCMe (second column from the left). Plot of the deformation densities Δρ with connected
stabilization energies ΔE of the four most important orbital interactions in B2(NHCMe)2 which indicate the associated charge flow red → blue.189
Reprinted with permission from ref 189. Copyright 2015 Royal Society of Chemistry.

8803 DOI: 10.1021/acs.chemrev.8b00722


Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Table 6. EDA Results of LBBL at the BP86/TZ2P Levelba


OCBBCO N2BBN2 [OBBBBO]2− OCBBCO N2BBN2 [OBBBBO]2−
4 − 4 − − 4 −
Interacting 2 BCO ( Σ ) 2 BNN ( Σ ) 2 [BBO] ( Σ ) B2 [(3) Σg ] + 2 CO ( Σ )
1 + 1 +
B2 [(3) Σg ] + 2 N2 ( Σg )
1 + 1 +
B2 [(3)1Σg+] + 2 BO− (1Σ+)
fragments
ΔEint −155.2 −155.3 −89.7 −267.7 −223.1 −326.1
ΔEPauli 103.9 93.6 144.9 203.6 221.3 195.0
ΔEelstatc −108.7 −97.4 (39.1%) −71.0 (30.2%) −119.1 (25.3%) −94.5 (21.3%) −209.0 (40.1%)
(42.0%)
ΔEorbc −150.4 −151.5 −163.6 (69.8%) −352.2 (74.7%) −349.9 (78.7%) −312.2 (59.9%)
(58.0%) (60.9%)
ΔEσd −92.1 (61.2%) −91.3 (60.3%) −102.2 (62.5%) −193.7 (55.0%) −168.3 (48.1%) −233.1 (74.7%)
ΔEπd −58.3 (38.8%) −60.2 (39.7%) −61.4 (37.5%) −158.5 (45.0%) −181.6 (51.9%) −79.1 (25.3%)
ΔEprep 5.5 10.2 6.7 6.1 (112.4)e 9.8 (123.1)e 7.6 (180.7)e
ΔE(=-De) 149.7 145.1 83.0 261.6 (155.3)e 213.3 (107.0)e 318.5 (145.4)e
a
Values are taken from ref 194. bEnergy values in kcal/mol. cThe percentage values in parentheses give the contribution to total attractive
interactions ΔEorb + ΔEelstat. dThe percentage values in parentheses give the contribution to total orbital interactions ΔEorb. eEnergy with respect to
B2 (X3Σ g−) ground state.

Figure 13. Coefficients and weights of the four most important electron configurations a−d in the full-valence CAS(8/8)SCF/cc-pVTZ calculation of
(X1Σg+) C2.195 Reprinted with permission from ref 195. Copyright 2016 John Wiley and Sons.

the bonding in C2 is not well described with only one CH. According to their VB calculations, the strength of the σ
configuration, because the occupied 1πu MO and the vacant interactions in C2 is 156.6 kcal/mol while σ bonding in acetylene
3σg+ MO are close in energy, which requests a multiconfigura- amounts to 138.7 kcal/mol.206−208
tional calculation of C2 in order to reliably account for the Following the reasoning of Shaik et al., the carbon−carbon
bonding situation. A (8/8) CASSCF (Complete Active Space) triple bond in HCCH stays essentially intact when the
calculation, where the 8 valence electrons occupy the complete hydrogen atoms are removed and the remaining unpaired
space of the 2s2p valence orbitals of C2, showed that the leading electrons couple with their inward lobe, which leads to a second
configuration (2σg+)2(2σu+)2(1πu)2(1πu′)2(3σg+)0(1πg)0 (Fig- σ bond that is weaker than the first one but strong enough to
ure 13a) has a weight of 71% of the total wave function.195 The establish a fourth bond. The alternative view, which is based on a
next important configuration, where only bonding σ and π MOs CASSCF wave function, identifies a significant change in the σ
are occupied, contributes 13.6% to the wave function (Figure space of the remaining C2 fragment of acetylene after removing
13b). The remaining configurations are less important. The the hydrogen atoms.195 The σ interactions comprise indeed two
formal bond order of 2, which comes from the π bonds in the rather than one component, but they are degenerate and very
leading configuration, becomes slightly higher by the second weak. The overall σ interactions in C2 are much weaker than in
important configuration that has a formal bond order of 4. acetylene and the bonding situation may be described like a Be2
The nature and strength of the bond in C2 has been a topic of species that is held together by a “π-clip”.209
an intensive and controversial debate in recent years, following Table 7 shows some relevant properties of C2 and HCCH
the suggestion by Shaik and co-workers that dicarbon has a for comparison. The carbon−carbon bond in C2 is longer and
quadruple bond comprising two σ and two π bonds.196−199 The has a smaller force constant and lower vibrational frequency than
assignment of a quadruple bond in C2 came from an analysis of in acetylene. Dicarbon has also a significantly lower bond
VB calculations. Several groups, who employed various MO and dissociation energy (BDE). The BDE may be questioned as a
VB methods, questioned in subsequent studies the suggestion of measure of the bond strength, because the electronic reference
a quadruple bond in C2.200−205 Calculations of bond orders state of the dissociated fragments is not the same as in the
using different approaches gave values between 2−4, and thus, molecule. However, Figure 14 shows that the energy curve for
they were not at all conclusive. The situation became variation of the C−C distance in HnCCHn (n = 0−3) places C2
substantiated when Shaik et al. quantified the strength of the between ethylene and acetylene. Thus, all observable quantities
intrinsic σ interactions in C2 and triply bonded acetylene HC suggest that the carbon−carbon bond in C2 is weaker than in
8804 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Table 7. Calculated Properties of C2 and HCCHa finally led to the isolation of the compounds C2(CAACR)2 with
different substituents R by two groups.214,215 The X-ray
C2 HCCH
structure analysis showed that the molecule slightly deviates
Re(C−C) [Å] 1.256 1.216 from a linear arrangement of the ligands along the central C2 axis.
kCC [mdyn/Å] 11.9 15.8 The compounds may thus either be described as cumulene
υCC [cm−1] 1837 1995 CAACCCCAAC or as complexes CAAC→CC←
De(C−C) [kcal/mol] 144.0 228.2 CAAC. EDA-NOCV calculations were carried out with the
a
Equilibrium distance Re(C−C), force constant kcc, vibrational fragments C2 and (CAAC)2 in the electronic singlet states for
frequencies υCC, and bond dissociation energy De. Values are taken dative bonding and quintet states for electron-sharing bonding
from refs 195 and 200. in the cumulene. Figure 15 shows schematically the electronic
reference states of C2. The (2)1Δg state, which has two vacant σ
HCCH. A recent theoretical study of the 13C isotropic orbitals, is the reference state for dative bonding L→CC←L
magnetic shieldings came to the conclusion that the bond in C2 while the 5Δg state with four unpaired electrons in two σ and two
has a higher multiplicity but it is weaker than in HCCH.210 π orbitals is the reference state for electron-sharing double
The suggestion that C2 has a quadruple bond that is stronger bonding.216 As mentioned in the section above about the EDA-
than the carbon−carbon triple bond in acetylene appears NOCV method, the absolute values of the ΔEorb term indicate
therefore questionable in the light of the properties of the two the more favorable description of the bonding situation.
molecules. Calculations were also carried out for C2(CAACMe)2 and for
The stabilization of C2 with two donor ligands in compounds the related molecule C2(NHCMe)2. The numerical results of the
C2L2 analogous to B2L2 has been a topic of theoretical and calculations are shown in Table 8.217
experimental studies. The PPh3 bound derivative has been The data in Table 8 suggest that the bonding situation in
generated at −78 °C but was reported to be unstable at C2(NHCMe)2 may be reasonably well described with either
temperatures above −40 °C.211 Theoretical studies suggest that bonding model. The ΔEorb value for dative interactions (−686.0
the stabilization by NHC ligands is stronger than that by kcal/mol) is only slightly smaller than the data for electron-
phosphanes and that compounds C2(NHCR) might be stable at sharing double bonding (−707.9 kcal/mol). In contrast, the
room temperature.212 Deprotonation of the dication [(C2H2)- ΔEorb value for dative interactions in C2(CAACMe)2 is much
(NHCR)]2+ with strong bases failed and gave the reduced larger (−720.5 kcal/mol) than that for electron-sharing double
product instead.213 But the use of the CAAC ligand instead bonding (−601.9 kcal/mol). This implies that the latter

Figure 14. (a) Calculated stretching force constants kCC at the full-valence CASSCF/cc-pVTZ level of HnCCHn (n = 3−0). (b) Potential energy curves
at CASSCF/cc-pVTZ for changing the C−C distances of HnCCHn (n = 3−0) from the equilibrium distance by ±0.15 Å in steps of 0.01 Å.195
Reprinted with permission from ref 195. Copyright 2016 John Wiley and Sons.

8805 DOI: 10.1021/acs.chemrev.8b00722


Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

16. Note that the deformation densities and the associated


energies are the sum of the unequal contributions of the α and β
electrons in opposite directions, with the differences between
them being less important for the present work. Figure 16 shows
only the net deformation densities and the sum of the strength of
the orbital interactions.
The two strongest orbital interactions ΔEσ(+,+) and ΔEσ(+,−)
come from the (+,+) in-phase and (+,−) out-of-phase
combinations of the singly occupied σ orbitals, which lead to
charge accumulation (blue) in the C−C bonding region
between the (CAAC)2 and C2 fragments. The two σ
contributions have similar strength. In contrast, the π
contribution, ΔEπ(+,−), is much larger than ΔEπ(+,+). This is
because the former interaction releases the antibonding π
interaction of the central C2 fragment and introduces π bonding
between C2 and the CAAC ligands, whereas the latter weakens
Figure 15. Schematic diagram of the electronic reference states of C2 the π bonding in the C2 fragment. Note that the energy levels of
for dative bonding and electron-sharing bonding in C2L2. (a) (2)1Δg the interacting singly occupied orbitals related to Δρ3 are much
state; (b) 5Δg state. closer than the SOMOs that are associated to Δρ4.
7.3.4. N2 and Related Molecules. The chemical bond in
Table 8. EDA-NOCV Results at BP86/TZ2P of the Donor− N2 has been discussed in the above section 7.2. Dinitrogen is a
Acceptor and Electron-Sharing Interactions in C2(NHCMe)2 prototype for an electron-sharing triple bond, which consists of a
and C2(CAACMe)2 Using Singlet (S) and Quintet (Q) σ and two degenerate π bonds. The EDA results in Table 4
Fragmentsa suggest that two-thirds of the covalent interactions come from σ
C2(NHCMe)2 C2(CAACMe)2 bonding while one-third is due to π bonding, which nicely agrees
C2 (S) C2 (Q) C2 (S) C2 (Q)
with chemical intuition. Dinitrogen is considered a weak donor
ligand, which usually binds in complexes in the X1Σg+ electronic
Molecule (NHCMe)2 (NHCMe)2 (CAACMe)2 (CAACMe)2 ground state through its σ lone-pair orbitals in an end-on
Fragments (S) (Q) (S) (Q)
fashion.219 The electronic excitation energies of N2 are rather
ΔEint −647.8 −485.8 −712.0 −486.7
high, and thermochemical reactions of N2 normally involve only
ΔEPauli 361.5 622.5 344.0 480.2
the electronic ground state. Dinitrogen as acceptor species was
ΔEelstatb −323.4 −400.5 −335.5 −347.0
(32.0%) (36.1%) (31.8%) (36.6% virtually unknown until recently.
ΔEorbb −686.0 −707.9 −720.5 −601.9 One decade ago, a series of complexes L→E2←L where E2 are
(68.0%) (63.9%) (68.2%) (63.4%) diatomic main-group compounds of groups 14 and 15 stabilized
ΔE1 L-C2-L σ −205.1 −228.0 −220.8 −185.3 by NHC ligands L were isolated and structurally characterized
bonding (+,+)c (29.9%) (32.2%) (30.6%) (30.8%) by X-ray crystallography.220−223 The starting point was the
ΔE2 L-C2-L σ −192.5 −164.5 −191.3 −163.6 synthesis of Si2(NHCR)2 by Robinson in 2008,224 which was
bonding (+,-)c (28.1%) (23.2%) (26.4%) (27.2%)
ΔE3 L-C2-L π*- −200.2 −148.1 −213.2 −146.7
followed by the germanium and tin homologues
bondingc,d (29.2%) (20.9%) (29.6%) (24.4%) Ge2(NHCR)2225 and Sn2(NHCR)2226 in 2009 and 2012. The
ΔE4 L-C2-L π- −42.5 −102.0 −50.7 −52.1 group-15 adducts P2(NHCR)2 and As2(NHCR)2 were isolated in
bondingc,d (6.2%) (14.4%) (7.0%) (8.7%) 2008 and 2010.227,228 A theoretical paper by Wilson and co-
ΔErestc −45.7 −65.3 −44.5 −54.2 workers reported in 2012 the calculated bond dissociation
(6.6%) (9.3%) (6.4%) (8.9%)
a
energies of the group-14 and group-15 complexes E2(L)2 with L
All energies in kcal/mol.217. bThe value in parentheses gives the = NHC, PPh3, which showed that the group-15 adducts are
percentage contribution to the total attractive interactions ΔEelstat + clearly less stable than the group-14 homologues.212 Most
ΔEorb. cThe value in parentheses gives the percentage contribution to
the total orbital interactions ΔEorb. dThe notation π and π* refer to
species should in principle be isolable, except for the dinitrogen
the π MO of the C2 fragment. complexes N2L2, which were calculated to be thermodynami-
cally unstable toward dissociation of the ligands (Table 9,
numbers in bold).
molecule is best described as cumulene CAACCC Inspection of the experimental literature reveals surprising
CAAC, which agrees with the finding that the singlet−triplet findings. The compounds N2(NHCR)2 and N2(PPh3)2 had
split of the CAAC carbene is significantly smaller than that for earlier been reported as stable species that can easily be
NHC.218 The values of ΔEorb in Table 8 are marked in red to prepared.229,230 The experimental study of the synthesis and
highlight the particular relevance of the orbital term, which gives properties of N2(PPh3)2 was in conflict with the theoretical
the best description of the type of orbital interaction. result that the molecule is unstable toward loss of the PPh3
It is instructive to compare the deformation densities, which ligands by more than 80 kcal/mol, but the compound was
are associated with the formation of the electron-sharing σ and π reported to be stable up to 215 °C when it slowly decomposes
bonds in C2(CAACMe)2 with the dative σ and π bonds in after melting.230 The synthesis was reported in 1964, and the
B2(NHCMe)2 (Figure 12). Table 8 shows that the major compound was poorly characterized by giving only its high
components of ΔEorb in C2(CAACMe)2 come from two σ and melting point of 184 °C. It was sketched with the formula
two π orbital interactions. The associated deformation densities Ph3PNNPPh3, which is obviously wrong, giving rise to
Δρ1−4 and the singly occupied orbitals of the fragments along questioning the identity of the compound. A joint experimental
with the resulting MOs of the molecule are displayed in Figure and theoretical study by Holzmann et al. reinvestigated the
8806 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Figure 16. Plot of the deformation densities Δρ of the pairwise orbital interactions for electron-sharing bonding between C2 in the excited 5Δg state
(see Figure 15) and two CAACMe ligands in C2(CAAC)2. (a) (CAACMe)−C2−(CAACMe) σ-interaction between the the 2σg+ MO of C2 and the
SOMO-3 of (CAACMe)2. (b) (CAACMe)−C2−(CAACMe) σ-interaction between the 1σu+ MO of C2 and the SOMO-2 of (CAACMe)2. (c)
(CAACMe)−C2−(CAACMe) π-interaction between the 1πg MO of C2 and the SOMO-1 of (CAACMe)2. (d) (CAACMe)−C2−(CAACMe) π-interaction
between the 1πu MO of C2 and the SOMO of (CAACMe)2. See ref 217.

Table 9. Calculated Free Energies (kcal/mol) at MP2/


TZVPP of the Metal−Ligand Bond Strengths in E2L2a
E2 + L → L-EE-L
E L = NHC L = PPh3
Group 14
C −158.2 −83.6
Si −84.2 −55.0
Ge −71.9 −49.4
Sn −66.8 −48.0
Pb −57.4 −44.4
Group 15
N 22.3 87.8
P −27.4 −16.2
As −21.3 6.4
Sb −25.9 Diss.
Bi −16.0 Diss.
a
The data are taken from ref 212.

structure and properties of the molecule.231 Following the


original simple synthetic protocol followed by an X-ray
measurement showed that the molecule is indeed N2(PPh3)2,
which has, however, a trans bent geometry of the core unit
(Figure 17a).
The inspection of the occupied orbitals (Figure 17b−e)
revealed a surprising electronic structure and bonding situation
of N2(PPh3)2. The out-of-plane π and π* orbitals of the N2
moiety are both occupied (HOMO−14 and HOMO), which Figure 17. (a) X-ray crystal structure of N2(PPh3)2. Selected bond
makes two in-plane MOs of the dinitrogen fragment available as lengths [Å] and angles [deg] with calculated values at RI-BP86/def2-
acceptor orbitals. The HOMO−1 and HOMO−15 are the in- TZVPP in brackets: N(1)−N(1) 1.497(2) [1.441], P(1)−N(1)
1.5819(12) [1.606], P(1)−N(1)−N(1a) 107.10(11) [112.6], P(1)−
phase (+,+) and out-of-phase (+,−) combination of the σ
N(1)−N(1a)−P(1a) 180.0 [180.0]. (b−e) Plot of relevant occupied
donation Ph3P→N2←PPh3. The electronic reference state of N2 orbitals of N2(PPh3)2.231 Reprinted with permission from ref 231.
is the highly excited (1)1Γg state which has the valence Copyright 2013 John Wiley and Sons.
configuration (1σg)2(1σu)2(1πu)2(2σg)2(1πg)2, where the 1πu′
and 1πg′ are vacant (Figure 18b). There is a formal excitation of
two electrons 1πu′ → 1πg from the X1Σg+ ground state (Figure combination of the σ lone-pair MOs of the PPh3 ligands. The
18a) to the (1)1Γg excited state. The bonding 1πu′ MO is the 1πu′ MO of N2 is much lower in energy than the antibonding
acceptor orbital for donation from the out-of-phase (+,−) 1πg′ orbital, which makes the associated HOMO−15 orbital of
8807 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Figure 18. Schematic view of the most important valence orbitals of N2


and the orbital occupation in (a) the X1Σg+ ground state and (b) 1Γg
excited state. Schematic view of (c) out-of-phase (+,−) donation of the
ligand σ orbitals into the vacant 1πu′ orbital of N2 and (d) in-phase (+,
+) donation of the ligand σ orbitals into the vacant 1πg′ orbital of N2.231 Figure 19. Plot of the deformation densities Δρ which are associated
Reprinted with permission from ref 231. Copyright 2013 John Wiley with most important orbital interactions in N 2 (PPh 3 ) 2 . (a)
and Sons. Deformation density Δρ1 due to donation Ph3P→N2←PPh3 into the
vacant in-plane π orbital of N2(1Γg). (b) Deformation density Δρ2 due
the complex more strongly stabilizing than the HOMO−1. to donation Ph3P→N2←PPh3 into the vacant in-plane π* orbital of
Figure 18c shows the Lewis structure of N2(PPh3)2, which N2(1Γg). (c) Deformation density Δρ3 due to π backdonation Ph3P←
depicts the bonding situation in the molecule. N2→PPh3 from the occupied out-of-plane π orbital of N2(1Γg).231
Table 10 shows the numerical results of the EDA-NOCV Reprinted with permission from ref 231. Copyright 2013 John Wiley
calculation of N2(PPh3)2 with N2 in the (1)1Γg reference state as and Sons.

Table 10. EDA-NOCV Results of N2(PPh3)2 Using the not compensate for the geometrical and electronic relaxation of
Fragments N2 in the (1)1Γg Reference State and 2 PPh3a the fragments, which amounts to ΔEprep = 349.3 kcal/mol. The
authors suggested that N2(PPh3)2 is a kinetically unusually
Energy terms N2 + 2 PPh3 stable molecule where the dissociation reaction for loss of the
ΔEint −300.1 PPh3 ligands may be a symmetry forbidden reaction.231 The
ΔEPauli 853.4 activation barrier could not be calculated, because it requires
ΔEelstatb −377.4 (32.7%) sophisticated approaches for a large molecule that were not
ΔEorbb −776.1 (67.3%) available to them. They estimated from an approximated
ΔEσ1c,d −357.9 (46.1%) transition state that the barrier was higher than 25.1 kcal/mol
ΔEσ2c,d −289.8 (37.3%) but lower than 67.6 kcal/mol, which is the calculated bond
ΔEπ1c,d −46.9 (6.0%) dissociation energy for breaking the N−N bond.
ΔErestc −81.5 (10.6%) 7.3.5. O2, F2 and Related Molecules. The peculiarity of
a
Energy values in kcal/mol. The data are taken from ref 231. bThe O2, which possesses a triplet (3Σg−) ground state, was already
percentage values in parentheses give the contribution to the total mentioned. It arises comprehensibly from the MO diagram in
attractive interactions ΔEelstat + ΔEorb. cThe percentage values in Figure 8, which indicates that dioxygen has a double bond. The
parentheses give the contribution to the total orbital interactions EDA results for O2 in Table 4 must be taken with some caution.
ΔEorb. dThe notation σ1, σ2, π1 refers to the orbital pairs which are It is not possible to assemble O2 in the (3Σg−) triplet state from
associated with the deformation densities Δρ1, Δρ2, Δρ3 that are two oxygen atoms in the 3P state without a concomitant spin
shown in Figure 19.
change of one π electron (see Figure 9). This is how the EDA
calculation of O2 (3Σg−) was done.28 Table 4 shows that the total
acceptor fragment. The orbital term ΔEorb comprises two-thirds π bonding contribution to the orbital interactions is 28.5%, a
of the total attraction between the donor and acceptor moieties. little less than in N2, where π bonding makes up for 34.4% of
There are three major components of ΔEorb, which can easily be ΔEorb. According to the EDA results shown in Table 4, the
identified with the help of the associated deformation densities largest degree of π bonding contributions of the diatomic species
Δρ (Figure 19) as in-phase (+,+) and out-of-phase (+,−) σ are found in B2 (40.8%) and C2 (48.2%). This is, because the σ
donation Ph3P→N2←PPh3 into the unoccupied 1π′ orbitals of orbital interactions in B2 and C2 come from dative bonding while
N2 and the weaker Ph3P←N2→PPh3 π backdonation. the σ bonds in N2 and O2 are due to stronger electron-sharing
Table 10 shows that the intrinsic bond strength between N2 in bonds.
the (1)1Γg reference state and the PPh3 ligands is very large The chemical bond in F2 comprises a single electron-sharing σ
(−300.1 kcal/mol). Thus, dinitrogen in the (1)1Γg state is a very bond, which is relatively weak. Difluorine is one of the rare
strong Lewis acid. This is due to the rather high electronegativity exceptions where the bond of a first octal-row atom is weaker
of nitrogen and because the nitrogen atoms in have only than in the heavier homologues. The BDE of Cl2 is calculated at
an electron sextet. However, the large interaction energy does the same level of theory with De = 62.0 kcal/mol.28 The weak
8808 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

bond in difluorine is usually explained with the repulsion equilibrium distance and the frozen distance of N2. Table 11
between the lone-pair electrons at short distances. But the shows that the Pauli repulsion in O22+ at the shorter equilibrium
strongest repulsion comes from the Pauli exchange interactions bond length (1.051 Å) is stronger than at the frozen N−N
between the electrons in the σ orbitals preventing shorter bonds. distance (1.102 Å), but this is overcompensated by the stronger
Table 4 shows also that the Coulombic attraction in F2 is much covalent interactions. The Coulomb repulsion at the shorter
weaker than in O2 and particularly in N2. The rather long F−F equilibrium distance is weaker than at the longer distance,
distance compared to O2 and N2 prevents an effective overlap of because the increase of the electron−nucleus attraction is larger
the p(σ) electrons with the nucleus of the other fluorine. The than the electron−electron and nucleus−nucleus repulsion. All
small stabilization energy of −6.2 kcal/mol of the π orbitals in F2 this is somewhat counterintuitive for a classical viewpoint, but it
comes from the change in the mixing of the p(π) fluorine AOs of becomes understandable using quantum theoretical argu-
the TZ2P basis set; it is an intraorbital polarization. ments.28
The highest lying occupied orbitals of F2 and O2 are Table 11 shows that the Coulomb contribution in O22+ is
antibonding. Removal of two electrons in the dications should repulsive, which makes the overall interaction energy ΔEint and
enhance the covalent bonding, which might overcome the the bond dissociation energy De positive. It means that the bond
Coulomb repulsion between the positively charged ions. In fact, dissociation energy is not always an adequate indicator for the
the covalent attraction in (X1Σg+) O22+ not only overcomes the strength of a chemical bond. This holds for all energy-rich
intrinsic Coulomb repulsion, it even enforces a shorter O−O molecules, where bond-breaking reactions are exergonic. It can
bond length in the dication (1.051 Å) than in neutral be argued that the activation energy for breaking the bond, ΔE⧧,
isoelectronic N2 (1.102 Å). This paradoxical finding has been
may be used instead, which would involve the calculation of the
noted before.232 The dioxygen dication O22+ has been
activation barrier. However, bond breaking involves a change in
experimentally observed and was the subject of several
the electronic structure when one goes from the molecule to the
theoretical studies.232−236 It is a metastable species, which is
thermodynamically unstable toward dissociation in two O+ fragments, which may not be related to the intrinsic binding
cations by 84.6 kcal/mol,232 but it lies in a deep potential well interactions at equilibrium. A case example is the CC double
of about 4.7 eV.233 Why does that O22+ have a shorter bond than bond in C2F4, which is discussed below in section 8.237 A direct
N2? Table 11 shows the EDA results for O22+, which may be probe for the intrinsic strength of a chemical bond is the local
mode force constant, which was introduced by Cremer and co-
Table 11. EDA Results of Singlet O22+ and Triplet F22+ workers as a universal measure of the bond strength.200,238−241
Species at the BP86/TZ2P-ZORA//BP86/def2-TZVPP The force constant is the second derivative of the energy with
Levela respect to the equilibrium coordinates, which does not require
the energy of the fragments or the transition state as a reference
O22+d O22+e F22+ for bond strength. It may thus be used for stable and metastable
ΔEint 56.7 59.2 129.4 species alike.
ΔEPauli 708.9 579.2 210.2 Table 11 also shows the EDA results for the difluorine
ΔEelstatb 126.5 144.0 233.0 dication F22+ in the triplet (X3Σg−) ground state.242,243 The F−F
ΔEorbb −778.8 (100%) −664.0 (100%) −313.9 (100%) bond in the dication (1.266 Å) is much shorter than in neutral F2
ΔE(σ)c −484.7 (62.2%) −418.4 (63.0%) −216.5 (69.0%) (1.420 Å), but it is not shorter than in isoelectronic O2 (1.224
ΔE(π)c −294.0 (37.8%) −245.6 (37.0%) −97.4 (31.0%) Å). The data for the energy values show that the Pauli repulsion
r(E-E) 1.051 1.102 1.266 in F22+ is much weaker (ΔEPauli = 210.2 kcal/mol) than in
De 56.7 59.2 129.4 neutral O2 (ΔEPauli = 464.9 kcal/mol, Table 4) but the orbital
a
Energies are in kcal/mol and bond lengths r(E-E) in Å. bThe interactions in F22+ (ΔEorb = −313.9 kcal/mol) are also weaker
percentage values in parentheses give the contribution to the total than in O2 (ΔEorb = −447.1 kcal/mol). It appears that the
attractive interactions ΔEelstat + ΔEorb. cThe percentage values in interactions in the isoelectronic pair F22+/O2 are somewhat
parentheses give the contribution to the total orbital interactions
different from the system O22+/N2, which may be due to the
ΔEorb. dCalculated at the equilibrium distance. eCalculated using the
frozen equilibrium distance of isoelectronic N2. occurrence of Pauli repulsion between π electrons in the latter
pair. But the conclusion remains that the removal of electrons
from antibonding orbitals is able to strongly enhance the
compared with the data for N2 (Table 4). The energy values covalent interactions, which may lead to significantly shorter
indicate that the Pauli repulsion in the dication (ΔEPauli = 708.9 bonds overcoming strong Coulomb repulsion.
kcal/mol) is weaker than in N2 (ΔEPauli = 802.4 kcal/mol) The following conclusions arise from this section:
whereas the stabilizing orbital interactions in O22+ (ΔEorb =
−778.8 kcal/mol) are stronger than in N2 (ΔEorb = −729.8 kcal/ • The trend of the bond strength of the first octal-row
mol). The orbital contraction in the dication due to the positive diatomic molecules can conclusively be explained with the
charge has obviously a much stronger effect on the Pauli results of the EDA-NOCV calculations, which provide
repulsion than on the orbital interactions. Although the further insight into the nature of the interactions in terms
interatomic distance in O22+ is smaller than in N2, the Pauli of Coulombic attraction/repulsion, Pauli repulsion, and
repulsion becomes weaker, because the overlap of the occupied orbital (covalent) bonding. Accordingly, the very weak
orbitals in the dication becomes smaller due to the charge attraction in Be2 is due to Pauli repulsion but not to
induced contraction. The overlap between occupied and vacant mutual energy compensation of the occupied bonding
orbitals in O22+ is reduced to a lesser extent, and therefore, the and antibonding valence orbitals. Likewise, the rather
orbital interactions in the dication become stronger than in N2. weak bond in F2 is caused by the Pauli repulsion of the
The effect of the different energy terms on the bond length of electrons in the σ orbitals rather than repulsion of the π
O22+ becomes obvious by comparing the calculated values at lone-pair electrons.

8809 DOI: 10.1021/acs.chemrev.8b00722


Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

• The covalent bonding in the dications (X1Σg+) O22+ and molecules. There is one particular feature of the atoms, which
(X3Σg−) F22+ is stronger and the bonds are shorter than in can be considered as essential for the different chemistry of the
the neutral parent molecules because the Pauli repulsion two groups. It concerns the ratio of the radii of the (n)s and (n)p
is weaker in the charged species due to the orbital valence AOs of the atoms, which significantly changes from the
contraction. The covalent bonding in O22+ is even first to the higher octal-row atoms. The 2p orbitals of the first
stronger than in N2. The Coulomb repulsion makes the octal-row atoms may penetrate rather deeply into the core,
dications thermodynamically unstable. The force con- because there are no lower lying occupied p functions, whereas
stant is more suitable as measure for the intrinsic strength the 2s orbitals are constrained to be orthogonal to the 1s AO.
of a bond than the bond dissociation energy. The radii of 2s and 2p orbitals are thus very similar, which leads
• The consideration of the occupied and vacant orbitals in to effective sp hybridization in chemical bonds of the lighter
the ground and excited states of E2 provides an atoms. In contrast, the occupied (n)p AOs (n > 2) of the heavier
understanding of complexes where the diatomic species atoms are antisymmetric to the p core orbitals, which causes
are stabilized by donor ligands L→E2←L. This led to the larger radii for the (n)p than for the respective ns AOs. The
synthesis of the first compound with a boron−boron spatial regions of (n)s and (n)p AOs (n > 2) are more separate
triple bond NHCR→B2←NHCR that is stable under than those of 2s and 2p AOs, which leads to less effective
ambient temperatures. hybridization of the heavier main-group atoms.246 Looking from
• The donor strength in complexes L→E2←L may an aufbau principle of atomic structure, chemical bonding of the
overcompensate the electronic excitation energy of E2 heavier main-group atoms exhibits “normal” behavior, whereas the
from the ground to the excited reference state in the f irst octal-row atoms are “anormal”, because the valence p AOs are
adduct, which leads to thermodynamically stable not affected by lower lying p orbitals. The difference between the
compounds. However, there are molecules like L→ relative size of the light and heavier main group atoms is
N2←L (L = NHC, PPh3) where N2 is in the high lying schematically shown in Figure 20.
(1)1Γg state, which are calculated to be unstable toward
dissociation of N2, and yet, they can easily be isolated.
This is likely because the extrusion reaction is a symmetry
forbidden process.
7.4. Chemical Bonding in Heavier Main-Group Compounds
There is ubiquituous chemical evidence that the structures and
reactivities of the first octal-row atoms exhibit siginificant
differences from the heavier homologues. This is particularly
evident when one compares the organic chemistry of carbon
with silicon chemistry. The statement “Silicon is weird” clearly
characterizes this finding from an organic chemist point of
Figure 20. Schematic representation of the relative size of the (n)s and
view.244 The difficulties to understand the large differences
(n)p AOs in atoms (a) of the first octal row and (b) in heavier main-
between molecules of the first octal row and higher homologues group atoms.
stem from the fact that the heuristic bonding models, which were
developed during the last century, received their input mainly
from observations which were made with compounds of the first
octal-row atoms, henceforth also called lighter elements, for The relative size of the (n)s and (n)p AOs of the main-group
which experimental data were available. The neural network of elements is a significant factor for the nature of the chemical
chemical information, which was used to develop such models, bonds and the reactivities of main-group compounds. An
was mainly fed with results that came from first octal-row approaching atom B “sees” the (n)p AOs earlier than the (n)s
molecules. Compounds of the heavier elements, which exhibit AOs (n > 2) of A. In less colloquial wording, it means that the
different properties, were initially considered as exceptions, for interference of the wave function between B and the valence p
which sometimes peculiar explanations were made. This led to AO of heavy atom A occurs earlier and is larger than with the
difficulties when the information about molecules of the heavier valence s AO. Kutzelnigg analyzed the Mulliken gross
elements increased. The unequal structures and reactivities of population at the central atoms in EH4 (E = C, Si) and found
heavier main-group compounds could no longer be considered that the hybridization at carbon is sp3 while it is ∼sp2 in SiH4,
as anomalies. But the bonding models that were suggested to although both molecules have Td symmetry.245 Table 12 shows
rationalize their behavior were still mainly based on extensions the hybridization of the E-X localized orbitals in tetrahedral (Td)
of the existing schemes with some ad-hoc additions. Some of EX4 (E = C−Pb; X = H, Cl) given by the NLMO (Natural
them were simple prescriptions for writing formulas that appear Localized MO) method.68−71 Only CH4 shows a perfect sp3
to address the peculiar bonding situations of the heavier hybridization, while the E−H bonds in the heavier EH4
molecules without which a sound theoretical basis for the homologues possess ∼sp2 hybridized orbitals at atom E. The
differences between lighter and heavier main-group compounds chlorine systems ECl4 deviate even more strongly from sp3
was given. hybridization. The percentage p-character of the E-Cl bonds at
The nature of the chemical bond in heavier main-group atom E is much higher than one would expect from the sp3
compounds has been discussed from a quantum chemical point model. The Pb−Cl bonds have even nearly equal s and p orbital
of view in a review article by Kutzelnigg in 1984.245 The author contribution at Pb. Note that the correct hybridization is only
presents the fundamental differences between the valence shells given by the NLMOs, which include the nondiagonal elements
of the lighter and the heavier main group atoms, and he discusses of the Fock matrix. The standard NBOs give perfect sp3
their effects on the structures and bonding situations of the hybridization for all EX4 species.
8810 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Table 12. NLMO Analysis of EX4 (E = C−Pb; X = H, Cl) at than for the respective (n)s AOs of the heavier main-
the HF/def2-TZVPP Level Using NBO 6.0a group atoms.
q Pol. Hybrid.
• Effective spx hybridization is only possible for the first
Molecule E X P(E−X) E% X% ns:np:(n−1)d
octal-row atoms, because the 2s and 2p AOs have similar
CH4 −0.74 0.19 0.97 59 41 25:75:0 radii. Chemical bonds of the heavier atoms have a rising
SiH4 0.85 −0.21 0.93 39 61 33:66:1 degree of higher %p character, and the (n)s orbitals
GeH4 0.74 −0.18 0.94 41 59 33:67:0 become gradually less prone to be engaged in chemical
SnH4 0.96 −0.24 0.91 38 62 35:65:0 bonding. This stabilizes low-valent compounds of the
PbH4 0.74 −0.18 0.92 41 59 36:64:0 heavier main-group atoms.
CCl4 −0.17 0.04 1.02 49 51 30:70:0
SiCl4 1.51 −0.38 0.88 26 74 37:60:3 7.4.1. Hypervalent Molecules: Valence, Orbital Sym-
GeCl4 1.64 −0.41 0.91 25 75 43:56:1 metry, and 3-Center, 4-Electron Bonding. In order to
SnCl4 1.87 −0.47 0.77 23 77 42:57:1 address the topic of hypervalence, one must first clarify what is
PbCl4 1.74 −0.44 0.76 25 75 49:51:0 meant by the term “valence”. According to the IUPAC, valence
a
Atomic partial charges q, Wiberg bond order P(E−X), polarization, is defined as “The maximum number of univalent atoms (originally
and hybridization of atom E in the E−X bond. hydrogen or chlorine atoms) that may combine with an atom of the
element under consideration, or with a f ragment, or for which an
atom of this element can be substituted.”251 Accordingly, molecules
The differences in the structures of molecules between first such as PF5 and SF6 are hypervalent, because they doubtlessly
and higher octal-row elements can thus be traced back to the possess five and six bonds at phosphorus/sulfur. However,
radii of the valence s and p AOs. This shall be demonstrated with quantum chemistry suggests that chemical bonding and thus
selected examples below. We also want to mention relevant valence are associated with the molecular orbitals, which are
theoretical studies by Kaupp, who analyzed the bonding used to describe the chemical bonds. Then the question arises,
situation in heavier main-group compounds paricularly of the which atomic orbitals are effectively engaged in the interference
sixth row of the periodic system.247−250 of the orbitals building the wave function. Since the s and p AOs
There are two prominent observations that distinguish can be occupied by a maximum of eight electrons, it seems
heavier main-group compounds from compounds of the lighter difficult to assign more than four bonds to an atom, which uses
homologues which shall be analyzed and discussed in this only its valence (n)s and (n)p AOs for covalent bonding. A
section. One phenomenon concerns the number of electron-pair simple explanation for the observation of stable molecules with
bonds and the coordination number of the atoms. Heavier main- more than four bonds such as PF5 and SF6 considers the (n)d
group atoms can have more than four bonds in stable molecules. orbitals, which are available for heavier main-group atoms where
This is essentially unknown for first octal-row atoms. Classical n ≥ 3. Early work suggested that hypervalent molecules possess
examples are SF6 and PF5, which are discussed below. The (n)sx(n)py(n)dz hybridized bonds. Accordingly, hypervalent
question if such molecules should be considered as hypervalent molecules would not obey the octet rule, which states that a
and if they obey the octet rule will also be considered. maximum of eight electrons are used for covalent bonding.
The second topic concerns the observation that compounds An alternative suggestion, which uses only (n)s and (n)p AOs
with multiple bonds between heavier main-group atoms are for covalent bonding in hypervalent molecules, was suggested
much more difficult to isolate and that they often possess very independently by Rundle252−254 and Pimentel.255 It is now
different equilibrium geometries than the first octal-row known as the 3-center, 4-electron bonding model, and it was
analogues. Classical examples are ethylene and acetylene and adapted by Coulson to VB theory.256 It is worthwhile to read the
their heavier group-14 homologues. The standard textbook original paper, because it already uses the symmetry of the
explanation suggests that the overlap of the p(π) AOs of the molecular orbitals for explaining chemical bonding in molecules
heavier atoms is much smaller than for lighter atoms, because of that possess an excess number of electron pairs than required by
the longer bonds of the former. It has been known for a long time the octet rule. The authors discuss the bonding in various
that this suggestion is not valid, because the p(π) AOs of the molecules such as I5−, XeF4, XeF2, HF2−, and X3−.252−255 It is
heavier atoms are more diffuse than for lighter atoms, which shown that the covalent bonds in the molecules may be
largely compensates for the longer bond. Furthermore, the described without invoking the use of higher energy basis
alleged weakening of the π bonds as reason for the experimental functions, as it was previously advocated by Pauling11 and by
findings is questionable, because the overlap of the σ AOs of the Kimball.257 The 3-center, 4-electron model can also be applied
heavier atoms is also affected by the longer distance. We shall to the π bonds in the allyl anion C3H5−. The basic feature is
present an explanation for the observations, which is based on displayed in Figure 21, which uses a model system of three atoms
the valence shell structure of the atoms. A3 possessing one AO each.
The following conclusions arise from this section: The three atomic basis functions lead to three MOs φ1−3,
• The most important difference between atoms of the first which are shown in ascending energy order. Covalent bonding
and higher octal rows of the periodic system, which comes from the occupation of φ1, which is totally symmetric.
determines the experimentally observed variance of their There is no covalent bonding between the central and the
structures and reactivities, is the relative size of the valence terminal atoms in φ2, which has one node. But there is still a
s and p orbitals and not the availability of d orbitals for the stabilizing contribution from electrons that occupy φ2, because
heavier elements. The (n)p AOs for n > 2 are, as of the Coulomb attraction between the nucleus of the central
determined by the Pauli principle, antisymmetric to the p atom and the electrons occupying the MO. The energy lowering
core orbitals, unlike the 2p AOs, which experience no due to occupation of φ2 is less than that of the electrons in φ1,
lower lying p orbitals. This causes larger radii for the (n)p but the total system is further stabilized. The occupation of φ2 in
8811 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

are stable only for the heavier atoms lies in the atomic size. First
octal row atoms A are too small and cannot accommodate more
than four atoms in AXn because the X−X Pauli repulsion
prevents the formation of stable species with n > 4. The 3-center
4-electron bonding model is an early version of the MO derived
models, which uses the symmetry of the orbitals for the
explanation of structures and reactivity that became so
prominent in the 1970s.
7.4.2. Hypervalent Molecules: SF6 and PF5. The
molecules SF6 and PF5 may be considered as archetypical
examples for the validity problem of the octet rule and the
chemical bonds in hypervalent compounds. There are definitely
six and five bonds in the respective molecules, which pose the
question about the correct representation of the bonding
Figure 21. Schematic representation of the molecular orbitals φ1 − φ3 situation with Lewis structures in accordance with the octet rule.
of a 3-center, 4-electron bond. There have been numerous discussions and suggestions about
the usefulness of Lewis structures and the octet rule for
A3 plays a similar role as the occupation of the 1eg MO in SF6, hypervalent molecules, which are perhaps better termed as
which is discussed below (Figure 22). hypercoordinated species. The problem can straightforwardly
The important result of these early works252−255 is the finding be addressed with the help of the symmetry of the molecular
that molecules with covalent bonds, which possess an excess orbitals describing the valence electrons in the molecule.
number of electron pairs as required by the octet rule, can be Figure 22a schematically shows the σ MOs of model
described using only their (n)s and (n)p orbitals for the compound EX6 where the central atom E has (n)s and (n)p
interference of the wave function. When accurate quantum valence functions. In the octahedral (Oh) ligand field in EX6, the
chemical calculations became available later, it was shown that (n)s and (n)p atomic orbitals of atom E split into an a1g and a
the (n)d AOs of heavy main-group atoms are not genuine triply degenerate set of t1u AOs, whereas the six σ orbitals of the
valence orbitals in hypervalent molecules but rather serve as ligand cage X6 are divided into a1g, t1u, and eg MOs. The
polarization functions. The reason why hypervalent molecules molecule SF6 has 12 valence electrons that occupy six σ MOs.

Figure 22. Splitting of the (n)s and (n)p AOs of a main-group atom E (a) in the octahedral field of six ligands X in EX6 and (b) in the trigonal pyramidal
field of five ligands X in EX5.

8812 DOI: 10.1021/acs.chemrev.8b00722


Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Figure 22a shows that only four MOs in EX6 (a1g and t1u) have
the right symmetry to enable interference between the AOs of
atom E and the ligands cages. The degenerate eg MO does not
induce covalent E−X bonding. This does not mean that the
electrons in the eg MO do not contribute to the stabilization of
the molecule. The electronic charge of the ligand cage X6 may be
strongly stabilized through Coulomb interactions with the
nucleus of atom E.258 This also shows the limitation of the
model of orbital interactions (covalent bonding) for describing
chemical bonding. There are other forces such as Pauli repulsion
and Coulombic interactions, which strongly influence the
chemical bond. It is a widely held belief that electrostatic
interactions between neutral atoms are weak or may even be
repulsive, which for most molecules is not correct.28,177,259 Thus, Figure 23. Lewis structures for (a) SF6 and (b) PF5 that obey the octet
there are actually only four electron-pair bonds in octahedral rule.
EX6 that come from the interference of the wave functions of the
atoms, which fully agrees with the octet rule. This information is tion in a molecule should not be used as evidence for the
only available by inspection of the symmetry of the MO wave correctness of the model.
function. But the orbitals are delocalized over more than two The following conclusions arise from this section:
atoms (seven atoms in the 1a1g MO and three atoms in each 1t1u
components) and the problem is to convert the delocalized • The question of hypervalence can only be addressed when
picture of the MO calculation into the model of localized Lewis valence is defined.
structures. Much of the difficulties and controversies about • Covalent bonding of main-group compounds usually
chemical bonding in molecules in the literature can be traced involves the atomic valence s and p orbitals. Molecular
back to the discrepancy between the viewpoints of delocalized orbitals can be constructed so that they are filled with
and localized bonds. The adherence to two-center bonds in the more than eight electrons, but the individual atoms do not
works of Gilbert Lewis and Linus Pauling can be identified as a receive more than eight electrons in their valence shell.
severe limitation of the viewpoint on chemical bonding, and it is • The Lewis model of localized electron pairs is not well
the source of much confusion still today. This is surprising, suited for describing hypervalent molecules.
because the ground-breaking study by Erich Hückel on • The 3-center, 4-electron MO model of Rundle and
aromaticity showed in the 1930s already the advantages if the Pimentel provides a simple rationale for the bonding in
description of chemical bonding is not restricted to a localized hypervalent molecules. It shows that the octet rule is
model.20 fulfilled in molecules where a main-group atom has more
The connection of the octet rule to the bonding situation in than four bonds. The stabilization of the excessive
PF5 is similar to SF6 but with an additional facet. Figure 22b electrons comes from the electrostatic stabilization,
which is an important factor in all covalent bonds.
shows that the AOs of E and σ MOs of X5 split in the D3h
• The bonding in SF6 and PF5 is an archetypical example for
symmetric model compound EX5 and give rise to σ orbitals
formally hypervalent molecules that still obey the octet
possessing a1′, a2″, and e′ symmetry, which are occupied by ten
rule. There are only four doubly occupied valence MOs in
electrons. The 1a1′, 1a2″, and the degenerate 1e′ MOs are
SF6, which have contributions from the s and p AOs of
bonding orbitals that give four delocalized electron-pair bonds,
sulfur. The symmetry of the MOs is the key for
which result from the interference of the wave functions. The understanding the interplay between the octet rule and
2a1′ orbital is the HOMO in EX5, but the coefficient of the (n)s the number of bonds. The excess electrons in PF5 occupy
valence AO is very small, which makes it effectively a an antibonding MO, which has a negligible contribution
nonbonding MO. There is negligible interference between the from the s and p AOs of phosphorus.
atomic orbitals of phosphorus and fluorine in the 2a1′ MO. 7.4.3. Multiple Bonds of Heavy Diatomic Molecules
Covalent bonding refers to the number of bonds, which are due Na2−Cl2 and N2−Bi2 and Related Molecules. A commonly
to the interference of the wave functions. There are four such made statement which is found in many textbooks of chemistry
doubly occupied MOs in the molecules. The bonding situation claims that the π bonds between heavy main-group atoms are
in PF5 is thus in agreement with the octet rule.63,260 weaker than between first octal-row elements. Table 13 gives the
How may the information gained by the MO model be EDA results for the diatomic molecues Na2−Cl2 (second-row
expressed by Lewis structures? The most common sketches of sweep), which may be compared with the data for the first-row
Lewis formulas for SF6 and PF5 obeying the octet rule depict one species Li2−F2 (Table 4).28 The BDEs of the heavier molecules
of 14 “resonance forms” for SF6 and one of five for PF5 as shown are smaller than those of the lighter systems, except for the
in Figure 23, which require formally charged fragments. It dihalogens F2 and Cl2, which agrees with the experimental
becomes obvious that the presentation of the chemical bonds is findings. The reason for the exceptionally weak F−F bond has
clumsy and the use of charged species may give rise to a been discussed in section 7.3.5. Note that Si2 has a X3Σg− ground
misleading interpretation of the nature of the bonding. The state whereas the valence isoelectronic C2 has a X1Σg+ ground
“ionic−covalent” resonance forms in Figure 23 must not be state. A comparison of the nature of the bonding between the
interpreted as indication for the appearance of ionic interactions. different atoms should be done for the same spin state.
The charges that show up in the Lewis structures are only formal, The trend of weaker bonds for heavier atoms, which is a
and they have no physical relevance. The sometimes accidental general feature for the elements of the periodic system, comes
agreement between formal charges and actual charge distribu- from the more weakly bonded valence electrons. The first
8813 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Table 13. Energy Partitioning Analysis of the Second Row Dimers E2 (E = Na−Cl) in C2v at BP86/TZ2Pa
E Na Mg Al Si Si P S Cl
State Σg+
1 1
Σg+ 3
Σg− 3
Σg− 1
Σg+ 1
Σg+ 3
Σg− 1
Σg+
ΔEint −16.3 −1.7 −35.6 −78.3 −56.5 −118.5 −112.2 −64.6
ΔEPauli 5.1 9.4 42.9 184.2 79.5 317.4 247.3 129.1
ΔEelstatb −10.7 (50.1%) −5.2 (46.2%) −15.6 (19.9%) −123.7 (47.1%) −4.4 (3.2%) −186.3 (42.7%) −121.8 (33.9%) −51.1 (26.4%)
ΔEorbb −10.7 (49.9%) −6.0 (53.8%) −62.9 (80.1%) −138.9 (52.9%) −131.6 (96.8%) −249.7 (57.3%) −237.7 (66.1%) −142.7 (73.6%)
ΔΕ(σ)c −10.7 (100.0%) −6.0 (100.0%) −26.1 (41.5%) −91.7 (66.0%) −50.1 (38.1%) −148.6 (59.5%) −154.0 (64.8%) −126.7 (88.8%)
ΔΕ(π)c 0.0 (<0.1%) 0.0 (<0.1%) −36.8 (58.4%) −47.2 (34.0%) −81.4 (62.0%) −101.0 (40.4%) −83.7 (35.2%) −15.9 (10.2%)
ΔEcorre 0.3 0.0 1.1 1.7 1.6 1.3 3.0 2.6
De d 16.0 (17.1) 1.7 (1.2) 34.5 (35.1) 76.6 (74.8) 54.8 117.2 (117.2) 109.2 (101.8) 62.0 (58.0)
E−Ed 3.095 (3.079) 3.607 (3.890) 2.510 (2.466) 2.303 (2.246) 2.071 1.911 (1.893) 1.921 (1.889) 2.023 (1.987)
a
All energies in kcal/mol, distances E−E in Å, all dissociation energies were calculated for the dissociation in the ground state atoms (Na: 2S; Mg:
1
S; Al: 2P; Si: 3P; P: 4S; S: 3P ; Cl: 2P). Data are taken from ref 28. bValues in parentheses give the percentage contribution to the total attractive
interactions ΔEelstat + ΔEorb. cValues in parentheses give the percentage contribution to the total orbital interactions ΔEorb. dExperimental values in
parentheses from ref 24. eCorrection for the spin polarization.

Table 14. Energy Partitioning Analysis of the E−E Bond for E = N−Bi (BP86/TZ2P)a
N2 P2 As2 Sb2 Bi2
ΔEint −240.2 −118.5 −97.3 −74.2 −64.6
ΔEPauli 802.2 317.6 276.8 219.3 201.9
ΔEElstatb −312.8 (30.0%) −186.3 (42.7%) −181.5 (48.5%) −159.0 (54.2%) −150.8 (56.6%)
ΔEOrbb −729.6 (70.0%) −249.8 (57.3%) −192.7 (51.5%) −134.5 (45.8%) −115.7 (43.4%)
ΔEσc −478.7 (65.6%) −148.7 (59.5%) −116.2 (60.3%) −81.0 (60.3%) −72.4 (62.6%)
ΔEπc −250.9 (34.4%) −101.1 (40.5%) −76.4 (39.7%) −53.4 (39.7%) −43.2 (37.4%)
Overlap σe 1.59 (68.2%) 1.49 (70.3%) 1.40 (70.7%) 1.36 (72.3%) 1.21 (70.8%)
Overlap πe 0.74 (31.8%) 0.63 (29.7%) 0.58 (29.3%) 0.52 (27.7%) 0.50 (29.2%)
Hybridization of the σ bond s: 37.5% s: 20.8% s: 15.4% s: 11.6% s: 8.0%
p: 61.9% p: 77.8% p: 83.9% p: 87.9% p: 91.9%
R(E−E)d 1.102 (1.098) 1.911 (1.893) 2.113 (2.103) 2.504 (2.48) 2.660 (2.660)
De 240.2 118.5 97.3 74.2 64.6
D0d 236.9 (225.0) 117.4 (116.1) 96.7 (91.3) 73.8 (71.3) 64.3 (47.0)
a
The interacting fragments are two atoms E in the 4S ground state (s2px1py1pz1). Energy values are given in kcal/mol. Bond lengths are given in Å.
The values are taken from ref 261. bThe values in parentheses give the percentage contribution to the total attractive interactions ΔEelstat + ΔEorb.
c
The values in parentheses give the percentage contribution to the total orbital interactions ΔEorb. dExperimental results are given in parentheses.
e
The values in parentheses give the percentage contribution to the total orbital overlap.

ionization energies decrease from top to bottom of the periodic slightly less, but the percentage value of π bonding in Bi2 is still
system within each group. Exceptions arise when inner shells are higher (37.4%) than in N2 (34.4%). The overlap of the p(π)
filled or when relativistic effects become important, which are orbitals decreases from nitrogen to the heavier atoms, but the
discussed further below. Table 13 shows that the percentage same trend is observed for the overlap of the σ orbitals. Note that
contribution of the electrostatic attraction ΔEelstat to the bonds the hybridization of the σ bonds changes significantly toward
in the second-row dimers is higher than in the first-row dimers. higher %p character from N2 to the heavier atoms. This comes
This can be explained with the larger nuclei and the bigger from the change in the radii ratio of the valence s and p AOs
electronic charges of the former, which compensate for the (Figure 20).
longer bonds. The most interesting result for the present topic Numerous complexes have been synthesized in the recent
arises from comparison of the π contribution to the orbital past where heavier diatomic species E2 of the third and higher
interactions ΔEorb between Al2−S2 and B2−O2. The data in octal row of the periodic system are stabilized by donor ligands
Table 13 suggest that the π bonding in the heavier diatomic L→E2←L where L is mostly a NHC or CAAC carbene ligand.262
molecules contributes more to the covalent bonding than in the The bonding situation in these adducts has been explained in
lighter species. terms of donor−acceptor interactions between E2 in an
The second-row sweep Na2−Cl2 shown in Table 13 shall be electronically excited reference state and the carbene σ-donor
complemented by a group-15 sweep of the diatomic species N2− ligands in a similar fashion as for the first octal-row adducts,
Bi2. The numerical results of the bonding analysis are shown in which were discussed above.263,264 Since there are no new
Table 14.261 The bond strength of the molecules given by the aspects about the nature of the bonding in the heavier
ΔEint (= −De) values exhibits the expected weakening from top complexes, we do not present them here. The interested reader
to bottom N2 ≫ P2 > As2 > Sb2 > Bi2. The relative Coulombic can find the details in the original publications and in recent
attraction to the overall binding shows the opposite trend, which review articles.49,55,56
is due to the increase of the positive charge of the nuclei. The The increase of the π contribution to the covalent interactions
contribution of π bonding to the overall orbital interaction in the heavier diatomic molecules is in contrast to the textbook
increases from nitrogen to phosphorus and then becomes knowledge that π bonding between heavier main-group atoms is
8814 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

particularly weak and, therefore, heavier main-group com- only stretched by 16.2% relative to P2 (1.911 Å). This is a hint
pounds with multiple bonds are much less stable than toward a possible explanation. More detailed information is
homologues of the lighter elements. Since chemical evidence available from EDA calculations of the diatomic and tetraatomic
for the latter finding is ubiquitous, other factors rather than molecules.
weaker π bonding must be responsible for the observation and Table 15 shows the EDA results for E4 (E = N, P). Note that
different explanations have to be provided. This is done in the the N−N and P−P bonds have a similarly small %s contribution;
following sections.
7.4.4. Multiple Bonds of Heavy Main-Group Atoms: Table 15. Energy Partitioning Analysis of E4 (E = N, P)
N2/N4 vs P2/P4. The most common allotropes of nitrogen and (BP86/TZ2P) and Comparison with E2a
phosphorus are gaseous N2 and tetrahedral P4 (white
E4 1/4 E4 1/2 E2
phosphorus), whose structures and relative energies may be
used as prototype to analyze the different chemical bonds of E=N
atoms of the first and second row of the periodic system. ΔEint −322.6 −80.7 −120.1
Tetrahedral N4 (Td) is an energetically high-lying minimum ΔEPauli 1636.6 409.2 401.1
species, which is experimentally unknown so far but has been the ΔEelstatb −705.4 (36.0%) −176.4 −156.4 (30.0%)
(36.0%)
subject of quantum chemical studies.265 Experimental data for
ΔEorbb −1253.7 −313.4 −364.8 (70.0%)
N2, P2, and P4 are available from the literature.24,266 (64.0%) (64.0%)
Figure 24 shows the calculated geometries of the diatomic and ΔEσc −1253.7 −313.4 −239.4 (65.6%)
tetraatomic molecules.267,268 The computed reaction energies (100.0%) (100.0%)
ΔEσ/nf −156.7 −104.5 −239.4
ΔEπc −125.5 (34.4%)
Hybridizatione of the s: 7.0% s: 7.0% s: 37.5%
σ bond p: 92.5% p: 92.5% p: 61.9%
R(N−N) 1.464 1.464 1.102 (1.098)g
E=P
ΔEint −293.1 −73.3 −59.3
ΔEPauli 972.3 243.1 158.8
ΔEelstatb −545.7 (43.1%) −136.4 −93.2 (42.7%)
(43.1%)
ΔEorbb −719.8 (56.9%) −180.0 −124.9 (57.3%)
(56.9%)
ΔEσc −719.8 −180.0 −74.4 (59.5%)
(100.0%) (100.0%)
ΔEσ/nf −120.0 −60.0 −74.4
ΔEπc −50.6 (40.5%)
Hybridizatione of the s: 5.2% s: 5.2% s: 20.8%
σ bond p: 93.2% p: 93.2% p: 77.8%
R(P−P) 2.221 (2.223)d 2.221 1.911 (1.893)g
a
The interacting fragments are four atoms E in the 4S ground state
(s2px1py1pz1). Bond lengths are given in Å. Energy values are given in
kcal/mol. The data are taken from refs 261. bThe values in
parentheses give the percentage contribution to the total attractive
interactions ΔEelstat + ΔEorb. cThe values in parentheses give the
Figure 24. Calculated bond lengths [Å] of E2 and (Td) E4 (E = N, P) percentage contribution to the total orbital interactions ΔEorb.
d
and dimerization energies De of E2 at BP86/TZ2P. Experimental bond Experimental value, ref 266. eThere is a small contribution from
lengths are given in parentheses. Reprinted with permission from ref the d polarization functions. fn = Number of bonds. gExperimental
267. Copyright 2014 Springer Nature. value, ref 24.

for formation of the tetrahedral species are striking evidence for


the drastic variation of nitrogen and phosphorus bonding: the change in the hybridization compared with the diatomic
molecules (Table 14) is particularly large for N2. Table 15 gives
2N2 → N4(Td) +158.9 kcal/mol (23) the values for E4 divided by four and the data for E2 divided by
2P2 → P4 (Td) −59.7 kcal/mol two for the direct comparison of the isomers. The values in the
(24)
two final columns refer to the energy values per atom. It becomes
The enormous energy difference of 218.6 kcal/mol between obvious that the nitrogen atoms in N2 are stabilized with respect
reactions 23 and 24 suggests a fundamental difference between to N4 whereas the phosphorus atoms in P4 are lower in energy
the strengths of the nitrogen and phosphorus bonds in the than in P2. Inspection of the three energy components indicates
diatomic and tetratomic isomers. The above results show that that the reverse in stability is also found for the orbital term
different π bond strength in the diatomic species cannot be the ΔEorb but not for ΔEelstat and ΔEPauli. The covalent interactions
reason for the diverse stability order. Inspection of Figure 24 stabilize the atoms in N2 much stronger than in N4, whereas the
shows that tetrahedral isomers of both elements have longer orbital interactions stabilize phosphorus in P4 more than in P2.
bonds than the diatomic species, but the bond lengthening of The diatomic molecules E2 have one σ and two degenerate π
nitrogen is more pronounced than that of phosphorus. The bonds, whereas E4 has six σ bonds. Thus, there are three bonds
calculated N−N bond length in N4 (1.464 Å) is 32.8% longer per atom in E2 (one σ, two π) and in E4 (three σ). Table 15
than in N2 (1.102 Å), whereas the P−P bond in P4 (2.221 Å) is shows that the σ bond strength per atom in E2 is higher than the
8815 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Figure 25. Variation of the energy terms of the EDA at different bond lengths of (a) N2 and (b) P2 at BP86/TZ2P. Reprinted with permission from ref
267. Copyright 2014 Springer Nature.

π bond strength for both N and P. The crucial difference But what is the driving force for the drastically weaker N−N σ
between nitrogen and phosphorus is the change of the σ bond bonds in the tetramer than in the dimer? The bonds in E4 are
strength from E2 to E4. The σ bond strength per atom in N4 longer than in E2. As illustrated in section 7.2, the equilibrium
(−104.5 kcal/mol) is drastically less than in N2 (−239.4 kcal/ bond length in N2 is determined by the Pauli repulsion and not
mol) and becomes even lower than the strength of the π bond in by the maximum overlap of the orbitals. Each nitrogen atom in
N2 (−125.5 kcal/mol). The σ bond strength in P4 (−60.0 kcal/ N4 encounters Pauli repulsion from three other atoms, but only
mol) is also lower than in diatomic P2 (−74.4 kcal/mol), but the from one atom in N2. The same holds for the phosphorus
weakening is clearly less than for nitrogen and it remains systems, but the Pauli repulsion in P2 has a different gradient
stronger than the π bond strength in P2 (−50.6 kcal/mol). The than in N2 with respect to the interatomic distance. Pauli
dramatic weakening of the chemical bonds in N4, which leads to the repulsion appears between electrons of the same spin, and the
overall much lower stability of the nitrogen tetramer and the reverse similar radii of the 2s and 2p AOs of nitrogen induce strong Pauli
energy shif t of the phosphorus homologues, is related to the σ bonds repulsion at the equilibrium distance of N2, which sharply
and not to π bonding. increases at shorter distances. The Pauli repulsion in P2 at
8816 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Figure 26. (a) and (b) Schematic representation of electronic 3B1 and 1A1 states of EX2. Schematic representation of the orbital interactions and
bonding models in various isomers of E2X4. (c) Electron-sharing σ and π bonding in planar E2X4 (model A). (d) Lone-pair donation in lp-bridged
structures E2X4 (model B). (e) Lone-pair donation in lp-bridged structures E2X4 (model C). (f) Bond pair donation in E-X-bridged structures E2X4
(model D).

equilibrium is weaker and increases less steeply, because the possess strongly bent nonlinear geometries.270 The latter aspect
singly occupied 3p AOs have a larger radius than the doubly is discussed in the following section. Here we investigate the
occupied 3s AO. This is shown in Figure 25, which displays the bonding situation in the parent systems H2EEH2 (E = Si − Pb)
energy terms of the EDA calculations of N2 and P2. The and in the fluorine compounds F2EEF2. Deviations of the
interaction of one nitrogen atom with three N in N4 induces geometries of compounds R2EER2, where R is a bulky group, are
stronger Pauli repulsion than the interaction of one P with the caused by steric effects. We focus on the underlying electronic
other three phosphorus atoms in P4. This is the reason why the differences between the carbon−carbon bond in ethylene and
N−N bonds in tetrahedral N4 are stretched by 32.8% relative to the E−E bonds in the heavier homologues.
N2 whereas the P−P bonds in the tetramer are stretched by only The key for understanding the different E-E bonding and the
16.2% compared to P2. The driving force for the particularly long resulting equilibrium structures of X2EEX2 between carbon and
and weak N−N bonds in N4 is thus the Pauli repulsion, which is the heavier group-14 atoms was provided 30 years ago by
due to the similar s/p ratio of the valence AOs. Trinquier and Malrieu298,299 and by Carter and Goddard,300
The following conclusions arise from this section: who suggested that the electronic states of the EX2 fragments are
• The frequently made statement that π bonds between crucial for the nature of the E−E bond.301 Figure 26a shows
heavier main-group atoms are particularly weak is not schematically the (3B1) triplet state and Figure 26b the (1A1)
justified. The relative contribution of π orbital interaction singlet state of EX2 and their relative energies. Methylene CH2
to covalent bonding in heavier main-group compounds is has a triplet ground state, which is perfectly suited to build an
even larger than in molecules of the first octal row. electron-sharing double bond X2EEX2 (model A) consisting
• The difficulty to isolate molecules of the heavier main- of a σ and a π bond (Figure 26c). CF2 and the heavier EH2 and
group atoms with multiple bonds is not due to weak π EF2 species have a singlet (1A1) ground state, and they must be
bonding. It is rather caused by the change of both the σ formally exited to the triplet (3B1) state in order to engage in
and π bond strength in addition reactions where π bonds electron-sharing bonding. Some ylidenes EH2 and EF2 with E =
are converted to σ bonds. P4 (Td) is significantly more Si−Pb bind through their singlet (1A1) ground state via donor−
stable than 2 P2, but N4 (Td) is much higher in energy than acceptor interactions (model B) where the lone-pair
2 N2. This is mainly due to the very long and weak N−N electrons of E interact with the vacant p(π) orbital of the other
bonds in N4 (Td), which are caused by the strong Pauli fragment as shown in Figure 26d. This leads to nonplanar C2h
repulsion that originates from the similar size of the 2s and structures of the respective X2EEX2 molecules where atom E has
2p orbitals. a pyrimidal ligand arrangement. This is a straightforward
7.4.5. Multiple Bonds of Heavy Main-Group Atoms: explanation for the experimentally observed and calculated
X2CCX2 vs X2EEX2 (E = Si − Pb; X = H, F). The history of nonplanar geometries of many heavier group-14 homologues of
experimental attempts to synthesize compounds that are heavier alkenes. Model C refers to a hypothetical situation with mutual
group-14 homologues of alkenes and alkynes is a fascinating dative bonds in a planar structure X2EEX2. It is used for the
chapter of chemical research. The various approaches, which discussion of electron-sharing and dative bonding in C2H4 and
finally suceeded in an albeit unexpected way, have been C2F4 in section 8.
described in several review articles.269−297 They shall not be There is an alternative method of dative bonding
discussed in this account, which focuses on the nature of between EX2 fragments in their (1A1) singlet state where the
chemical bonding. Experiments showed that heavy alkene donation occurs from the E−X bonding orbital rather than the
species X2EEX2 (E = Si − Pb) possess nonplanar equilibrium lone-pair electrons of E. This is shown in Figure 26e (model D).
structures that are significantly different from the planar D2h Textbook knowledge suggests that lone-pair electrons are
arrangement of ethylene.269 Similarly, the experimentally energetically higher lying and are therefore better donors than
observed geometries of the heavier alkyne compounds XEEX bonding electrons, but this is strictly valid only for atoms of the
8817 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Figure 27. Calculated geometries and relative energies of E2H4 and E2F4 isomers (E = C−Pb) at BP86/def2-TZVPP. Number of imaginary frequencies
i. Bond lengths in Å, angles in deg, energies in kcal/mol. The red numbers indicate the lowest energy structures.

Table 16. Calculated Bond Dissociation Energies De of Planar X2EEX2 into 2 EX2 in the 3B1 Triplet State and (in Parentheses)
BDE of the Lowest Energy Structure of E2X4 to 2 EX2 in the 1A1 Singlet Statea
E2H4 E2F4
De ΔE(S→T) De − 2ΔE(S→T) De ΔE(S→T) De − 2ΔE(S→T)
C 178.7 (210.3)c −15.8b 178.7 176.7 (74.3)c 51.2 (74.3)
Si 100.0 (67.8)c 16.5 67.0 111.5 (8.3)c 69.7 −27.9
Ge 93.7 (51.3)c 23.3 47.1 91.0 (18.8)c 81.5 −72.0
Sn 74.4 (40.5)c 27.3 19.8 65.9 (31.9)c 80.2 −94.5
Pb 60.4 (38.5)c 35.5 −10.6 -d (37.8)c 89.9 -
a
Excitation energies ΔE(S → T) of EX2 at BP86/def2-TZVPP. All values in kcal/mol. bThe triplet state is the ground state. cDissociation energy of
the lowest energy structure of E2X4 to 2 EX2 in the singlet state. dPlanar Pb2F4 dissociates during the geometry optimization into 2 PbF2.

first octal row. As mentioned above, the (n)p AOs of higher order suggests that the E−X bonds become better donors than
main-group atoms have larger radii than the respective (n)s AO, the lone-pair electrons for heavier E and that the turning point
which leads to an increasingly higher %p character of the heavier depends on the nature of X. This occurs in the case of E2F4
atoms in their bonding orbitals. As a consequence, the lone-pair already for the germanium system, because fluorine is more
orbitals of the heavier atoms have a higher %s character which electronegative than hydrogen. This leads to a higher %s
lowers their energy. At some point the lone-pair orbitals are lower character of the lone-pair orbitals.
in energy and less prone to charge donation than the bonding electron The strength of the electron-sharing double bonds X2EEX2
pairs of E−X bonds. This is the reason why many low-valent main- in the hypothetical planar forms of the molecules may be
group compounds of the heavier elements possess bridging structures. estimated from the calculated bond dissociation energies De of
The rich and seemingly exotic chemistry of the cluster the reaction X2EEX2 (D2h) → 2 EX2 (3B1), which refers to EX2
compounds of the heavier main group elements is also based in the electronic reference state. The comparison with the BDEs
on the stronger donor capabilities of the bonding orbitals than of of the most stable nonplanar structures reveals the different
the lone pairs. This is elaborated and discussed in the paper by stabilization energies that come from electron-sharing and
Dehnen and co-workers in this issue.302 donor−acceptor interactions. Table 16 gives the computed data
Figure 27 shows the calculated geometries and the relative together with the singlet−triplet excitation energies. Subtracting
energies of the planar D2h form of E2X4 (E = C−Pb), which are the requested promotion energies ΔE(S → T) for obtaining the
3
not energy minima, and the isomeric equilibrium structures with B1 state of EX2 from the De values indicates the net stabilization
trans-bent geometry and doubly X-bridging isomers, all of C2h of the electron-sharing interactions. It becomes obvious that the
symmetry. The most stable isomer of the E2H4 compounds for E energy preference for the nonplanar structures of Si2H4 and
= Si, Ge is the trans-bent form whereas the hydrogen-bridged Ge2H4 surpasses only slightly the stronger electron-sharing
trans isomer is the energetically lowest lying species for E = Sn, bonding in the planar forms, which requires excitation to the 3B1
Pb. The most stable isomer for E2F4 is the trans-bent form for E state of EH2. The difference of the binding interactions becomes
= Si and the F-bridged isomer for E = Ge, Sn, Pb. There are also much larger for Sn2H4 and Pb2H4, where the dative bonding in
cis isomers of the bent structures and the bridged forms of E2X4, the lowest lying trans-bridged form occurs via E-H donation.
which are slightly higher in energy than the respective trans The situation in fluorine-substituted systems E2F4 reveals
isomers. They are not discussed here, because their bonding significant differences to the E2H4 compounds. The BDE of
situation is similar to those in the trans isomers. The stability planar C2F4 (74.3 kcal/mol), which is even less than the
8818 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

calculated BDE of the F−F single bond in C2F6 (87.3 kcal/ The comparison of the EDA-NOCV results for the lp-bridged
mol),237 is much smaller than the intrinsic interaction energy form of Sn2H4(B) with the H-bridged isomer Sn2H4(C) reveals
(176.7 kcal/mol) due to the rather large ΔE(S → T) value for the origin of the change in the relative energy of the two isomers.
CF2 (51.2 kcal/mol). The bonding situation in C2F4 and the The lp-bridged form Sn2H4(B) encounters identical Pauli
change in the electronic structure during bond rupture are repulsion and stronger electrostatic attraction than Sn2H4(C)
discussed in the section on dative bonding below. The electron- (Table 17). However, the orbital interactions in Sn2H4(C) are
sharing interactions in the planar structures of the heavier stronger than in Sn2H4(B), and they overcompensate the former
systems E2F4 have a similar magnitude as in the respective two effects. Thus, the EDA-NOCV results suggest that the
hydrogen systems E2H4, but the singlet → triplet excitation higher stability of Sn2H4(C) compared with Sn2H4(B) is due to
energies of EF2 are much higher than for EH2. This makes the the stronger charge donation of the Sn−H bonding orbital than
planar structures of the former much less stable than the bridged the lone-pair MO. The orbital interactions in Sn2H4(C) also
species. As mentioned above, the E−X bonds at some point comprise two major components ΔEorb(1) and ΔEorb(2), which
become stronger donors than the lone-pair orbitals. The gradual are the (+,+) and the (+,−) combinations of the Sn−H donor
change of the relative strength of the two types of interactions orbitals. This becomes obvious by inspection of the associated
can be monitored by EDA-NOCV calculations. deformation densities, which are displayed in Figure 28b. The
Table 17 shows the numerical results of the EDA-NOCV shape of Δρ and the connected HOMO and LUMO orbitals
calculations of the energetically lowest lying bridged isomers of show that the (+,+) combination of the Sn−H donor orbitals
provides, in this case, more stabilization than the (+,−)
Table 17. EDA-NOCV Results of the Most Stable Structures combination.
of E2H4 at BP86-D3/TZ2P+//BP86/def2-TZVPP The fluorine-bridged compounds differ significantly from the
hydrogen-bridged molecules, because fluorine carries lone-pair
Lp Lp Lp-
bridged bridged bridged H-bridged H-bridged electrons. Table 18 shows the EDA-NOCV results for the
(C2h) (C2h) (C2h) (C2h) (C2h) energetically lowest lying bridged isomers of E2F4. The turning
Isomer Si2H4(B) Ge2H4(B) Sn2H4(B) Sn2H4(C) Pb2H4(C) point from lp-bridging to more favorable F-bridging occurs
2 SiH2 2 GeH2 2 SnH2 2 SnH2 2 PbH2 already for E = Ge. The stabilizing interaction energy in the F-
fragments (1A1) (1A1) (1A1) (1A1) (1A1)
bridged isomer Ge2F4(C) is substantially larger (ΔEint = −41.7
ΔEint −81.3 −60.5 −40.7 −49.5 −48.0 kcal/mol) than in the lp-bridged species Ge2F4(B), which is
ΔEPauli 294.8 202.6 124.3 124.3 125.4 barely stabilized (ΔEint = −2.3 kcal/mol). Table 18 shows that
ΔEdisp −3.3 −2.8 −3.2 −3.7 −3.9 the Pauli repulsion in Ge2F4(C) is much larger than in
ΔEelstata −181.2 −138.2 −89.2 −86.1 −91.7 Ge2F4(B), but the electrostatic stabilization and the stronger
(48.6%) (53.1%) (55.2%) (50.6%) (54.1%)
orbital interaction compensate for the increase of ΔEPauli. The
ΔEorba −191.6 −122.2 −72.5 −84.1 −77.8
(51.4%) (46.9%) (44.8%) (49.4%) (45.9%) largest contribution to the orbital interactions comes from the
ΔEorb(1)b −149.9 −81.0 −39.8 −47.4 −45.2 donation of the σ lone-pair MOs of the bridging fluorine (Figure
(78.2%) (66.3%) (54.8%) (56.4%) (58.1%) 29a). There are several smaller orbital contributions, which are
ΔEorb(2)b −37.6 −36.8 −29.3 −26.8 −25.1 associated with a significant polarization of the electronic charge.
(19.6%) (30.1%) (40.5%) (31.9%) (32.3%)
This becomes visible by inspection of the deformation density
ΔEorb(rest)b −4.2 −4.4 −3.4 −9.9 −7.5
(2.2%) (3.6%) (4.7%) (11.7%) (9.7%) Δρ2 (Figure 29b). The red area of charge depletion is localized,
ΔEprep 14.1 9.7 4.7 5.5 5.3 but the blue area of charge accumulation is distributed over the
-De −67.2 −50.8 −35.9 −44.0 −42.6 whole molecule.
a 7.4.6. Multiple Bonds of Heavy Main-Group Atoms:
The values in parentheses give the percentage contribution to the
total attractive interactions ΔEelstat + ΔEorb. bThe values in HCCH vs HEEH (E = Si − Pb). The peculiar equilibrium
parentheses give the percentage contribution to the total orbital structures of the heavier group-14 homologues of actylene can
interactions ΔEorb. be explained in the same fashion as the nonplanar bridged
homologues of ethylene. Since the structures and bonding
situation in E2H2 (E = Si − Pb) have been discussed in recent
review articles,261,303 only the essential points are presented
E2H4. The tin compound Sn2H4 is the turning point where the here. More details can be found there and in the original
H-bridged form C becomes lower in energy than the lp-bridged work.304 Figure 30 summarizes the most important features of
isomer B (Figure 27). Therefore, we present for comparison the the E2H2 species.
results for both species Sn2H4(B) and Sn2H4(C). The calculated The nonplanar doubly hydrogen-bridged structure A is the
energy terms for the lp-bridged isomers B of E2H4 (E = Si − Sn) global energy minimum for all E2H2 molecules. The next
suggest that the attractive E−E interactions arise with similar energetically stable isomer is the singly bridged form B, which
strength from electrostatic forces and from orbital interactions. exhibits a peculiar inward bending of the terminal hydrogen
The ΔEorb term has two major components, which contribute atom. Species B has all three atoms on the same side of the triply
>95% to the dative bonding. Figure 28a shows the deformation coordinated atom E. This is a highly unusual bonding situation
densities Δρ of Sn2H4, which are associated with the individual for a group-14 atom. Both E2H2 isomers A and B have been
orbital interaction. The strongest contribution ΔEorb(1) comes observed for all elements E = Si − Pb in low-temperature matrix
from the mutual donation of the (+,−) combination of the lone- studies.305−310 The vinylidene species C is the next energetically
pair orbitals whereas ΔEorb(2) is due to the charge donation from higher isomer. It is the only energy minimum structure that is
the (+,+) lp combination. The associated interacting orbitals common for carbon and the heavier group-14 atoms Si−Pb.
HOMO and LUMO are also shown. It becomes obvious that the Finally, there are two trans-bent structures D1 and D2, which
interfragment interactions lead also to a significant charge possess significantly different E−E bond lengths and E−E−H
alteration (polarization) at each fragment. angles. Substituted homologues E2R2 with bulky groups R
8819 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Figure 28. Plot of deformation densities Δρ (isovalue = 0.005) of the pairwise orbital interactions ΔEorb(1) and ΔEorb(2) and the associated fragments
orbitals in the (a) Lp-bridged isomer of Sn2H4 and (b) H-bridged structure of Sn2H4. The energy values are given in kcal/mol. The direction of the
charge flow is red → blue.

Table 18. EDA-NOCV Results of the Most Stable Structures EH requires formally the a4Σ− excited state of EH for the
of E2F4 at BP86-D3/TZ2P+//BP86/def2-TZVPP formation of an electron-sharing triple bond. The excitation
energy X2Π → a4Σ− of the heavier EH species is significantly
Lp Lp-
bridged bridged F-bridged F-bridged F-bridged larger than for CH. Table 19 shows the calculated bond
(C2h) (C2h) (C2h) (C2h) (C2h) dissociation energies De of the linear structures HEEH
Isomer Si2F4(B) Ge2F4(B) Ge2F4(C) Sn2F4(C) Pb2F4(C) yielding (a4Σ−) EH. Subtracting the excitation energies for two
2 SiF2 2 GeF2 2 GeF2 2 SnF2 2 PbF2 EH fragments to the a4Σ− state ΔEexc from the De values gives
fragments (1A1) (1A1) (1A1) (1A1) (1A1)
the net stabilization energy of the electron-sharing triple bond.
ΔEint −9.7 −2.3 −41.7 −49.7 −53.3 This value of 240.0 kcal/mol for CH is rather high; it clearly
ΔEPauli 68.6 21.7 161.7 135.4 113.6 surpasses the stabilization of an electron-sharing single bond,
ΔEdisp −3.4 −2.9 −3.3 −3.9 −4.2 which could be provided by coupling the unpaired electrons of
ΔEelstata −29.4 −5.0 −112.0 −109.6 −99.4 the EH fragments in the X2Π ground state. The net stabilization
(39.3%) (23.8%) (56.0%) (60.5%) (61.1%)
ΔEorba −45.5 −16.1 −88.0 −71.6 −63.3
energy De − 2ΔEexc for Si and Ge is much smaller (Table 19). It
(60.7%) (76.2%) (44.0%) (39.5%) (38.9%) does not compensate for the energy gain of a E−E single bond.
ΔEorb(1)b −26.0 −11.1 −52.0 −39.5 −34.9 The excitation energies of two SnH and PbH fragments are even
(57.3%) (69.3%) (59.0%) (55.2%) (55.1%) higher than the bond dissociation energy of linear HEEH. It
ΔEorb(2)b −17.9 −4.2 −13.7 −11.1 −10.1 follows that the bonding interactions between two EH species
(39.4%) (26.2%) (15.6%) (15.5%) (15.9%)
for E = Si − Pb occur from the X2Π ground state. The isomer C
ΔEorb(rest)b −1.5 −0.7 −22.3 −21.0 −18.3
(3.4%) (4.5%) (25.4%) (29.3%) (29.0%) is not considered, because it has a different atomic connectivity
ΔEprep 0.3 0.2 20.3 14.7 12.1 than the other species.
−De −9.4 −2.1 −21.4 −35.0 −41.1 Figure 32 shows three possible arrangements of the EH
a fragments in their X2Π ground state where the unpaired
The values in parentheses give the percentage contribution to the
total attractive interactions ΔEelstat + ΔEorb. bThe values in electrons are coupled to an electron-sharing σ bond. The
parentheses give the percentage contribution to the total orbital assignments (a) and (c) are unfavorable, because they leave the
interactions ΔEorb. unoccupied p(π) orbitals in the resulting species vacant.
Consequently, the structures are not energy minima but
transition states (i = 1). The arrangement (c) leads to structure
featuring D1 have been synthesized and structurally charac- D2, which becomes an energy minimum for E = Pb when bulky
terized for E = Si, Ge, Sn.311−314 The lead analogue Pb2R2 substituents prevent the formation of the electronically more
exhibits structure D2.315 The linear form E is an energetically favored isomers.316 The arrangement (b) in Figure 32 places the
high lying second order saddle point (number of imaginary EH bonds in perfect position to donate the associated electron
frquencies i = 2) for all heavy group-14 species E2H2. pairs into the respectively vacant p(π) orbitals. This leads to the
Figure 31 schematically shows the X2Π ground and the a4Σ− global energy minimum form A of the heavier E2H2 molecules. It
excited state of EH. It is obvious that the linear structure HE becomes clear that the structure A has a triple bond, which
8820 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Figure 29. Plot of deformation densities Δρ (isovalue = 0.005) of the pairwise orbital interactions ΔEorb(1) and ΔEorb(2) and the associated fragments
orbitals in the (a) Lp-bridged isomer of Ge2F4 and (b) F-bridged structure of Ge2F4. The energy values are given in kcal/mol. The direction of the
charge flow is red → blue.

consists of one electron-sharing σ component and a degenerate modified Lewis type structures for the isomers A, B, D1, and D2
EH donor component. as shown in Figure 34. It turns out that the Lewis model should
Figure 33 shows three different placements of the (X2Π) EH not be discarded or replaced, but it should be used in
fragments where the unpaired electrons yield a π bond. conjunction with quantum chemical calculations.
Placement (a) shows that the electron-sharing π bond is The following conclusions arise from these sections:
supported by a lone-pair donor bond and a EH donor bond. The
optimal EH donation requires a tilting of the vacant p(π) orbital • The nonplanar equilibrium geometries of the heavier
of the right EH fragment, which leads to the inward bending of ethylene homologues E2X4 (E = Si − Pb) can be
the associated E−H bond. The formation of the EH donor bond understood in terms of dative interactions between
nicely explains the unusual equilibrium structure of the isomer ylidene fragments EX2 in the electronic (X1A1) ground
B, which has a triple bond that includes an electron-sharing π state. The excitation energy from the (X1A1) ground state
bond, a lone-pair donor bond, and a EH donor bond. Rotation of to the 3B1 excited state is too high to make the planar
one EH fragment leads to structure G, which is not an energy species with electron-sharing double bonds X2EEX2
minimum, although it is lower in energy than B. Structure G energy minima.
possesses two EH donor bonds, which are stronger than one EH
and one lone-pair donor bond in B. But G is the transition state • The dative bonds in E2X4 can be formed either via the
for the degenerate flip−flap rearrangement of the global energy lone-pair electrons or by the E−X bond pairs as donors,
minimum A. Finally, there is the energetically highest lying which become favorable when E is a heavier atom. The
energy minimum D1, which has a triple bond consisting of an turning point from lone-pair donation to E−X donation
electron-sharing π bond and two lone-pair donor bonds. depends also on the nature of X.
Substitution of hydrogen by bulky groups in E2H2 stabilizes • The unusual equilibrium structures of the heavier
D1 so much that it can be isolated. When E = Pb, the isomeric acetylene homologues E2H2 (E = Si − Pb) can be
form D2 becomes lower in energy than D1, because the explained analogously to the heavier ethylene systems
electron-sharing π bond is replaced by a stronger σ bond and the with the interaction of the EH fragments in the (2Π)
lone-pair donation of lead is rather weak. doublet ground state. The E−E bonds in the lowest lying
The model presented here straightforwardly explains the energy minima A, B, and D1 possess electron-sharing σ or
experimentally observed structures of E2H2, which possess π single bonds, which are completed by lone-pair or E−H
equilibrium geometries that are not directly accessible by the dative bonds.
Lewis bonding model. They can be straightforwardly explained
with an extension of the model by Trinquier and Malrieu298,299 • The trans-bent equilibrium geometries of the isolated
and by Carter and Goddard,300 that was suggested for heavier systems E2R2 are enforced by steric repulsion of the bulky
homologues of ethylene. With the information that is gained groups R. In the case of E = Si, Ge, Sn, they have the
from the analysis of the orbital interactions in E2H2 it is possible structure D1 with a triple bond whereas Pb2R2 features
to display the bonding situation in the isomers in terms of structure D2 with a single bond.

8821 DOI: 10.1021/acs.chemrev.8b00722


Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Figure 30. Optimized structures of E2H2 isomers A−E at BP86/QZ4P. Bond lengths are given in Å, angles in deg. The values for Θ give for structure A
the dihedral angle between the E2H and E2H′ planes. The relative energies with respect to A are given at the bottom of each entry in kcal/mol together
with the number of imaginary frequencies i. Reproduced from ref 304. Copyright 2005 American Chemical Society.

8. DATIVE BONDING IN MAIN-GROUP COMPOUNDS decade following the suggestion that the chemical bonds and the
reactivity of carbodiphosphorane C(PPh3)2 can best be
In the examples of chemical bonding that were discussed so far, a
understood in terms of donor−acceptor interactions Ph3P→
distinction was often made between electron-sharing bonding
C←PPh3.50 The proposal not only led to an understanding of
and dative bonding. As mentioned above, the concept of dative the sometimes unusual geometries of related molecules such as
bonding and the associated notation with an arrow A→B in the bent equilibrium geometry of carbon suboxid C(CO)2;52,53
contrast to an electron-sharing bond A−B is quite old, but it was it was also the starting point for the synthesis of new molecules
neglected by the strong influence of Pauling on the general view with uncommon features.49,55,56 The term “carbone” was
of chemical bonding in the following period. Arne Haaland was a suggested for divalent carbon(0) compounds CL2, which exhibit
lonely caller in the desert when he published a review article on a distinctly different chemical behavior than the carbenes CR2,
donor−acceptor interactions in main-group chemistry in which are carbon(II) compounds.54 Carbones possess two lone
1989.48 It received little attention, which is probably due to electron pairs whereas carbenes have only one. A particular class
the fact that the work was restricted to well-known standard of carbones CL2 are carbodicarbenes C(NHC)2, which possess
molecules. The model of dative bonding was revived in the past carbon−carbon dative bonds (NHC)→C←(NHC). Carbodi-
8822 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

guidance for synthesizing one-center complexes EL2, it can


also be applied to multicenter complexes. Several examples
where diatomic species E2 are stabilized by ligands L in stable
complexes L→E2←L have been discussed in section 7.3 above.
The basic principle of the stabilizing interactions in the two-
center adducts E2L2 is the same as in the one-center complexes
EL2.49 The mono- or diatomic fragment E or E2 is formally
excited to an electronic reference state, which is usually an
excited state and which makes the fragment a good acceptor.
The dative interactions L→En←L are strong enough to stabilize
the complex so it can be experimentally observed or even
isolated in bulk form. Recently, the concept was extended to
systems with triatomic cyclic systems E3L3 (Figure 36). The
triatomic Si(0) cluster Si3(CAAC)3, which has three Si⇆CAAC
donor−acceptor bonds, could be isolated and structurally
characterized by X-ray analysis.334 The triboron complexes
[B3(NN) 3] + and [B 3(CO)3 ]+ have been observed and
spectroscopically identified in the gas phase.335 The molecules
Figure 31. Schematic representation of the electron configuration of feature the smallest π-aromatic system B3+, which is stabilized by
the 2Π electronic ground state and the a4Σ− excited state of EH (E = C three dative bonds with the ligands N2 and CO.
− Pb). The experimental24 and calculated (BP86/QZ4P) excitation The usefulness to distinguish between electron-sharing and
energies are given in kcal/mol. dative bonding for understanding and also explaining chemical
reactions can be demonstrated with a recent example, in which
Table 19. Calculated Bond Dissociation Energies De (kcal/ very similar substrates led to completely different products.
mol) of Linear HEEH → 2 EH (a4Σ−) and X2Π → a4Σ− Roesky and co-workers reported in 2009 that the reaction of
Excitation Energies ΔEexc (kcal/mol) of EH at BP86/QZ4P NHC with HSiCl3 leads to the synthesis of the complex NHC→
SiCl2 (A), which was the first ligand stabilized Si(II) compound
E De ΔEexc De − 2xΔEexc
that was isolated and structurally characterized by X-ray analysis
C 270.9 15.44 240.0 (Reaction 25, Figure 37).336 In a subsequent work, the authors
Si 121.6 38.56 44.5 reacted A with CAAC, which according to quantum chemical
Ge 113.3 47.09 19.0 calculations binds more strongly to SiCl2 than NHC.
Sn 89.4 45.87 −2.3 Surprisingly, the reaction gave the compound (SiCl2)(CAAC)2
Pb 69.0 52.01 −35.0 (B) where silicon is four-coordinated, rather than the expected
complex (CAAC)→SiCl2 via 1:1 ligand exchange (Reaction 26,
carbenes were theoretically predicted as stable compounds in Figure 37)337 The calculations showed that the Si−C bonds in
2007,51 and they were synthesized for the first time in (SiCl2)(CAAC)2 are significantly shorter than in (CAAC)→
2008.317,318 It was recently found that carbodicarbenes may be SiCl2.
used as catalysts in a variety of reactions, which are a topic of
intense experimental investigations.319−324 HSiCl3 + 2NHC → NHC→SiCl 2(A) + NHC(H+)(Cl−)
Following the discovery of carbones, heavier group-14 (25)
homologues EL2 (E = Si − Pb) were proposed as synthetically
achievable targets.325,326 Stable silylones SiL2, germylones GeL2, NHC→SiCl 2(A) + 3CAAC
and stannylones SnL2 have been isolated in the mean-
time.327−329 Perhaps the most amazing outcome of the carbone → (SiCl 2)(CAAC)2 (B) + NHC‐CAAC (26)
model CL2 was the synthesis of the isoelectronic borylene
complex (BH)L2 with L = CAAC, which has a three-coordinated The unexpected formation of four-coordinated (SiCl2)-
boron atom that possesses an electron pair CAAC→(BH)← (CAAC)2 rather than (CAAC)→SiCl2 can straightforwardly
CAAC.330 Substituted homologues (BR)L2 with different be explained with the presence of Si−C electron-sharing bonds
ligands L have also been isolated in the meantime, of which in the former compound in contrast to the latter adduct, which
the dicarbonyl complex (BR)(CO)2 is the most surprising has a Si←C dative bond. The formation of electron-sharing Si−
example.331,332 It is the first stable dicarbonyl complex of boron C in (SiCl2)(CAAC)2 (B) requires open-shell fragments in the
and the only dicarbonyl complex of any main-group atom electronic triplet state, which leaves one unpaired electron at
besides carbon suboxide C(CO)2 and the isoelectronic each CAAC ligand (Figure 37). According to Hund’s rule, the
nitronium cation N+(CO)2.333 The examples demonstrate the overall electronic state of (SiCl2)(CAAC)2 should be a triplet.
value of a bonding model, which is based on a quantum chemical The EPR spectrum actually suggests a triplet state for the
analysis of the electronic structure of the molecules. Figure 35 compound. Interestingly, compound B possesses two poly-
gives an overview of the calculated and experimental structures morphic forms, where only one exhibits the EPR spectrum.337 It
of carbones CL2 and isoelectronic boron and nitrogen appears that the unpaired electrons in one isomer are weakly
homologues (BH)L2 and N+L2 with various ligands. The trend coupled via intermolecular interactions, making it diamagnetic.
in the bond angles can be explained with the relative strength of There remains the question why SiCl2 binds two CAAC ligands
the π backdonation from the ligands to the central atom.49,58 with covalent bonds while it only binds one NHC ligand with a
The model of dative bonding is not only useful as a means of dative bond. The associated energies of bond formation and
interpretation of the bonding situation and as predictive electronic excitation provide a straightforward answer.
8823 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Figure 32. Qualitative model for the orbital interactions between two EH molecules in different orientations where the unpaired electrons yield a σ
bond. Reproduced from ref 304. Copyright 2005 American Chemical Society.

Figure 34. Suggested Lewis structures for isomers A, B, D1, and D2 of


E2H2 isomers.

gives only a net stabilization of 7.3 kcal/mol, because the singlet


→ triplet excitation energy of NHC (88.9 kcal(mol) is much
Figure 33. Qualitative model for the orbital interactions between two
higher than for CAAC (49.9 kcal/mol). The energies calculated
EH molecules in different orientations where the unpaired electrons on the basis of the two bond types provide not only an
yield a π bond. Reproduced from ref 304. Copyright 2005 American explanation for the experimental observation. They serve also as
Chemical Society. a guideline for predicting the experimental outcome of further
reactions with different carbene ligands, which mainly depends
on the energy difference between their singlet and triplet state.
Figure 37 gives the calculated bond dissociation energies of How can a dative bond, A→B, be distinguished from an
the mono and bis adducts of SiCl2 with NHC and CAAC ligands electron-sharing one, A−B? The question seems to be easily
and the singlet → triplet excitation energies.337 The bottom end answered by looking at the fragments that arise after breaking
shows the energy balance for electron-sharing bonding of SiCl2 the bond. Dative bonds tend to dissociate heterolytically while
with NHC and CAAC. The sum of energy gained due to bond electron-sharing bonds usually break up in a homolytic fashion.
formation and singlet → triplet excitation energies for the CAAC However, the situation is more complicated, because the nature
ligand results in a net stabilization of 67.3 kcal/mol for of the bonding interactions may change during the bond
(SiCl2)(CAAC)2, which surpasses the BDE of (SiCl2)(CAAC) breaking process. A good example is the carbon−carbon double
(42.5 kcal/mol). In contrast, the formation of (SiCl2)(NHC)2 bond F2CCF2, which is a classical electron-sharing bond that
8824 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Figure 35. List of isoelectronic molecules EL2 with calculated bond angles and partial charges Δq of the central fragments E = BH, C, N+. Experimental
bond angles of isolated molecules are given in parentheses. Reproduced from ref 49. Copyright 2017 Elsevier.

consists of a σ and a π component. But the fragments, which are revealed by the values of the orbital interactions ΔEorb values.
eventually formed during the bond-breaking process, are two As mentioned above, those fragments, whose electronic energies
CF2 species in the singlet (1A1) state. Since bond breaking is the change least during the covalent bond formation, provide the
reverse process of bond formation, the initial interaction must best description of the chemical bond. This criterion was used
have a dative nature (see model B in Figure 26d). A different for a variety of chemical bonds where the description in terms of
situation exists for ethylene H2CCH2, because the interacting dative or electron-sharing interactions was not clear.133−146
methylene fragments have a triplet (3B1) ground state, which Table 21 shows that the ΔEorb values using triplet fragments are
straightforwardly leads to electron-sharing σ and π bonding in smaller than with singlet fragments for H2CCH2 and F2C
the molecule (model A, Figure 26c). The different electronic CF2, which supports the notion that also the latter compound
states of the carbene fragments have a drastic effect on the BDE has an electron-sharing double bond. The change from electron-
of the alkenes. Table 20 shows the calculated values of the sharing bonding at equilibrium to dative bonding during
alkenes C2X4 and alkanes C2X6 (X = H, F). The De values for the carbon−carbon bond breaking can be monitored by EDA
hydrogen bonded parent systems show the expected order; that calculations along the dissociation profile. Figure 38 shows the
is, the H2CCH2 double bond is much stronger than the H3C− calculated reaction pathway for the carbon−carbon bond
CH3 single bond. In contrast, the F2CCF2 double bond is breaking/bond formation reaction of C2F4 at different levels of
weaker than the F3C−CF3 single bond, although the former
theory.237 The calculations suggest a smooth reaction profile
bond is even shorter than in H2CCH2. It becomes obvious
where the CF2 fragments initially approach each other in a side-
that the bond dissociation energy is not a reliable measure for
way orientation. The molecule becomes planar only at the final
the strength of a chemical bond.
The carbon−carbon double bonds in H2CCH2 and F2C stadium of the bond formation. There is no activation barrier for
CF2 have recently been studied in terms of dative and electron- the reaction; the small hump at the CASSCF(8,8) level is due to
sharing bonds with the EDA-NOCV method.237 Table 21 gives the lack of dynamic correlation.
the numerical results of the calculations, where the carbene Table 22 shows the calculated ΔEorb values at different C−C
fragments were arranged in the singlet and triplet states distances using the singlet and triplet electronic states of CF2,
according to bonding models A and C (Figures 26c and 26e). which refer to the three different bonding models A, B, and C.
It becomes obvious that the calculated interaction energies ΔEint The electron-sharing model A provides the best description for
between the triplet fragments in H2CCH2 and F2CCF2 are shorter distances as suggested by the ΔEorb values, but for longer
a more faithful description than the bond dissociation energies distances the dative model B becomes better than A. Model C is
De. The ΔEint values give the strength of the stabilizing not relevant at any carbon−carbon distance. The trend of the
interactions in the eventually formed bond while the De values ΔEorb values shows that the change from electron-sharing to
for C2F4 include the electronic relaxation of the CF2 fragments dative bonding can be illustrated with the EDA-NOCV method.
from the triplet to the singlet state. It should be kept in mind that the transition involves a change of
The preference for describing the double bonds in terms of the model and not a change in the nature of the interatomic
electron-sharing interactions rather than dative bonds is interactions.
8825 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Figure 36. (a) Geometry and bonding situation of the isolated complex Si3(CAACR)3 with experimental and (in parentheses) calculated bond lengths
(Å) and angles (deg). Reprinted with permission from ref 334. Copyright 2016 John Wiley and Sons. (b) Calculated geometry and bonding situation of
the experimentally observed aromatic cations B3L3+ (L = CO, N2). Reprinted with permission from ref 335. Copyright 2016 John Wiley and Sons.

Finally, we would like to discuss three systems LX2L with for compound 2, where the dative interactions bz-NHCMe→
valence isoelectronic species X2 = N2, C2H2, and Si2H2, where C2H2←bz-NHCMe (2b) lead to a smaller ΔEorb value than the
the dichotomy of dative and electron-sharing bonds is a electron-sharing model 2a, which refers to a tetraaminobuta-
challenge for chemical intuition. Figure 39 shows three diene. There is chemical evidence that 2b is indeed a more
molecules 1−3, which may be written in either of the two faithful representation of the chemical bonds than 2a. When the
depicted ways. Compound 1 (E = N) with a bz-NHCMe ligand dication 22+ is reacted with silver or gold compounds, the in situ
(bz-NHCMe denotes a benzoannelated NHCMe ligand with generated 2 reacts and gives a major product the metal
methyl groups at nitrogen) was synthesized in 2012, where it was complexes bz-[NHCMe→M←bz-NHCMe]+ (M = Ag, Au)
termed as a guanidine compound, which was sketched with the where the metal cation replaces the central C2H2 moiety.140
classical Lewis structure 1a.338 The above-described bonding Such reaction is unknown for tetraaminobutadiene compounds.
analysis of N2(PPh3)2, which was recognized as donor−acceptor The EDA-NOCV calculations for Si2H2(CAACDip) (Dip = 2,6-
complex Ph3P→N2←PPh3, let it seem possible that compound diisopropylphenyl) suggest that the description in terms of
1 is better described with dative bonds bz-NHCMe→N2←bz- dative and electron-sharing bond have essentially equal weight.
Consequently, the synthesis of 3 was reported with the title “The
NHCMe (1b). The choice of the bz-NHCMe ligands was made,
Structure of the Carbene Stabilized Si2H2 May Be Equally Well
because the molecule was supposed to be compared with the
Described with Coordinate Bonds as with Classical Double
isoelectronic species C2H2(bz-NHCMe) (2), which was reported
Bond”.136
to have some surprising chemical behavior.140 Very recently, the The following conclusions arise from this section:
silicon system 3 was reported, which may be considered as
ligand stabilized disilaacetylene complex 3b or as substituted • The distinction of the covalent electron-pair bonding into
2,3-disilabutadiene 3a.136 electron-sharing and dative interactions is a helpful model
Table 23 shows the results of the EDA-NOCV calculations of for understanding the structures and reactivities of
1−3. The data suggest that the chemical bonds at the central N2 molecules. It is also a powerful tool for predicting stable
fragment in 1 are indeed better described in terms of dative molecules with unusual chemical bonds.
interactions bz-NHCMe→N2←bz-NHCMe (1b) than with • Dative bonds A→B usually tend to dissociate heterolyti-
electron-sharing bonds (1a). The same conclusion is drawn cally and electron-sharing bonds A−B normally break up
8826 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Figure 37. Reactions of SiCl2 with NHC and CAAC and schematic view of the bonding situation in the complex NHC→SiCl2 (A) and the molecule
SiCl2(CAAC)2 (B) in the triplet state. Below are the calculated bond dissociation energies at M05-2X/def2-TZVPP of the complexes SiCl2(NHC) and
SiCl2(CAAC) and the compounds in the triplet state SiCl2(NHC)2 and SiCl2(CAAC)2 as well as the singlet−triplet gaps of the fragments. The bottom
lines give the net stabilization energies ΔE for the formation of the electron-sharing bonds in SiCl2(NHC)2 and SiCl2(CAAC)2 in comparison with the
dative bond. Reprinted with permission from ref 337. Copyright 2013 John Wiley and Sons.

Table 20. Calculated (Experimental) C−C Bond Lengths Re Table 21. EDA Calculations of C2F4 and C2H4 at the M06-L/
[Å] and Calculated (Experimental) Bond Dissociation TZ2P Level Using Triplet and Singlet Fragments According
Energies De [kcal/mol]a to Models A and C (Figure 26)a
Re De C2F4 C2H4
H3CCH3 1.532 (1.522) 93.1 (89.7) Fragments Triplet (A) Singlet (C) Triplet (A) Singlet (C)
H2CCH2 1.333 (1.336) 178.2 (172.1) ΔEint −197.7 −279.9 −196.7 −278.1
F3CCF3 1.567 (1.545) 87.3 (96.4) ΔEMetaGGA 6.2 −4.8 7.9 7.6
F2CCF2 1.329 (1.311) 73.3 (70.3) ΔEPauli 305.0 292.9 291.4 282.8
a
Calculated values were obtained at the M06-L/TZ2P level of theory. ΔEelstatb −182.7 −175.6 −183.6 −181.2
Taken from ref 237. (35.9%) (30.9%) (37.0%) (31.9%)
ΔEorbb −326.1 −392.5 −312.4 −387.2
(64.1%) (69.1%) (63.0%) (68.1%)
homolytically, but the dissociation products are not
ΔEorb(σ)c −209.2 −221.5 −218.0 −241.2
always reliable probes for the type of the bonding. This is (64.2%) (56.4%) (69.8%) (62.3%)
because the nature of the interatomic interactions may ΔEorb(π)c −87.0 −142.0 −79.4 −129.9
strongly change during the bond formation in a way that (26.7%) (36.2%) (25.4%) (33.5%)
the initially dative interactions eventually turn into ΔEorb restc −29.9 (9.2%) −29.0 (7.4%) −15.0 (4.8%) −16.1 (4.2%)
electron-sharing bonding. ΔEprep 122.2 204.4 20.7 102.1
ΔE = −De −75.5 −75.5 −176.0 −176.0
• Dative and electron-sharing bonds are models, which are a
helpful tools for the description and prediction of the Energy values in kcal/mol. Taken from ref 237. bThe values in
structures and reactivities of molecules. There may be parentheses give the percentage contribution to the total attractive
cases where both descriptions have similar validity. interactions ΔEelstat + ΔEorb. cThe values in parentheses give the
percentage contribution to the total orbital interactions ΔEorb.

9. THE OCTET RULE AND THE ATOMIC VALENCE earlier suggested to explain the stabilty of so-called hypervalent
SPACE OF MAIN-GROUP ATOMS molecules, was discarded on the basis of quantum chemical
There is ample evidence that the atoms of groups 1, 2, and 13− calculations. Atomic orbitals with higher angular momentum
18 of the periodic system called main-group elements use only such as d and f functions only serve as polarization functions for
their s and p valence orbitals for chemical bonding. Since the the sp space, but they are not genuine valence orbitals in main-
exchange (Pauli) principle allows at most two electrons in one group compounds.339,340 The evaluation of the symmetry of the
orbitals, the four valence orbitals can accommodate a maximum molecular orbitals shows that molecular wave functions may be
of eight electrons. This is the quantum theoretical basis for the constructed that assign more than four bonds to an atom
octet rule, which was suggested by Langmuir already in 1921.6 although they only use its four atomic valence s and p functions
The extension of the valence space to d-orbitals, which was for chemical bonding. This has been discussed in section 7.4.
8827 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Figure 38. Calculated reaction pathway for rupture of the carbon−


carbon bond of C2F4 with different theoretical methods. The energy is
given with respect to the dissociation product 2 CF2 (1A1). Reprinted
with permission from ref 237. Copyright 2018 John Wiley and Sons.

Recently, joint experimental/theoretical studies reported the Figure 39. Schematic representation of alternative bonding situations
with dative or electron-sharing bonds for molecules with valence
observation and identification of the earth alkaline carbonyl
isoelectronic fragments. (a) N2(bz-NHCMe)2; (b) (C2H2)(bz-
complexes M(CO)8 (M = Ca, Sr, Ba) in low-temperature NHCMe)2; (c) (Si2H2)(CAACDip)2 (Dip = 2,6-diisopropylphenyl).
matrices, which possess chemical bonds that are typical for
transition metals.102 The molecules fulfill the 18-electron rule,
and the metal−carbonyl bonds were explained with the Dewar− Figure 40 shows the calculated equilibrium geometries of
Chatt−Duncanson (DCD) model1,341,342 in terms of M ← CO neutral M(CO)8 and the cations M(CO)8+ (M = Ca, Sr, Ba),
σ donation and M → CO π backdonation. Surprisingly, the latter which were observed in the gas phase.102 The molecules were
interactions involve the valence d orbitals of the earth alkaline identified by comparing experimental infrared spectra with
atoms, which are vacant in the electronic ground state. The calculated wave numbers. The complexes are stable with respect
electronic reference state of the metal atoms is the excited triplet to loss of one CO ligand by 8.4−11.5 kal/mol. The neutral
(3F) state with the electron configuration (n)s0(n−1)d2, which species exhibit strong red shifts of the single IR active C−O
is stabilized via strong M(d) → CO π backdonation. In a stretching mode by 130−160 cm−1. Also the cations show a red
forerunner of the work, the authors had found that the barium shift by 30−50 cm−1, which can be explained with one and two
carbonyl cation Ba(CO)+ exhibits a strong red shift of the C−O occupied d-orbitals in the cations and neutral molecules,
stretching mode, which was explained with large Ba(5d)+ → CO respectively. Figure 41 shows the correlation diagram of a
π backdonation.343 It is highly unusual that a cation serves as a metal with a (n)s, (n)p, (n−1)d shell and eight CO in a cubic
donor to a neutral species. The possible involvement of d (Oh) field. The qualitative model suggests that a total of 14
orbitals in barium compounds was earlier suggested by Pyykkö, electrons are donated from the CO lone-pair MOs to the metal
who coined the term “honorary transition metal” for barium.344 into (a) the triply degenerate t2g (n−1)d-AOs, (b) the triply
But it seems that transition metal-like bonding may not be degenerate t1u (n)p-AOs, and (c) the (n)s AO. One σ electron
restricted to barium. pair of the ligand cage has a2u symmetry, which does not match

Table 22. Calculated EDA Values at M06-L/TZ2P of the Orbital Term ΔEorb [kcal/mol] for the Interactions between CF2 with
Different Electronic States and Different C−C Distances dC−C [Å]a

dC−C 1.30 1.326a 1.40 1.50 1.60 1.70 1.80 1.90 2.00 2.20 2.40 3.00
Model A
ΔEorb −339.5 −326.1 −290.9 −245.6 −217.3 −195.0 −178.4 −165.9 −156.5 −144.3 −137.7 −131.4
Model B
ΔEorb −856.4 −802.0 −665.8 −420.9 −275.1 −188.0 −133.3 −96.4 −70.4 −38.1 −20.7 −3.9
Model C
ΔEorb −406.0 −392.5 −357.1 −359.4 −365.5 −349.3 −326.1 −303.3 −283.7 −255.0 −237.1 −215.6

a
The bold values depict the smallest ΔEorb value at the respective C−C distance. Data taken from ref 237.

8828 DOI: 10.1021/acs.chemrev.8b00722


Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Table 23. EDA-NOCV Calculations at the BP86-D3(BJ)/TZ2P Level of Theory of N2(bz-NHCMe)2, C2H2(bz-NHCMe)2, and
Si2H2(CAACDip)2 and Using the Interacting Fragments in the Singlet (S) or Quintet (Q) Electronic Statea
N2(bz-NHCMe)2 C2H2(bz-NHCMe)2 Si2H2(CAACDip)2
2 bz-NHCMe (S) Me
2 bz-NHC (Q) Me
2 bz-NHC (S) Me
2 bz-NHC (Q) Me
2 CAAC (S) 2 CAACMe (Q)
Fragments + N2 (S)c + N2 (Q) + C2H2 (S)c + C2H2 (Q) + Si2H2 (S)c + Si2H2 (Q)
Bonding dative electron-sharing Dative electron-sharing dative electron-sharing
ΔEint −393.5 −633.5 −465.3 −436.0 −193.0 −323.5
ΔEPauli 1275.3 1083.2 636.5 924.8 665.2 436.0
ΔEdisp −7.2 −7.2 −10.4 −10.4 −33.4 (3.9%) −33.4 (4.4%)
ΔEelstatb −547.8 (33.0%) −487.4 (28.5%) −385.4 (36.8%) −503.9 (37.3%) −423.9 (49.4%) −328.6 (43.3%)
ΔEorbb −1113.9 (67.0%) −1222.1 (71.5%) −706.1 (63.2%) −846.5 (62.7%) −401.0 (46.7%) −397.5 (52.3%)
a
Energy values are given in kcal/mol. The data are taken from refs 49, 136, and 140. bThe values in parentheses give the percentage contribution to
the total attractive interactions ΔEelstat + ΔEorb. cThe π and π* MOs in the electronically excited reference state are occupied.

Figure 40. Calculated equilibrium geometries of alkaline earth octacarbonyls. (A) M(CO)8 (M = Ca, Sr, or Ba), (B) [M(CO)8]+ (M = Ca or Sr), and
(C) [Ba(CO)8]+. Bond lengths are in angstroms. The D0 values in roman type are the ZPE-corrected bond dissociation energies for loss of one CO
ligand; the italicized values are the corresponding energies for the loss of eight CO ligands and M/M+ in the ground state. The values without
parentheses are from M06-2X-D3/def2-TZVPP calculations; the values in parentheses are from CCSD(T)/def2-TZVPP using M06-2X-D3/def2-
TZVPP optimized geometries. Reprinted with permission from ref 102. Copyright 2018 American Association for the Advancement of Science.

any valence orbital of the metal. The earth alkaline atoms have characteristic for transition metal compounds. The possible
therefore only 16-electrons in their valence shells. Although the extension of this concept to other atoms than the alkaline earth
complexes are formally 18-electron species, they are effectively elements and further ligands has yet to be explored. A guideline
16-electron complexes. The HOMO is a degenerate orbital, and for the search of other stable molecules is the excitation energy
the molecules are therefore in agreement with Hund’s rule from the occupied sp shell to the d orbitals and the strength of
triplets. The singly occupied MO of the cations induces due to the bonds to a ligand. This concept proved to be useful for
Jahn−Teller distortion a lower symmetry of the equilibrium finding the class of carbones and related compounds, which are
structure. mentioned in section 8. It might also be helpful for the search of
Table 24 shows the results of EDA-NOCV calculations on molecules where classical main-group atoms bind like transition
M(CO)8, which give among others a quantitative account of the metals.
individual orbital contributions that are sketched in Figure 41. The following conclusions arise from this section:
The data show that the most important valence orbitals of the
metals are the (n−1)d functions, which contribute 85%−95% to • The valence space of the main group atoms usually
the total orbital interactions. The dominant contribution is due comprises the outermost s and p orbitals, which is the
to the (n−1)d → (CO)8 π backdonation. Therefore, the alkaline basis for the octet rule. However, main-group elements
earth atoms calcium, strontium, and barium bind the CO ligands can also form stable molecules with covalent bonds using
in M(CO)8 like typical transition metals. The intrinsic their d-orbitals when the excitation energy to the d-shell is
interaction energies ΔEint between the metal atoms in the compensated by the binding energies. This has recently
excited 3F state and the (CO)8 ligand cage is very strong; it been demonstrated by the synthesis of the octacarbonyl
overcompensates the (n)s → (n−1)d excitation energies of the complexes M(CO)8 (M = Ca, Sr, Ba).
metals. The bonding analysis of the recently reported 20-
electron systems M(CO)8− (M = Sc, Y, La), which also have
cubic (Oh) symmetry, showed a very similar bonding pattern.345
10. RELATIVISTIC EFFECTS IN MAIN-GROUP
The HOMO in Figure 41 is fully occupied in the latter COMPOUNDS
compounds, which are thus effectively 18-electron complexes. When one moves to the heaviest elements in the periodic table,
The above results indicate that some atoms of the sp shell are relativistic effects need to be considered in chemical bonding.
able to use their d orbitals, which are empty in the electronic Since relativistic quantum theory is relatively a newcomer to the
ground state, for chemical bonds in stable molecules that are field of quantum chemistry, and from the theoretical point of
8829 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Figure 41. Bonding scheme and shape of the occupied valence orbitals of M(CO)8 (M = Ca, Sr, or Ba). Splitting of the spd valence orbitals of an atom
M with the configuration (n−1)d2(n)s0(n)p0 in the octacoordinate cubic (Oh) field of eight CO ligands is also given. Only the occupied valence orbitals
that are relevant for the M−CO interactions are shown. Up and down arrows indicate electrons with opposite spin. Reprinted with permission from ref
102. 2018 American Association for the Advancement of Science.

view quite demanding for chemists, we briefly outline the most Loosely speaking, the two large components ΨLα and ΨLß of
important qualitative view points of this field. the relativistic wave function ΨD refer to electrons with α and ß
The Schrödinger equation is usually presented in quantum spin, and the remaining two small components ΨSα and ΨSß are
chemistry in its time-independent form (eq 27). The kinetic required to allow for the existence of positrons (the antiparticles
energy requests the second derivatives with regard to the spatial of the electrons with the same mass and spin but opposite
coordinates (x,y,z). The time-dependent Schrödinger equation charge), which were not yet known in 1928 when Dirac
(eq 28) involves the derivative with respect to the time developed his theory and whose existence was predicted by
originating from the energy quantization, but the time him.355 Positrons were only identified in 1932.356 Thus, the
differential is only of first order. Relativistic theory demands Dirac equation directly yields the electron (and positron) spin,
that time and space are equivalent coordinates in the four whereas spin (and its resulting spin−orbit coupling) must be
dimensional space-time continuum, and thus, they must be introduced ad-hoc into the nonrelativistic Schrödinger equation.
treated equally in physics. The Schrödinger equation is therefore Calculations involving the Dirac equation with the full four-
not compatible with relativity, but it naturally emerges from the component wave function ΨD are significantly more compli-
nonrelativistic limit (velocity of light c → ∞). cated and demanding than those of the nonrelativistic
Schrödinger equation, mainly due to the fact of the appearing
[−h2 /2m(∂ 2/∂x 2 + ∂ 2/∂y 2 + ∂ 2/∂z 2) + V (x , y , z)]
negative energy continuum and increasing computational costs
Ψ(x , y , z) = E Ψ(x , y , z) (27) of the matrix form of the Dirac operator. Nevertheless, a number
of four-component methods have been developed over the past
[−ℏ2 /2m(∂ 2/∂x 2 + ∂ 2/∂y 2 + ∂ 2/∂z 2) + V (x , y , z , t )] 30 years, applicable routinely to smaller systems.357 For larger
systems, approximate two-component methods have been
Ψ(x , y , z , t ) developed, which involve the separation of the large from the
small component in the Dirac wave function and a trans-
= ih∂/∂t Ψ(x , y , z , t ) (28)
formation to a two-component formalism.358−360 Two-
The problem to bring the Schrödinger equation in harmony component methods such as the recently introduced X2C
with special relativity was solved by Dirac. He introduced a (4 × approximation are less computer time-consuming than four-
4) matrix operator, which brings the spatial and time coordinates component procedures but are still computationally expensive
on equal footing. Mathematical details on the Dirac operator D especially when electron correlation is included.346−354,361,362
shall not be presented in this work; they are discussed in several Therefore, the presently most commonly used quantum
review articles and monographs.346−354 The most important chemical approaches including relativistic effects in chemistry
result of the Dirac eq 29 is the manifestation of ΨD = are based on one-component (scalar relativistic) approxima-
(ΨLαΨLßΨSαΨSß) being a four-component wave function tions, such as the Douglas−Kroll method modified by Hess363 or
(spinor). the Zeroth Order Regular Approximation (ZORA).364 A very
popular time-efficient and quite accurate alternative is the use of
DΨD(x , y , z , t ) = E ΨD(x , y , z , t ) (29) relativistic effective core potentials (ECPs), also called
8830 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Table 24. EDA-NOCV Results for Triplet State M(CO)8 (M This is because valence electrons can have substantial density in
= Ca, Sr, Ba) Complexes at the M06-2X/TZ2P-ZORA//M06- the core region where relativistic perturbation operators
2X-D3/def2-TZVPP Level Taking (CO)8 in Singlet Ground act.375,376 This gives rise to three major effects of relativity on
State and M in Triplet Excited State with a (n)s0(n−1)d2 the valence electrons, which are important for the electronic
Valence Electronic Configuration as Interacting Fragmentsa structure and bonding situation in molecules:368 (a) The lowest
lying s orbitals (and often p orbitals but to a lesser extend)
Ca + Sr + Ba +
Energy term (CO)8 (CO)8 (CO)8 stabilize and contract; that is, their shell radii become smaller
(direct relativistic effect). (b) Because of this direct relativistic
ΔEint −243.1 −224.1 −145.2
ΔEhybridb 41.8 46.9 37.4
effect the nucleus becomes more screened lowering the effective
ΔEPauli 19.5 30.5 30.8
nuclear charge for each orbital, and most notably the higher
ΔEelstatc −65.3 −61.7 −78.5
angular momentum (more diffuse) functions expand and the
(21.5%) (20.5%) (36.8%) orbital energies increase. This effect is called the indirect
ΔEorbc −239.1 −239.7 −135.0 relativistic ef fect and is more important for the d and higher
(78.5%) (79.5%) (63.2%) angular momenta shells. (c) The third effect is the coupling of
ΔEorb(1)d,e (eg) [M(d)]→(CO)8 π −206.2 −206.4 −95.0 the electron spin with its angular momentum. This effect is
backdonation (86.2%) (86.2%) (69.8%)
called spin−orbit coupling (correctly derived only from the Dirac
ΔEorb(2)d,e (t2g) [M(d)]←(CO)8 σ −21.3 −20.7 −22.8
donation (9.0%) (8.7%) (16.8%) equation), which, for example, leads to a splitting of the p
ΔEorb(3)d (a1g) [M(s)]←(CO)8 σ −2.4 −2.9 −3.2 orbitals into a p1/2 and p3/2 component. We note that the
donation (1.0%) (1.2%) (2.4%) contraction of the p1/2 component of the orbital is much larger
ΔEorb(4)d,e (t1u) [M(p)]←(CO)8 σ −0.9 −0.9 −2.7 than for the p3/2 component. Figure 42 shows the variation of the
donation (0.3%) (0.3%) (2.1%) orbitals under the influence of relativistic effects.377
ΔEorb(5)d (a2u) (CO)8 polarization −0.6 −1.0 −2.3 In many-electron systems the underlying shell structure has
(0.3%) (0.4%) (1.7%)
ΔEorb(rest)d −7.7 −7.8 −9.0
significant influence on the magnitude of relativistic effects.378
(3.2%) (3.3%) (6.7%) For example, in neutral Tl the 6p1/2 orbital contracts (−10.4%)
ΔEprep(a) 8 CO → (CO)8 13.9 7.7 3.3 while the 6p3/2 orbital expands slightly (+2.2%). Spin−orbit
ΔEprep(b)f M, (n)s2 → (n) 159.5 150.9 68.2 splitting in the p-levels can be substantial; for example, for the
s0(n−1)d2 (T) heavy p-block row we get for the spin−orbit splitting (in kcal/
ΔE (ΔEint + M(CO)8 → M (1S) −69.7 −65.5 −73.7 mol) in the spectroscopic levels379 of Tl 22.3 (2P1/2/2P3/2), Pb
ΔEprep) = −De + 8 CO
a
30.4 (3P0/3P2), Bi+ 48.7 (3P0/3P2), Po 48.1 (3P2/3P1) and At
Energy values are given in kcal/mol. bContribution of the 60.3 (2P3/2/2P1/2).380 It becomes obvious that such large energy
metahybrid term in M06-2X. cThe values within the parentheses differences due to relativity become manifest also in chemical
show the contribution to the total attractive interactions ΔEelstat +
bonding. Interestingly, subtle shell structure effects378 can lead
ΔEorb. dThe values within the parentheses show the contribution to
the total orbital interaction, ΔEorb. eThe sum of the two (eg) or three to a much larger percentage contraction of the valence s-orbital
(t2g, t1u) components is given. fThe EDA calculations give a triplet compared to the core−shell s-orbitals. For example, for the
state with spherically symmetrical distribution of the d electrons. The neutral atom Tl we get the following s-contractions/orbital
experimental values for excitation into the energetically lowest lying stabilizations:381 1s −12.7/10.7%, 2s −13.4/17.4%, 3s −10.6/
3
F state with (n)s0(n−1)d2 configuration are 124.2 and 59.8 kcal/mol 18.1%, 4s −9.4/20.1%, 5s −9.5/21.7%, 6s −13.1/24.4%.
for Ca and Ba. There is no experimental value for the relevant 3F state Moreover, relativistic effects in the s-levels of the p-block
of Sr. The data are taken from ref 102. elements are particularly large from filling the lower lying core d-
shell.382 This rather strong relativistic valence s-shell contraction
pseudopotentials.365,366 The basic idea behind the pseudopo- leads to a large separation between the ns/np levels and to a
tential approximation goes back to Hellmann, who introduced reduced mixing (hybridization) between the two, thus altering
the concept already in 1933.367 Here, the chemically inactive chemical bonding in the p-block heavy-element section
core electrons of the atom are replaced by an effective potential, substantially. This comes on top of the reduced s/p mixing
which is optimized in relativistic atomic calculations mimicing from the first to the second octal-row atoms discussed above.
the influence of relativistic effects on the valence electrons. Thus, Interestingly, these perhaps unexpected large relativistic effects
ECP calculations do not only consider relativistic effects but also and their consequence for chemical and physical properties have
reduce the number of electrons in quantum chemical been accepted far earlier by the chemistry than by the physics
calculations. Small-core ECPs explicitely retain the outer community.
core−electrons in the calculations, whereas the less accurate Spin−orbit effects can be large in heavy elements but are
large-core ECPs consider only the immediate valence shell. The suppressed in strong ligand fields, which is best described by
relativistic ECP approximation transfers relativistic effects ionic bonding models. This is obvious as we formally move
extremely well from the atomic into the molecular environment; electron density from orbitals undergoing spin−orbit splitting at
thus, the change of the relativistic influence on the electronic the central atom to the electronegative ligand. For the p-block
structure during the interatomic interactions is accurately elements in high oxidation states with strong ionic bonding
described. partners (such as fluorine) spin−orbit effects become rather
For a very long time it was thought that relativistic effects are small. However, for covalent bonding spin−orbit effects can also
unimportant for chemistry, because the relativistic increase in be substantially reduced by mixing of p1/2 with p3/2 orbitals.
kinetic energy of the electrons in heavy atoms takes place mainly These relativistic orbitals are complex spin α/β mixtures; that is,
at the core electrons close to the nucleus. It was only in the 1970s two p1/2 combinations at different centers give 1/3σ and 2/3π*
when it was discovered that the valence electrons in heavier bonding or a 2/3π and 1/3σ* bonding, while two p3/2
atoms are also affected by relativity and to the extend where the combinations at different centers give one π or a 2/3π and
chemistry of heavy elements is changed significantly.354,368−374 1/3σ* bonding.383 Thus, s-p hybridization and the formation of
8831 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Figure 42. Radial densities D(r) = 4πr2ρ(r) (r in atomic units) for the 1s, 2s, 3s, and 2p states of a hydrogen-like atom with Z = 80. The dashed curves
are nonrelativistic (NR) and the full curves relativistic. The contractions for 1s, 2s, 2p1/2, and 3s are of the same order of magnitude while that for 2p3/2
is much smaller. Reproduced with permission from ref 377. Copyright 1967 Elsevier.

strong σ bonding can be severely hampered if we do not strong relativistic effects. For example, in the compound CsAu
correctly mix p1/2 with p3/2.383 This is especially the case when the gold atom acts like an electronegative ligand inducing ionic
spin−orbit splitting becomes very large, as for the superheavy p- bonding best described by Cs+Au−,395−397 which is due to the
block elements. For example, the 2P3/2/2P1/2 splitting in the relativistically increased electronegativity of Au (from 1.9 to
heaviest rare-gas element known, Og+, is 229 kcal/mol,384 larger 2.4).398 This leads to the interesting situation where solid
than most bond dissociation energies. Spin−orbit effects have Cs+Au− adopts a rocksalt structure rather than being an
major influence in systems, which are sensitive to small changes intermetallic compound as one would expect399 (two metals
in energy such as in (first or second-order) Jahn−Teller do not necessarily give a metallic bond!). CsAu can be dissolved
distortion385,386 or in dispersive types of interactions. in liquid ammonia like several other ionic alkali or alkaline earth
All the important bonding models for molecules or the solid halides,400 and related to this, the interesting coordination
state developed in nonrelativistic quantum chemistry can be compound AuCs·NH3 has been isolated by Jansen and co-
transferred and applied to the relativistic domain, even if spin− workers in 2002.401 A similar interesting example is the closed-
orbit coupling is considered. Furthermore, like electron- shell interaction between Ba and Au− with an unprecedented
correlation effects which can be switched on and off in high dissociation energy of 34 kcal/mol due to relativistic
calculations, relativistic effects are extremely useful in discussing effects.402
anomalies in chemical bonding situations and periodic Far more interesting for relativistic effects in chemical
trends.387,388 bonding are the p-block elements. The direct relativistic s-
Concerning relativistic effects in the group 1 and 2 elements of shell contraction, enhanced by filling the core d-shell, results in
the periodic table,389,390 which are dominated by the direct the valence (n)s AOs of the heavy atoms becoming less prone to
relativistic s-shell contraction/stabilization, these are rather chemical bonding; that is, the (n)s electrons retain as lone-pairs
small compared to the corresponding group 11 and 12 atoms rather than engaging in chemical bonding (inert-pair effect,
(except for the superheavy elements with nuclear charge 119 and originally proposed by Sidgwick403−405). This reduces s−p
120).391−393 This was already recognized early on by Pyykkö mixing in heavy main group compounds and, as a consequence,
and Desclaux.394 In these groups we expect for example small destabilizes high oxidation states.406 For example, the non-
bond contractions and bond stabilizations for the dimers, which relativistic 6s population for TlCl3 was calculated to be 0.73,
however will not significantly alter the bonding picture. The while the relativistic one gave 0.96.406 It is well-known that low
situation completely changes when the electropositive group 1 valencies in heavy p-block element compounds arise naturally
and 2 elements bind to atoms such as gold, which exhibit very from the periodic trend down the group in the periodic table
8832 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

because of smaller overlap between the more diffuse higher distance by 0.17 Å and reduce the bond energy by about 19 kcal/
principal quantum number orbitals; however, relativistic effects mol in Tl2.417 This feature of weak Tl−Tl bonding propagates
significantly enhance this trend. For example, the TlCl3 → TlCl into the solid state with Tl having a rather small cohesive energy
+ Cl2 decomposition energy reduces substantially by 42 kcal/ of 43 kcal/mol. Further, Tl−Tl interactions in coordination
mol (Table 25) due to relativistic effects to a final experimental compounds (termed thallophilic interactions) are rare and
extremely weak, and mostly counterion-mediated, requiring the
Table 25. Relativistic and Nonrelativistic Reaction Energies anions to provide favorable electrostatics,418 or due to crystal
(in kcal/mol) of TlX3 (X = F−I) for the Process TlX3(D3h) → packing effects.419,420 Similar large spin−orbit coupling effects
X2 + TlX at the QCI Level Using the Pseudopotential Method are observed for lead. For example for solid lead, nonrelativistic
(from Ref 407) PW91 results give a cohesive energy of 74 kcal/mol, scalar
relativistic effects reduces it slightly to 70 kcal/mol changing the
Reaction Nonrelativistic ΔE Relativistic ΔE
crystal structure from diamond to face center cubic, and finally
TlF3 → F2 + TlF 154.9 105.4 spin−orbit effects lead to a substantial reduction to reach a final
TlCl3 → Cl2 + TlCl 97.0 54.6 calculated value of 46 kcal/mol (exp. 47 kcal/mol) and a change
TlBr3 → Br2 + TlBr 82.0 45.2 in crystal structure to the hexagonal close packing arrange-
TlI3 → I2 + TlI 63.7 33.5 ment.421 Kraka et al. studied the energetics and mechanism for
the hydrogenation reaction XHn + H2 → XHn+1 + H with X
value of 50 kcal/mol.407 Solid-state effects shift the equilibrium belonging to group 4, 5, 6, or 7 of the periodic table, and found
even more toward the lower oxidation state for thallium; that is, that spin−orbit coupling becomes very important for the heavier
solid TlCl3 is rather unstable and disproportionates at 40 °C, elements.422,423
loosing chlorine. This originates from the large relativistic The chemistry of short-lived isotopes of superheavy elements
increase of the TlCl dipole moment by more than 1 D,407 can be studied using atom-at-a-time experiments.424 To prepare
leading to stronger ionic interactions in the solid. Concerning for such experiments, theoretical predictions about their
TlI3, it only exists as thallium(I)triiodide in solid form.408 chemical and physical behavior are required.425 Here, however,
Similarly large relativistic effects are obtained for the halides of relativistic effects in the bonding of the superheavy elements
lead.406 For example, Sn(II) and Pb(II) compounds are quite become so dominant that they drastically change their chemical
stable, whereas C(II) and Si(II) require electronic stabilization and physical behavior.391−393 The inert-pair effect becomes so
in order to be isolated. The influence of relativity on the stability strong for the higher oxidation state that even the fluorides NhF3
of group-14 compounds in the oxidation states E(II) and E(IV) and FlF4 decompose easily.426,427 For example, the 7s
becomes obvious by the calculated reaction energies for the population in NhF3 increases dramatically from 0.3 to 1.2 due
decomposition ECl4 → Cl2 + ECl2 (E = C − Pb) shown in Table to relativistic effects. In fact, NhF3 adopts a T-shape instead of
26. It is clear that there is already a sizable difference between the expected trigonal-planar structure. Because of strong spin−
orbit splitting in Fl (the 3P0/3P2 splitting is 83 kcal/mol428), Fl
Table 26. Relativistic and Nonrelativistic Free Reaction has a closed-shell (7p1/2)2 configuration, and p3/2 mixing into the
Energies (in kcal/mol) of ECl4 (E = C−Pb) for the Process Fl−Fl bond to form strong σ- and π- bonding becomes unlikely,
ECl4 → Cl2 + ECl2 at the BP86-D3(BJ)/TZ2P Level Using predicting only very weak van der Waals type of interactions.
the Scalar ZORA Approach Indeed, solid-state calculations gave a cohesive energy of only 12
Reaction Nonrelativistic ΔG298 Relativistic ΔG298 kcal/mol for Fl.421 Pitzer already suggested in 1975 that Fl
CCl4 → Cl2 + CCl2 60.9 60.8
should be chemically inert and could even be a gas at room
SiCl4 → Cl2 + SiCl2 98.0 97.0
temperature.429
GeCl4 → Cl2 + GeCl2 61.4 55.9
Moving to the elements where the spin−orbit destabilized
SnCl4 → Cl2 + SnCl2 65.2 49.4
7p3/2 level becomes occupied, B3LYP ZORA calculations by
PbCl4 → Cl2 + PbCl2 72.4 14.4
Mitin and Wüllen show a very large spin−orbit destabilization
for the dimer of tennessine (eka-At), Ts2, from 37 to 11 kcal/
mol due to the inaccessability of the 7p1/2 electrons to form
nonrelativistic and relativistic values for Ge, which becomes proper σ-bonds.430 This might be termed as a spin−orbit caused
much larger for Sn and particularly large for Pb. For example, 7p1/2 inert pair ef fect. The rare-gas element oganesson (eka-Rn)
while CCl4 is a stable compound, PbCl4 decomposes quickly at is even more spectacular. The extremely large spin−orbit
50 °C into PbCl2 and Cl2 due to relativistic effects (the gas phase splitting results in a rather diffuse and polarizable 7p3/2 shell and
decomposition energy of PbCl4 has been calculated to be 24 increases the dispersive type of interactions. Hence, Og2 has the
kcal/mol at the CCSD(T) level of theory).409 A nice peculiarity highest dissociation energy of any rare-gas dimer predicted at 1.8
concerns the lead acid battery, for which we have the overall kcal/mol.431 In fact, the electron localization function (ELF)
reaction Pb(s) + PbO2(s) + 2H2SO4(aq) → 2PbSO4(s) + exhibits a uniform electron gas-like behavior in the valence
2H2O(l). The relativistic 6s stabilization destabilizes PbO2, thus region not seen for any of the lighter elements shown in Figure
raising the voltage from 0.4 V to the standard value of 2.1 V.410 43.384 The ELF clearly shows the shell structure for the heavier
Thus, lead batteries only work because of relativistic effects. rare gases Xe and Rn, but for Og the density becomes smeared
To illustrate the role of spin−orbit effects, we choose two out over the whole atom. Moreover, Og has a positive electron
prime examples, the dimers Tl2 and Pb2.411−414 Tl2 has an affinity of 1.5 kcal/mol due to the very large relativistic 8s shell
unusually small dissociation energy of 0.43 ± 0.04 eV.415 contraction, while all the other rare-gas elements do not bind an
Christiansen and Pitzer stated in 1981 that “Tl2 is only weakly extra electron.432,433 One might therefore speculate that solid
bound... The cause is undoubtedly related to the large spin-orbit Og has a small band gap and becomes semiconducting or even
splitting between the 6p1/2 and 6p3/2 thallium spinors”.416 Han and metallic due to relativistic effects.434 Further, the diffuse four
Hirao calculated that spin−orbit effects increase the bond 7p3/2 electrons can easily be removed to form compounds like
8833 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Spin−orbit coupling becomes very important for the fifth


and higher octal rows.
• Quantum chemical methods are now available, which
consider relativistic effects with high accuracy. The
presently most common methods are the ZORA
approximation, Douglas−Kroll−Hess approach, X2C,
and relativistic pseudopotentials. Spin−orbit coupling
can be either included at the Hartree−Fock or Kohn−
Sham level or at a later stage at the correlated level (e.g.,
multireference configuration interaction).

11. CONCLUDING REMARKS


The approach and the relationship of many chemists to the
quantum theoretical explanation of chemical bonding may be
seen as example of Plato’s cave allegory, which is described in his
work Republica.444 It is written as a fictitious dialogue where
Socrates pictures a group of imprisoned people who spent all
their life chained in a cave and facing a blank wall. The people
watch moving shadows projected on the wall in front of them
from a fire behind them. Since these are the only processes
accessible to their senses, the shadows are seen as real objects.
They develop a science of shadows and try to determine laws in
their occurrence and movements and to derive prognoses from
them. They give praise and honor to the one who makes the best
predictions.
Figure 43. Electron localization functions from nonrelativistic (NR, One day a prisoner manages to break his chains and he
left) and Dirac−Hartree−Fock calculations (R, right) for the heavy rare escapes from the cave. He discovers the sun and realizes that
gas atoms Xe (top), Rn (middle), and Og (bottom). Reprinted with their reality was not what they thought it was. But the sun “would
permission from ref 384. Copyright 2018 American Physical Society. hurt his eyes, and he would escape by turning away to the things
which he was able to look at, and these he would believe to be clearer
than what was being shown to him.”444 Slowly his eyes are
OgF4. This structure, however, does not adopt the expected D4h adjusting to the light, and the freed prisoner understands that the
symmetry but becomes tetrahedral due to strong spin−orbit real world outside the cave is superior to the world he
coupling.435 This indicates that we expect rather unusual experienced in the cave. He is happy about the change, and he
bonding features in the superheavy element region due to wants to free his fellow cave dwellers out of the cave and into the
relativistic effects, which need yet to be explored. sunlight. But returning to the cave would again hurt his eyes,
As already mentioned for the Group 1 and 2 elements, strong which have become accustomed to the sunlight, and the
relativistic effects in molecular properties also appear when prisoners left behind would infer from the returning man’s pain
elements are bound to “relativistic” atoms such as platinum, gold that leaving the cave had harmed him and that they should not
or mercury. To mention a few examples for the p-block undertake a similar journey. Thus, anyone who attempted to
elements, the strong Pt−C-bond (∼110 kcal/mol) in [Pt- drag them out of the cave would therefore be attacked and
(CH2)]+ profits from a 50 kcal/mol relativistic stabilization with cursed.444−448 The resistance and even hostility of experimental
the two Pt configurations d8s1 and d9 being involved in the chemists toward quantum chemistry is a painful experience of
formation of the Pt−C double bond.436,437 That relativity can many theoretical chemists.
dictate enzymatic reaction was nicely demonstrated by This work is an attempt to connect the present understanding
Kozlowski et al., who showed that mercury methylation by of the nature of the chemical bond in terms of quantum theory
cobalt corrinoids is dictated by spin−orbit splitting of the Hg 6p with the heuristic bonding models, which were derived from
orbitals.438 In this respect, (CH3)2Hg, a highly volatile experimental observations. We do not consider the empirical
compound and a well-known toxin, shows very large relativistic models as irrelevant or unimportant shadows on the wall. On the
contrary, they are central ingredients and key variables in
effects in the Hg−C bond strength.439,440 For many further
chemical research, which serve as indispensable guidelines for
examples on relativistic effects including main-group elements,
further development. However, the models must not be
we refer to a number of review articles.368,370−374,394,441−443
identified with the physical reality of chemical bonding and
The following conclusions arise from this section:
interatomic interactions. They should be used on the basis of
• Relativistic effects have a strong influence on the and in conjunction with accurate quantum chemical calcu-
structures and reactivities of heavy main-group atoms. lations. The sometimes ambiguous interpretation of exper-
They become already important for atoms of the fourth imental results can be better focused in favor of more targeted
octal row of the periodic system, and they have a planning of further experiments. The laws of physics are only the
dominant effect on atoms of the fifth and higher rows. framework within which chemical research, with the help of
• Relativistic effects make the s-orbitals less inclined to form human creativity and fuzzy concepts, always creates something
chemical bonds. This leads, among other things, to the new, often with different viewpoints about the interpretation of
stabilization of compounds in low oxidation states and chemical bonding. This is not a weakness of chemistry but shows
enhances the inert pair effect for the p-block elements. the limitation of the human mind, which has been expressed by
8834 DOI: 10.1021/acs.chemrev.8b00722
Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Niels Bohr with the saying “We must be clear that when it comes to Foundation to join Auckland University and the Australian National
atoms, language can be used only as in poetry. The poet, too, is not University. He earned many awards, amongst them the Prince &
nearly so concerned with describing facts as with creating images and Princess of Wales Science Award, the Hector medal, Fellowships of the
establishing mental connections.”113 Royal Society New Zealand and the International Academy of
Quantum Molecular Science, a James Cook Fellowship and most
AUTHOR INFORMATION recently a Humboldt Research Prize, the Fukui Medal and the
Corresponding Authors Rutherford medal. His research interests are in fundamental chemistry
and physics.
*E-mail: peter.schwerdtfeger@gmail.com.
*E-mail: frenking@chemie.uni-marburg.de. Gernot Frenking, FRSC, studied chemistry at the universities Aachen,
Kyoto, and TU Berlin where he received his doctoral degree in 1979.
ORCID
After obtaining his habilitation in theoretical organic chemistry at the
Lili Zhao: 0000-0003-2580-6919 TU Berlin in 1984, he moved to the USA. Following one year as a
Sudip Pan: 0000-0003-3172-926X visiting scientist with Fritz Schaefer at UC Berkeley he worked at the
Gernot Frenking: 0000-0003-1689-1197 Stanford Research Institute in Menlo Park, California. In 1989 he
returned to Germany as Associate Professor for Computational
Notes
Chemistry at the Philipps University Marburg, where he was appointed
The authors declare no competing financial interest. Full Professor for Theoretical Chemistry 1998 and became Hans-
Biographies Hellmann-Professor from 2011−2014. He is also Visiting Research
Professor at the DIPC (Donostia International Physics Center) at San
Lili Zhao received her Ph.D. degree at the Graduate University of Sebastian, Spain. Awards include the Elhuyar-Goldschmidt Prize of the
Chinese Academy of Sciences in 2012. She then worked in the IHPC, Royal Society of Chemistry of Spain (2007) and the Schrödinger Medal
A*STAR in Singapore as a Scientist during 2012−2014, modeling of the WATOC (2009). In 2019 he received the International Solvay
homogeneous catalysis. From 2014 to 2016, she held a Humboldt Chair of Chemistry of the International Solvay Institutes. His research
research fellowship at Philipps-Universität Marburg in Germany. She interests lie in the field of Chemical Bonding Theory and Theoretical
joined Nanjing Tech University (China) in the end of 2016 and worked Inorganic Chemistry. Major topics of his work are the nature of the
independently as a full professor. Her research interest include the chemical bond and molecules with unusual electronic structures.
homogeneous catalysis, bonding analysis by using state-of-the-art
methods (i.e., EDA, EDA-NOCV, QTAIM), as well as carbone
chemistry.
ACKNOWLEDGMENTS
Sudip Pan obtained his Ph.D. degree from the Indian Institute of
G.F. wants to express his gratitude to Dr. Alfred Paulus for
Technology Kharagpur (India) in 2016 under the supervision of Prof.
inspiring suggestions and Prof. Eugen Schwarz as well as Klaus
Pratim K. Chattaraj. In the same year, he moved to work as a
Ruedenberg for enlightening discussions about the nature of the
Postdoctoral Fellow at Cinvestav Merida (Mexico) under Prof. Gabriel
chemical bond. He also expresses his gratitude for financial
Merino. Toward the end of 2017, he moved to Nanjing Tech University
support by the Alexander von Humboldt foundation and the
(China) for another postdoctoral stay under Prof. Gernot Frenking and
Deutsche Forschungsgemeinschaft. L.Z and G.F. acknowledge
Prof. Lili Zhao, where he is currently working. His research interests
financial support from Nanjing Tech University (grant numbers
include the theoretical predictions of viable noble gas compounds,
39837123 and 39837123) and a SICAM Fellowship from
effective hydrogen storage material, clusters with unusual bonding,
Jiangsu National Synergetic Innovation Center for Advanced
nanomachinery, compounds with boron−boron triple bonds and
Materials, National Natural Science Foundation of China (grant
reactivity, catalysis and reaction mechanism, planar hypercoordinate
no. 21703099), and Natural Science Foundation of Jiangsu
carbon and boron systems, and ligand stabilized species and reactivity.
Province for Youth (grant no. BK20170964). S.P. thanks
Nanjing Tech University for a postdoctoral fellowship and the
Nicole Holzmann studied chemistry at the universities Darmstadt, High Performance Computing Center of Nanjing Tech
Bristol/UK, and Marburg. In 2013 she received her doctoral degree in University for providing computational resources. P.S. thanks
Theorerical Chemistry at the Philipps-Universität Marburg in the group the Royal Society of New Zealand for financial support in terms
of Prof. Dr. Gernot Frenking, where she was working on bonding of a Marsden Fund (17-MAU-021) and Prof. Helmut Schwarz
analyses of donor−acceptor complexes. From 2014 to 2015 she was a for useful discussions.
postdoc with the CNRS (Centre National de la Recherche Scientifique)
at the University of Lorraine with Prof. Chris Chipot and Dr. François REFERENCES
Dehez. During this time she carried out MD simulations of the
membrane proteins. In her current position in the group of Dr. Gilberto (1) Frenking, G.; Fröhlich, N. The Nature of the Bonding in
Transition-Metal Compounds. Chem. Rev. 2000, 100, 717−774.
Teobaldi at the STFC (Science and Technology Facilities Council), she
(2) Lewis, G. N. The Atom and the Molecule. J. Am. Chem. Soc. 1916,
works in close collaboration with user groups and instrument scientists 38, 762−785.
at the Rutherford Appleton Laboratory facilities and in the area of ab (3) Langmuir, I. The Arrangements of Electrons in Atoms and
inito molecular dynamics simulations and TD-DFT. Molecules. J. Am. Chem. Soc. 1919, 41, 868−934.
Peter Schwerdtfeger currently holds a chair in Theoretical Chemistry at (4) Langmuir, I. Isomorphism, Isosterism and Covalence. J. Am. Chem.
Massey University and serves as Director of the Center of Theoretical Soc. 1919, 41 (1919), 1543−1559.
(5) Langmuir, I. The Octet Theory of Valence and its Applications
Chemistry and Physics within the Institute for Advanced Study. He
with Special Reference to Organic Nitrogen Compounds. J. Am. Chem.
took his first degree in Chemical Engineering (Aalen) and studied both Soc. 1920, 42, 274−292.
chemistry and mathematics at Stuttgart University, where he received (6) Langmuir, I. Types of Valence. Science 1921, 54, 59−67.
his Ph.D. in theoretical chemistry, at habilitated at Marburg University. (7) The particular stability of molecules where atoms have eight
He held a position as a software analyst at Stuttgart University before electrons in their valence shell was already pointed out by Lewis, who
receiving a Feodor-Lynen fellowship of the Alexander von Humboldt wrote: “The atom tends to hold an even number of electrons in the

8835 DOI: 10.1021/acs.chemrev.8b00722


Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

shell, and especially to hold eight electrons which are normally arranged (34) Bitter, T.; Wang, S. G.; Ruedenberg, K.; Schwarz, W. H. E.
symmetrically at the eight comers of a cube”.2 Toward a Physical Understanding of Electron-Sharing Two-Center
(8) Heitler, W.; London, F. Wechselwirkung Neutraler Atome und Bonds. II. Pseudo-Potential Based Analysis of Diatomic Molecules.
Homöopolare Bindung nach der Quantenmechanik. Eur. Phys. J. A Theor. Chem. Acc. 2010, 127, 237−257.
1927, 44, 455−472. (35) Bacskay, G. B.; Nordholm, S.; Ruedenberg, K. The Virial
(9) Heisenberg, W. Ü ber Quantentheoretische Umdeutung Kine- Theorem and Covalent Bonding. J. Phys. Chem. A 2018, 122, 7880−
matischer und Mechanischer Beziehungen. Eur. Phys. J. A 1925, 33, 7893.
879−893. (36) For an illustrative demonstration see: Rioux, F. Vibrational
(10) Schrödinger, E. Quantisierung als Eigenwertproblem. Ann. Phys. Analysis for C60 and Other Fullerenes. J. Chem. Educ. 2003, 80, 1380.
1926, 79, 361−376. (37) Mierzecki, R. The Historical Development of Chemical Concepts;
(11) Pauling, L. The Nature of the Chemical Bond and the Structure of Kluwer Academic Publishers: Dordrecht, 1991.
Molecules and Crystals; Cornell University Press: Ithaca, NY, 1939. (38) Couper, A. S. On a New Chemical Theory. Philos. Mag., 4th
(The latest issue was published in 1960.) series, 1858, 16, 104−116.
(12) The universe is composed of only 4.6% of hadrons (protons, (39) Lewis, G. N. Valence and the Structure of Atoms and Molecules;
neutrons), which make only 4.6% of the total mass of the universe; the American Chemical Society Monograph Series; New York, 1923. The
large remainder comes from dark matter (24%) and dark energy book is out of print but an online version is found at: http://babel.
(71.4%). hathitrust.org/cgi/pt?id=uc1.$b35072;view=1up;seq=1.
(13) Frenking, G.; Krapp, A. Unicorns in the World of Chemical (40) Kohler, R. The Lewis-Langmuir Theory of Valence and the
Bonding Models. J. Comput. Chem. 2007, 28, 15−24. Chemical Community, 1920−1928. Hist. Stud. Phys. Sci. 1976, 6, 431−
(14) Kutzelnigg, W. The Physical Mechanism, of the Chemical Bond. 468.
Angew. Chem., Int. Ed. Engl. 1973, 12, 546−562. (41) Lewis, G. N. The Chemical Bond. J. Chem. Phys. 1933, 1, 17−28.
(15) Schmidt, M. W.; Ivanic, J.; Ruedenberg, K. The Physical Origin of (42) For a quantum theoretical foundation of the Lewis electron-pair
the Chemical Bond; In The Chemical Bond. 1. Fundamental Aspects of model see: Zhao, L.; Schwarz, W. H. E.; Frenking, G. The Lewis
Chemical Bonding; G. Frenking, G., Shaik, S., Eds.; Wiley-VCH: electron-pair bonding model: The physical background, a century later.
Weinheim, 2014; p 1−67. Nature Rev. Chem. 2019, 3, 35−47.
(16) Maksic, Z. B., Ed. Theoretical Model of Chemical Bonding. Part 2: (43) Lewis, G. N. Acids and bases. J. Franklin Inst. 1938, 226, 293−
The Concept of the Chemical Bond; Springer Verlag: Berlin, 1990. 313.
(17) Mulliken, R. S. Electronic Structures of Polyatomic Molecules (44) Sidgwick, N. V. Co-ordination Compounds and the Bohr Atom.
and Valence. II. General Considerations. Phys. Rev. 1932, 41, 49−71. J. Chem. Soc., Trans. 1923, 123, 725−730.
(18) Hund, F. Zur Frage der Chemischen Bindung. Eur. Phys. J. A (45) Sidgwick, N.V. The Electronic Theory of Valency; Clarendon Press:
1932, 73, 1−30. Oxford,1927.
(19) Lennard-Jones, J. E. The Electronic Structure of Some Diatomic (46) Sidgwick, N. V. Structure of Divalent Carbon Compounds.
Molecules. Trans. Faraday Soc. 1929, 25, 668−686. Chem. Rev. 1931, 9, 77−88.
(20) Hückel, E. Quantentheoretische Beiträge zum Benzolproblem. (47) Frenking, G.; Loschen, C.; Krapp, A.; Fau, S.; Strauss, S. H.
Eur. Phys. J. A 1931, 70, 204−286. Electronic Structure of CO - An Exercise in Modern Chemical Bonding
(21) Fukui, K. Theory of Orientation and Stereoselection; Springer Theory. J. Comput. Chem. 2007, 28, 117−126.
Verlag: Berlin, 1975. (48) Haaland, A. Covalent versus Dative Bonds to Main Group
(22) Woodward, R. B.; Hoffmann, R. The Conservation of Orbital Metals, a Useful Distinction. Angew. Chem., Int. Ed. Engl. 1989, 28,
Symmetry. Angew. Chem., Int. Ed. Engl. 1969, 8, 781−853. 992−1007.
(23) For a detailed discussion see refs 14 and 15. (49) Zhao, L.; Hermann, M.; Holzmann, N.; Frenking, G. Dative
(24) Huber, K. P.; Herzberg, G. Molecular Spectra and Molecular Bonding in Main Group Compounds. Coord. Chem. Rev. 2017, 344,
Structure IV. Constants of Diatomic Molecules; Springer US: New York, 163−204.
1979. (50) Tonner, R.; Ö xler, F.; Neumüller, B.; Petz, W.; Frenking, G.
(25) Gillespie, R. J., Hargittai, I. The VSEPR Model of Molecular Carbodiphosphoranes: The Chemistry of Divalent Carbon(0). Angew.
Geometry; Allyn & Bacon Boston, 1991. Chem., Int. Ed. 2006, 45, 8038−8042.
(26) For a critical discussion see: (b) Frenking, G. Book Review: (51) Tonner, R.; Frenking, G. C(NHC)2: Divalent Carbon(0)
Chemical Bonding and Molecular Geometry from Lewis to Electron Compounds with N-Heterocyclic Carbene LigandsTheoretical
Densities. Angew. Chem., Int. Ed. 2003, 42, 143−147. Evidence for a Class of Molecules with Promising Chemical Properties.
(27) Reply: Gillespie, R. J.; Popelier, P. L. A. Chemical Bonding and Angew. Chem., Int. Ed. 2007, 46, 8695−8698.
Molecular Geometry: Comments on a Book Review. Angew. Chem., Int. (52) Tonner, R.; Frenking, G. Divalent Carbon(0) Chemistry, Part 1:
Ed. 2003, 42, 3331−3334. Parent Compounds. Chem. - Eur. J. 2008, 14, 3260−3272.
(28) For a detailed discussion see: Krapp, A.; Bickelhaupt, F. M.; (53) Tonner, R.; Frenking, G. Divalent Carbon(0) Chemistry, Part 2:
Frenking, G. Orbital Overlap and Chemical Bonding. Chem. - Eur. J. Protonation and Complexes with Main Group and Transition Metal
2006, 12, 9196−9216. Lewis Acids. Chem. - Eur. J. 2008, 14, 3273−3289.
(29) As a necessary condition the wave function in quantum (54) Frenking, G.; Tonner, R. Divalent Carbon(0) Compounds. Pure
mechanics must be complex in order to fulfill the continuity equation. Appl. Chem. 2009, 81, 597−614.
We loosely adopt the word “square of the wave function” when we (55) Frenking, G.; Tonner, R.; Klein, S.; Takagi, N.; Shimizu, T.;
mean absolute value square. Krapp, A.; Pandey, K. K.; Parameswaran, P. New Bonding Modes of
(30) Hellmann, H. Einführung in die Quantenchemie (Introduction to Carbon and Heavier Group 14 Atoms Si − Pb. Chem. Soc. Rev. 2014, 43,
Quantum Chemistry); Deuticke: Leipzig and Wien, 1937. 5106−5139.
(31) Ruedenberg, K. The Physical Nature of the Chemical Bond. Rev. (56) Frenking, G.; Hermann, M.; Andrada, D. M.; Holzmann, N.
Mod. Phys. 1962, 34, 326−376. Donor−Acceptor Bonding in Novel Low-Coordinated Compounds of
(32) Schmidt, M. W.; Ivanic, J.; Ruedenberg, K. The Physical Origin of Boron and Group-14 Atoms C − Sn. Chem. Soc. Rev. 2016, 45, 1129−
Covalent Bonding. In The Chemical Bond: Fundamental Aspects of 1144.
Chemical Bonding; Wiley-VCH Verlag GmbH & Co. KGaA, 2014; p 1− (57) Himmel, D.; Krossing, I.; Schnepf, A. Dative Bonds in Main-
68. Group Compounds: A Case for Fewer Arrows! Angew. Chem., Int. Ed.
(33) Bitter, T.; Ruedenberg, K.; Schwarz, W. H. E. Toward a Physical 2014, 53, 370−374.
Understanding of Electron-Sharing Two-Center Bonds. I. General (58) Frenking, G. Dative Bonds in Main-Group Compounds: A Case
Aspects. J. Comput. Chem. 2007, 28, 411−422. for More Arrows! Angew. Chem., Int. Ed. 2014, 53, 6040−6046.

8836 DOI: 10.1021/acs.chemrev.8b00722


Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

(59) Himmel, D.; Krossing, I.; Schnepf, A. Dative or Not Dative? (84) Gritsenko, O. V.; Mentel, Ł. M.; Baerends, E. J. On the Errors of
Angew. Chem., Int. Ed. 2014, 53, 6047−6048. Local Density (LDA) and Generalized Gradient (GGA) Approx-
(60) Nesterov, V.; Reiter, D.; Bag, P.; Frisch, P.; Holzner, R.; Porzelt, imations to the Kohn-Sham Potential and Orbital Energies. J. Chem.
A.; Inoue, S. NHCs in Main Group Chemistry. Chem. Rev. 2018, 118, Phys. 2016, 144, 204114.
9678−9842. (85) Grafenstein, J.; Cremer, D. The combination of Density
(61) Patel, N.; Sood, R.; Bharatam, P. V. NL2+ Systems as New- Functional Theory with Multi-Configuration Methods − CAS-DFT.
Generation Phase-Transfer Catalysts. Chem. Rev. 2018, 118, 8770− Chem. Phys. Lett. 2000, 316, 569−577.
8785. (86) Grafenstein, J.; Cremer, D. Can Density Functional Theory
(62) Hermann, M.; Frenking, G. Gilbert Lewis and the Model of Describe Multi-Reference Systems? Investigation of Carbenes and
Dative Bonding. Struct. Bonding (Berlin, Ger.) 2016, 169, 131−156. Organic Biradicals. Phys. Chem. Chem. Phys. 2000, 2, 2091−2103.
(63) An excellent presentation of orbital correlation diagrams and (87) Gagliardi, L.; Truhlar, D. G.; Manni, G. L.; Carlson, R. K.; Hoyer,
molecular structures for a variety of main-group compounds and C. E.; Bao, J. L. Multiconfiguration Pair-Density Functional Theory: A
transition metal complexes is given by: Albright, T. A.; Burdett, J. K.; New Way To Treat Strongly Correlated Systems. Acc. Chem. Res. 2017,
Whangbo, M.-H. Orbital Interactions in Chemistry, 2nd ed.; Wiley: New 50, 66−73.
York, 2013. (88) Laqua, H.; Kussmann, J.; Ochsenfeld, C. Communication:
(64) Textbooks on Quantum Chemistry: Szabo, A.; Ostlund, N. Density Functional Theory Model for Multi-Reference Systems Based
Modern Quantum Chemistry; MacMillan: New York, 1982. on the Exact-Exchange Hole Normalization. J. Chem. Phys. 2018, 148,
(65) Levine, I. N. Quantum Chemistry, 7th ed.; Pearson: New York, 121101.
2016. (89) The historical development of the VB method has been discussed
(66) Helgaker, T.; Jorgensen, P.; Olsen, J. Molecular Electronic- in: Löwdin, P. O. On the Historical Development of the Valence Bond
Structure Theory; Wiley: New York, 2000. Method and the Non-Orthogonality Problem. J. Mol. Struct.:
(67) Koch, W.; Holthausen, M. C. A Chemist’s Guide to Density THEOCHEM 1991, 229, 1−14.
Functional Theory, 2nd ed.; Wiley-VCH: Weinheim, 2001. (90) Shaik, S., Hiberty, P. C. A Chemist’s Guide to Valence Bond Theory;
(68) Landis, C. R.; Weinhold, F. The NBO View of Chemical John Wiley & Sons, Inc., 2007.
Bonding. In The Chemical Bond: Fundamental Aspects of Chemical (91) Cooper, D. Valence Bond Theory; Elsevier: Amsterdam, 2002.
Bonding; Frenking, G., Shaik, S., Eds.; Wiley-VCH: Weinheim, 2014; p (92) In a recent review about VB theory, S. Shaik called the 1916
91−120. publication by Gilbert Lewis the first birth of Valence Bond Theory.
(69) Weinhold, F.; Landis, C. R. Discovering Chemistry With Natural This is a misleading identification of a bonding model with a quantum
Bond Orbitals; Wiley: NJ, 2012. chemical method that can lead to serious misinterpretation of the
(70) Landis, C. R.; Weinhold, F. Valency and Bonding: A Natural Bond nature of the chemical bond: Shaik, S. A Personal Story on a
Orbital Donor-Acceptor Perspective; Cambridge University Press: Renaissance in Valence Bond Theory: A Theory Coming of Age!
Cambridge, 2005. Comput. Theor. Chem. 2017, 1116, 2−31.
(71) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Intermolecular (93) Glendening, E. D.; Landis, C. R.; Weinhold, F. NBO 6.0: Natural
Interactions from a Natural Bond Orbital, Donor-Acceptor Viewpoint. Bond Orbital Analysis Program. J. Comput. Chem. 2013, 34, 1429−
Chem. Rev. 1988, 88, 899−926. 1437.
(72) Streitwieser, A., Jr. Molecular Orbital Theory for Organic Chemists; (94) Reed, A. E.; Schleyer, P.v.R. Chemical Bonding in Hypervalent
Wiley: New York, 1961. Molecules. The Dominance of Ionic Bonding and Negative Hyper-
(73) Dewar, M. J. S. Molecular Orbital Theory for Organic Chemistry; conjugation over d-orbital Participation. J. Am. Chem. Soc. 1990, 112,
McGraw-Hill: New York, 1969. 1434−1445.
(74) Borden, W. T. Modern Molecular Orbital Theory for Organic (95) Landis, C. R.; Firman, T. K.; Root, D. M.; Cleveland, T. A
Chemists; Prentice-Hall: Englewood Cliffs, NJ, 1975. Valence Bond Perspective on the Molecular Shapes of Simple Metal
(75) Hohenberg, P.; Kohn, W. Inhomogeneous Electron Gas. Phys. Alkyls and Hydrides. J. Am. Chem. Soc. 1998, 120, 1842−1854.
Rev. 1964, 136, B864−B871. (96) Landis, C. R.; Cleveland, T.; Firman, T. K. Valence Bond
(76) Kohn, W.; Sham, L. J. Self-Consistent Equations Including Concepts Applied to the Molecular Mechanics Description of
Exchange and Correlation Effects. Phys. Rev. 1965, 140, A1133−A1138. Molecular Shapes. 3. Applications to Transition Metal Alkyls and
(77) This is discussed in detail in: Bickelhaupt, F. M.; Baerends, E. J. Hydrides. J. Am. Chem. Soc. 1998, 120, 2641−2649.
Kohn-Sham Density Functional Theory: Predicting and Understanding (97) Firman, T. K.; Landis, C. R. Structure and Electron Counting in
Chemistry. Rev. Comput. Chem. 2007, 15, 1−86. Ternary Transition Metal Hydrides. J. Am. Chem. Soc. 1998, 120,
(78) Koch, W.; Holthausen, M. C. A Chemist’s Guide to Density 12650−12656.
Functional Theory, 2nd ed.; Wiley-VCH: Weinheim, 2001. (98) Maseras, F.; Morokuma, K. Application of the Natural
(79) Mardirossian, N.; Head-Gordon, M. Thirty Years Of Density Population Analysis to Transition-Metal Complexes. Should the
Functional Theory in Computational Chemistry: An Overview and Empty Metal p Orbitals be Included in the Valence Space? Chem.
Extensive Assessment Of 200 Density Functional. Mol. Phys. 2017, 115, Phys. Lett. 1992, 195, 500−504.
2315−2372. (99) Bayse, C. A.; Hall, M. B. Prediction of the Geometries of Simple
(80) Goerigk, L.; Hansen, A.; Bauer, C.; Ehrlich, S.; Najibi, A.; Transition Metal Polyhydride Complexes by Symmetry Analysis. J. Am.
Grimme, S. A Look At The Density Functional Theory Zoo with the Chem. Soc. 1999, 121, 1348−1358.
Advanced GMTKN55 Database for General Main Group Thermo- (100) Diefenbach, A.; Bickelhaupt, F. M.; Frenking, G. The Nature of
chemistry, Kinetics and Noncovalent Interactions. Phys. Chem. Chem. the Transition Metal−Carbonyl Bond and the Question about the
Phys. 2017, 19, 32184−32215. Valence Orbitals of Transition Metals. A Bond-Energy Decomposition
(81) Cremer, D. Density Functional Theory: Coverage of Dynamic Analysis of TM(CO)6q (TMq = Hf2‑, Ta−, W, Re+, Os2+, Ir3+). J. Am.
and Non-Dynamic Electron Correlation Effects. Mol. Phys. 2001, 99, Chem. Soc. 2000, 122, 6449−6458.
1899−1940. (101) Chi, C.; Pan, S.; Meng, L.; Luo, M.; Zhao, L.; Zhou, M.;
(82) Chong, D. P.; Gritsenko, O. V.; Baerends, E. J. Interpretation of Frenking, G. Alkali Metal Covalent Bonding in Nickel Carbonyl
the Kohn−Sham Orbital Energies as Approximate Vertical Ionization Complexes ENi(CO)3−. Angew. Chem. 2019, 131, 1746−1752. Angew.
Potentials. J. Chem. Phys. 2002, 116, 1760−1772. Chem., Int. Ed. 2019, 58, 1732−1738.
(83) Gritsenko, O. V.; Baerends, E. J. The Spin-Unrestricted (102) Wu, X.; Zhao, L.; Jin, J.; Pan, S.; Li, W.; Jin, X.; Wang, G.; Zhou,
Molecular Kohn−Sham Solution and the Analogue of Koopmans’s M.; Frenking, G. Observation of Alkaline Earth Complexes M(CO)8
Theorem for Open-Shell Molecules. J. Chem. Phys. 2004, 120, 8364− (M = Ca, Sr, Ba) that Mimic Transition Metals. Science 2018, 361, 912−
8372. 916.

8837 DOI: 10.1021/acs.chemrev.8b00722


Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

(103) Reed, A. E.; Weinhold, F. On the Role of d Orbitals in Sulfur (129) Zhao, L.; von Hopffgarten, M.; Andrada, D. M.; Frenking, G.
Hexafluoride. J. Am. Chem. Soc. 1986, 108, 3586−3593. Energy Decomposition Analysis. WIREs Comput. Mol. Sci. 2018, 8,
(104) Glendening, E. D.; Weinhold, F. Natural Resonance Theory: I. No. e1345.
General formalism. J. Comput. Chem. 1998, 19, 593−609. (130) Frenking, G.; Bickelhaupt, F. M. The EDA Perspective of
(105) Glendening, E. D.; Weinhold, F. Natural Resonance Theory: II. Chemical Bonding. In The Chemical Bond: Fundamental Aspects of
Natural Bond Order and Valency. J. Comput. Chem. 1998, 19, 610−627. Chemical Bonding; Frenking, G., Shaik, S., Eds.; Wiley-VCH:
(106) Glendening, E. D.; Badenhoop, J. K.; Weinhold, F. Natural Weinheim, 2014; p 121−158.
Resonance Theory: III. Chemical Applications. J. Comput. Chem. 1998, (131) Bickelhaupt, F. M.; Baerends, E. J. Kohn-Sham Density
19, 628−646. Functional Theory: Predicting and Understanding Chemistry. In Rev.
(107) See the reviewer comments that were published along with the Comput. Chem.; Lipkowitz, K. B., Boyd, D. B., Eds.; Wiley-VCH: New
original work in ref 104. York, 2000; Vol. 15, pp 1−86.
(108) Weinhold, F.; Klein, R. A. Anti-Electrostatic Hydrogen Bonds. (132) te Velde, G.; Bickelhaupt, F. M.; Baerends, E. J.; Guerra, C. F.;
Angew. Chem., Int. Ed. 2014, 53, 11214−11217. Van Gisbergen, S. J. A.; Snijders, J. G.; Ziegler, T. Chemistry with ADF.
(109) Frenking, G.; Caramori, G. F. No Need for a Re-examination of J. Comput. Chem. 2001, 22, 931−967.
the Electrostatic Notation of the Hydrogen Bonding: A Comment. (133) Yang, T.; Andrada, D. M.; Frenking, G. Dative versus Electron-
Angew. Chem., Int. Ed. 2015, 54, 2596−2599.
Sharing Bonding in N-oxides and Phosphane Oxides R3EO and
(110) Weinhold, F.; Klein, R. A. Improved General Understanding of
Relative Energies of the R2EOR Isomers (E = N, P; R = H, F, Cl, Me,
the Hydrogen-Bonding Phenomena: A Reply. Angew. Chem., Int. Ed.
Ph). A Theoretical Study. Phys. Chem. Chem. Phys. 2018, 20, 11856−
2015, 54, 2600−2602.
(111) Zubarev, D. Y.; Boldyrev, A. I. Developing Paradigms of 11866.
Chemical Bonding: Adaptive Natural Density Oartitioning. Phys. Chem. (134) Zhang, Q.; Li, W.-L.; Xu, C.-Q.; Chen, M.; Zhou, M.; Li, J.;
Chem. Phys. 2008, 10, 5207−5217. Andrada, D. M.; Frenking, G. Formation and Characterization of the
(112) Pendás, A. M.; Francisco, E. From Quantum Fragments to Boron Dicarbonyl Complex [B(CO)2]−. Angew. Chem., Int. Ed. 2015,
Lewis Structures: Electron Counting in Position Space. Phys. Chem. 54, 11078−11083.
Chem. Phys. 2018, 20, 21368−21380. (135) Andrada, D. M.; Frenking, G. Stabilization of Heterodiatomic
(113) Taken from https://en.wikiquote.org/wiki/Niels_Bohr. SiC through Ligand Donation: Theoretical Investigation of SiC(L)2 (L
(114) Bader, R. F. W. Atoms in Molecules. A Quantum Theory; Oxford = NHCMe, CAACMe, PMe3). Angew. Chem., Int. Ed. 2015, 54, 12319−
University Press: Oxford, 1990. 12324.
(115) Matta, C. F., Boyd, R. J., Eds. The Quantum Theory of Atoms in (136) Mohapatra, C.; Kundu, S.; Paesch, A. N.; Herbst-Irmer, R.;
Molecules; Wiley-VCH: Weinheim, 2007. Stalke, D.; Andrada, D. M.; Frenking, G.; Roesky, H. W. The Structure
(116) Bader, R. F. W. Bond Paths Are Not Chemical Bonds. J. Phys. of the Carbene Stabilized Si2H2 May Be Equally Well Described with
Chem. A 2009, 113, 10391−10396. Coordinate Bonds as with Classical Double Bonds. J. Am. Chem. Soc.
(117) Bonyhady, S. J.; Collis, D.; Holzmann, N.; Edwards, A. J.; Piltz, 2016, 138, 10429−10432.
R. O.; Frenking, G.; Stasch, A.; Jones, C. Anion Stabilised Hypercloso- (137) Li, Z.; Chen, X.; Andrada, D. M.; Frenking, G.; Benkö, Z.; Li, Y.;
Hexaalane Al6H6. Nat. Commun. 2018, 9, 3079. Harmer, J. R.; Su, C.-Y.; Grützmacher, H. (L)2C2P2: Dicarbondiphos-
(118) Mousavi, M.; Frenking, G. Bonding Analysis of Trimethylene- phide Stabilized by N-Heterocyclic Carbenes or Cyclic Diamido
methane (TMM) Complexes [(CO)3M−TMM] (M = Fe, Ru, Os, Carbenes. Angew. Chem., Int. Ed. 2017, 56, 5744−5749.
Rh+). Absence of Expected Bond Paths. J. Organomet. Chem. 2013, 748, (138) Scharf, L. T.; Andrada, D. M.; Frenking, G.; Gessner, V. H. The
2−7. Bonding Situation in Metalated Ylides. Chem. - Eur. J. 2017, 23, 4422−
(119) Kraka, E.; Cremer, D. Theoretical Models of Chemical Bonding. 4434.
The Concept of the Chemical Bond, Vol. 2; Maksic, Z. B., Ed.; Springer (139) Hermann, M.; Frenking, G. Carbones as Ligands in Novel
Verlag: Heidelberg, 1990; p 453. Main-Group Compounds E[C(NHC)2]2 (E = Be, B+, C2+, N3+, Mg,
(120) Cremer, D.; Kraka, E. Chemical Bonds without Bonding Al+, Si2+, P3+): A Theoretical Study. Chem. - Eur. J. 2017, 23, 3347−
Electron Density  Does the Difference Electron-Density Analysis 3356.
Suffice for a Description of the Chemical Bond? Angew. Chem., Int. Ed. (140) Georgiou, D. C.; Zhao, L.; Wilson, D. J. D.; Frenking, G.;
Engl. 1984, 23, 627−628. Dutton, J. L. NHC-Stabilised Acetylene - How Far Can the Analogy Be
(121) Vogel, E.; Tuckmantel, W.; Schogl, K.; Widhalm, M.; Kraka, E.; Pushed? Chem. - Eur. J. 2017, 23, 2926−2934.
Cremer, D. Zur Konfigurativen Stabilität Syn/Anti-Isomerer Ü ber- (141) Krapp, A.; Pandey, K. K.; Frenking, G. Transition Metal−
brückter [14]Annulene Mit Anthracen-Perimeter. Tetrahedron Lett. Carbon Complexes. A Theoretical Study. J. Am. Chem. Soc. 2007, 129,
1984, 25, 4925−4928.
7596−7610.
(122) Matta, C. F.; Boyd, R. J. The Quantum Theory of Atoms in
(142) Celik, M. A.; Frenking, G.; Neumüller, B.; Petz, W. Exploiting
Molecules: From Solid State to DNA and Drug Design; Wiley-VCH:
the Two fold Donor Ability of Carbodiphosphoranes: Theoretical
Weinheim, 2007.
(123) Mitoraj, M. P.; Michalak, A.; Ziegler, T. A Combined Charge Studies of [(PPh3)2C→EH2]q (Eq = Be, B+, C2+, N3+, O4+) and
and Energy Decomposition Scheme for Bond Analysis. J. Chem. Theory Synthesis of the Dication [(Ph3P)2CCH2]2+. ChemPlusChem 2013, 78,
Comput. 2009, 5, 962−975. 1024−1032.
(124) Michalak, A.; Mitoraj, M.; Ziegler, T. Bond Orbitals from (143) Wu, Z.; Xu, J.; Sokolenko, L.; Yagupolskii, Y. L.; Feng, R.; Liu,
Chemical Valence Theory. J. Phys. Chem. A 2008, 112, 1933−1939. Q.; Lu, Y.; Zhao, L.; Fernández, I.; Frenking, G.; Trabelsi, T.; Francisco,
(125) Kitaura, K.; Morokuma, K. A New Energy Decomposition J. S.; Zeng, X. Parent Thioketene S-Oxide H2CCSO: Gas-Phase
Scheme for Molecular Interactions within the Hartree-Fock Approx- Generation, Structure, and Bonding Analysis. Chem. - Eur. J. 2017, 23,
imation. Int. J. Quantum Chem. 1976, 10, 325−340. 16566−16573.
(126) Ziegler, T.; Rauk, A. On the Calculation of Bonding Energies by (144) Petz, W.; Andrada, D. M.; Hermann, M.; Frenking, G.;
the Hartree Fock Slater Method. Theor. Chim. Acta 1977, 46, 1−10. Neumüller, B. A C2 Fragment as Four-Electron σ Donor. Z. Anorg. Allg.
(127) Mitoraj, M.; Michalak, A. Natural Orbitals for Chemical Chem. 2017, 643, 1096−1099.
Valence as Descriptors of Chemical Bonding in Transition Metal (145) Pecher, L.; Pan, S.; Frenking, G. Chemical Bonding in the
Complexes. J. Mol. Model. 2007, 13, 347−355. Hexamethylbenzene−SO2+ Dication. Theor. Chem. Acc. 2019, 138, 47.
(128) Mitoraj, M.; Michalak, A. Donor−Acceptor Properties of (146) Pan, S.; Saha, R.; Osorio, E.; Chattaraj, P. K.; Frenking, G.;
Ligands from the Natural Orbitals for Chemical Valence. Organo- Merino, G. Ligand-Supported E3 Clusters (E = Si-Sn). Chem.Eur. J.
metallics 2007, 26, 6576−6580. 2017, 23, 7463−7473.

8838 DOI: 10.1021/acs.chemrev.8b00722


Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

(147) Martín Pendás, A.; Francisco, E.; Blanco, M. A. Binding (165) Weinhold, F.; Landis, C. R.; Glendening, E. D. What is NBO
Energies of First Row Diatomics in the Light of the Interacting Analysis and How is It Useful? Int. Rev. Phys. Chem. 2016, 35, 399−440.
Quantum Atoms Approach. J. Phys. Chem. A 2006, 110, 12864−12869. (166) Heisenberg, W. Der Teil und das Ganze, Gespräche im Umkreis
(148) Blanco, M. A.; Martín Pendás, A.; Francisco, E. Interacting der Atomphysik; Piper: München, 2001.
Quantum Atoms: A Correlated Energy Decomposition Scheme Based (167) Clark, T.; Murray, J. S.; Politzer, P. A Perspective on Quantum
on the Quantum Theory of Atoms in Molecules. J. Chem. Theory Mechanics and Chemical Concepts in Describing Noncovalent
Comput. 2005, 1, 1096−1109. Interactions. Phys. Chem. Chem. Phys. 2018, 20, 30076−30082.
(149) Francisco, E.; Martín Pendás, A.; Blanco, M. A. A Molecular (168) Ivanic, J.; Atchity, G. J.; Ruedenberg, K. Intrinsic Local
Energy Decomposition Scheme for Atoms in Molecules. J. Chem. Constituents of Molecular Electronic Wave Functions. I. Exact
Theory Comput. 2006, 2, 90−102. Representation of the Density Matrix in Terms of Chemically
(150) Menéndez, M.; Á varez Boto, R.; Francicso, E.; Pendás, A. M. Deformed and Oriented Atomic Minimal Basis Set Orbitals. Theor.
One-Electron Images in Real Space: Natural Adaptive Orbitals. J. Chem. Acc. 2008, 120, 281−294.
Comput. Chem. 2015, 36, 833−843. (169) Schrödinger, E. Quantisierung als Eigenwertproblem. Ann. Phys.
(151) Hirshfeld, E. L. Bonded-Atom Fragments for Describing 1926, 81, 109−139.
Molecular Charge Densities. Theor. Chim. Acta 1977, 44, 129−138. (170) A nice didactical discussion about the paradoxical roles of
(152) For the connection of the Lewis electron-pair model with the kinetic and potential Energy contributions to the chemical bond has
EDA-NOCV method see: Zhao, L.; Hermann, M.; Schwarz, W. H. E.; been published by: Rioux, F. Kinetic Energy and the Covalent Bond in
Frenking, G. The Lewis electron-pair bonding model: Modern energy H2+. Chem. Educ. 1997, 2, 1−14.
decomposition analysis. Nature Rev. Chem. 2019, 3, 48−63. (171) Another didactically skilful presentation is given by: Burdett, J.
(153) Bollermann, T.; Cadenbach, T.; Gemel, C.; von Hopffgarten, K. Chemical Bonds: A Dialog; Wiley: Chichester, 1997; p 3f.
M.; Frenking, G.; Fischer, R. A. Molecular Alloys: Experimental and (172) For a detailed analysis of the physical origin of the bond
Theoretical Investigations on the Substitution of Zinc by Cadmium and formation in diatomic molecules, see refs 14 and 15.
Mercury in the Homologous Series [Mo(M′R)12] and [M(M′R)8] (M (173) For a detailed discussion of the physical mechanisms of bond
= Pd, Pt; M′=Zn, Cd, Hg). Chem.Eur. J. 2010, 16, 13372−13384. formation in H2+, H2, Li2+, and Li2, see ref 34.
(154) von Hopffgarten, M.; Frenking, G. Building a Bridge between (174) For a recent controversy see: Gershoni-Poranne, R.; Chen, P.
Coordination Compounds and Clusters: Bonding Analysis of the The Carbon−Nitrogen Bonds in Ammonium Compounds Are Charge
Icosahedral Molecules [M(ER)12] (M = Cr, Mo, W; E = Zn, Cd, Hg). J. Shift Bonds. Chem. - Eur. J. 2017, 23, 4659−4668 and refs 175 and 176 .
Phys. Chem. A 2011, 115, 12758−12768. (175) Frenking, G. Covalent Bonding and Charge Shift Bonds:
(155) Nguyen, T. A. N.; Frenking, G. Transition-Metal Complexes of Comment on “The Carbon−Nitrogen Bonds in Ammonium
Tetrylones [(CO)5W-E(PPh3)2] and Tetrylenes [(CO)5W-NHE] (E = Compounds Are Charge Shift Bonds. Chem. - Eur. J. 2017, 23,
C−Pb): A Theoretical Study. Chem.Eur. J. 2012, 18, 12733−12748. 18320−18324.
(156) Celik, M. A.; Dash, C.; Adiraju, V. A. K.; Das, A.; Yousufuddin, (176) Chen, P.; Gershoni-Poranne, R. Response to “Covalent
M.; Frenking, G.; Dias, H. V. R. End-On and Side-On π-Acid Ligand Bonding and Charge Shift Bonds: Comment on ‘The Carbon−
Adducts of Gold(I): Carbonyl, Cyanide, Isocyanide, and Cyclooctyne Nitrogen Bonds in Ammonium Compounds Are Charge Shift Bonds.
Gold(I) Complexes Supported by N-Heterocyclic Carbenes and Chem. - Eur. J. 2017, 23, 18325−18329.
Phosphines. Inorg. Chem. 2013, 52, 729−742. (177) Spackman, M. A.; Maslen, E. N. Chemical Properties from the
(157) Mousavi, M.; Frenking, G. Bonding Analysis of the Promolecule. J. Phys. Chem. 1986, 90, 2020−2027.
Trimethylenemethane (TMM) Complexes [(η6-C6H6)M-TMM] (M (178) Kutzelnigg, W. In The Concept of the Chemical Bond; Maksic, Z.
= Fe, Ru, Os), [(η5-C5H5)M-TMM] (M = Co, Rh, Ir), and [(η4- B., Ed.; Springer Berlin/Heidelberg, 1990; Vol. 1, p 1.
C4H4)M-TMM] (M = Ni, Pd, Pt). Organometallics 2013, 32, 1743− (179) We remind the reader that the terms σ and π bonds refer to the
1751. symmetry of the orbitals, which do not automatically identify single and
(158) Das, A.; Dash, C.; Celik, M. A.; Yousufuddin, M.; Frenking, G.; double bonds as it is sometimes assumed. Saturated molecules such as
Dias, H. V. R. Tris(alkyne) and Bis(alkyne) Complexes of Coinage CH4 and C2H6 have σ and π orbitals but they have no double bonds.
Metals: Synthesis and Characterization of (cyclooctyne)3M+ (M = Cu, (180) The repulsive forces between the atomic nuclei become
Ag) and (cyclooctyne)2Au+ and Coinage Metal (M = Cu, Ag, Au) effective only at very short distances; they are negligible at longer
Family Group Trends. Organometallics 2013, 32, 3135−3144. distances.
(159) Weinberger, D. S.; Melaimi, M.; Moore, C. E.; Rheingold, A. L.; (181) Scuseria, G. E.; Miller, M. D.; Jensen, F.; Geertsen. The dipole
Gernot, F.; Jerabek, P.; Bertrand, G. Isolation of Neutral Mono- and moment of carbon monoxide. J. Chem. Phys. 1991, 94, 6660−6666.
Dinuclear Gold Complexes of Cyclic (Alkyl)(amino)carbenes. Angew. (182) Cui, Z.-h.; Yang, W.-s.; Zhao, L.; Ding, Y.-h.; Frenking, G.
Chem., Int. Ed. 2013, 52, 8964−8967. Unusually Short Be−Be Distances with and without a Bond in Be2F2
(160) Mondal, K. C.; Samuel, P. P.; Roesky, H. W.; Carl, E.; Herbst- and in the Molecular Discuses Be2B8 and Be2B7−. Angew. Chem., Int. Ed.
Irmer, R.; Stalke, D.; Schwederski, B.; Kaim, W.; Ungur, L.; Chibotaru, 2016, 55, 7841−7846.
L. F.; Hermann, M.; Frenking, G. Stabilization of a Cobalt−Cobalt (183) Dasent, W. E. Inorganic Energetics, 2nd ed.; Cambridge
Bond by Two Cyclic Alkyl Amino Carbenes. J. Am. Chem. Soc. 2014, University Press: New York, 1982; p35.
136, 1770−1773. (184) Efimenko, J. A. Mass Spectrometric Study of the BeO-BeF2
(161) Jerabek, P.; Roesky, H. W.; Bertrand, G.; Frenking, G. Coinage System at High Temperatures. J. Res. Natl. Bur. Stand., Sect. A 1968,
Metals Binding as Main Group Elements: Structure and Bonding of the 72A, 75−80.
Carbene Complexes [TM(CAAC)2] and [TM(CAAC)2]+ (TM = Cu, (185) Holzmann, N.; Stasch, A.; Jones, C.; Frenking, G. Structures
Ag, Au). J. Am. Chem. Soc. 2014, 136, 17123−17135. and Stabilities of Group 13 Adducts [(NHC)(EX 3 )] and
(162) Caramori, G. F.; Piccoli, R. M.; Segala, M.; Muñoz-Castro, A.; [(NHC)2(E2Xn)] (E = B to In; X = H, Cl; n = 4, 2, 0; NHC = N-
Guajardo-Maturana, R.; Andrada, D. M.; Frenking, G. Cyclic Heterocyclic Carbene) and the Search for Hydrogen Storage Systems:
Trinuclear Copper(I), Silver(I), and Gold(I) Complexes: A Theoreti- A Theoretical Study. Chem. - Eur. J. 2011, 17, 13517−13525.
cal Insight. Dalton Trans 2015, 44, 377−385. (186) Braunschweig, H.; Dewhurst, R. D.; Hammond, K.; Mies, J.;
(163) Couzijn, E. P. A.; Lai, Y.-Y.; Limacher, A.; Chen, P. Intuitive Radacki, K.; Vargas, A. Ambient-Temperature Isolation of a Compound
Quantifiers of Charge Flows in Coordinate Bonding. Organometallics with a Boron-Boron Triple Bond. Science 2012, 336, 1420−1422.
2017, 36, 3205−3214. (187) Frenking, G.; Holzmann, N. A Boron-Boron Triple Bond.
(164) Landis, C. R.; Hughes, R. P.; Weinhold, F. Bonding Analysis of Science 2012, 336, 1394−1395.
TM(cAAC)2 (TM = Cu, Ag, and Au) and the Importance of Reference (188) Köppe, R.; Schnöckel, H. The Boron−Boron Triple Bond? A
State. Organometallics 2015, 34, 3442−3449. Thermodynamic and Force Field based Interpretation of the N-

8839 DOI: 10.1021/acs.chemrev.8b00722


Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

heterocyclic Carbene (NHC) Stabilization Procedure. Chem. Sci. 2015, (211) Stang, P. J.; Arif, A. M.; Zhdankin, V. V. Reaction of E-1,2-
6, 1199−1205. bis[triphenyl(trifluoromethanesulfonyloxy)phospho]ethylene,
(189) Holzmann, N.; Hermann, M.; Frenking, G. The Boron−Boron Ph3PCH = CHPPh3·2Otf with Bases: Unusual Products and Evidence
Triple Bond in NHC→B≡B←NHC. Chem. Sci. 2015, 6, 4089−4094. for C2-diylide, Ph3P=CC=PPh3, Formation. Tetrahedron 1991, 47,
(190) Böhnke, J.; Braunschweig, H.; Ewing, W. C.; Hörl, C.; Kramer, 4539−4546.
T.; Krummenacher, I.; Mies, J.; Vargas, A. Diborabutatriene: An (212) Wilson, D. J. D.; Couchman, S. A.; Dutton, J. L. Are N-
Electron-Deficient Cumulene. Angew. Chem., Int. Ed. 2014, 53, 9082− Heterocyclic Carbenes “Better” Ligands than Phosphines in Main
9085. Group Chemistry? A Theoretical Case Study of Ligand-Stabilized E2
(191) Böhnke, J.; Braunschweig, H.; Dellermann, T.; Ewing, W. C.; Molecules, L-E-E-L (L = NHC, phosphine; E = C, Si, Ge, Sn, Pb, N, P,
Hammond, K.; Jimenez-Halla, J. O. C.; Kramer, T.; Mies, J. The As, Sb, Bi). Inorg. Chem. 2012, 51, 7657−7668.
Synthesis of B2(SIDip)2 and its Reactivity Between the Diboracumu- (213) Georgiou, D. C.; Stringer, B. D.; Hogan, C. F.; Barnard, P. J.;
lenic and Diborynic Extremes. Angew. Chem., Int. Ed. 2015, 54, 13801− Wilson, D. J. D.; Holzmann, N.; Frenking, G.; Dutton, J. L. The Fate of
13805. NHC-Stabilized Dicarbon. Chem. - Eur. J. 2015, 21, 3377−3386.
(192) Zhou, M.; Tsumori, N.; Li, Z.; Fan, K.; Andrews, L.; Xu, Q. (214) Li, Y.; Mondal, K. C.; Samuel, P. P.; Zhu, H.; Orben, C. M.;
OCBBCO: A Neutral Molecule with Some Boron−Boron Triple Bond Panneerselvam, S.; Dittrich, B.; Schwederski, B.; Kaim, W.; Mondal, T.;
Character. J. Am. Chem. Soc. 2002, 124, 12936−12937. Koley, D.; Roesky, H. W. C4 Cumulene and the Corresponding Air-
(193) Li, S.-D.; Zhai, H.-J.; Wang, L.-S. B2(BO)22‑ Diboronyl Stable Radical Cation and Dication. Angew. Chem., Int. Ed. 2014, 53,
Diborene: A Linear Molecule with a Triple Boron−Boron Bond. J. 4168−4172.
Am. Chem. Soc. 2008, 130, 2573−2579. (215) Jin, L.; Melaimi, M.; Liu, L.; Bertrand, G. Singlet Carbenes as
(194) Ducati, L. C.; Takagi, N.; Frenking, G. Molecules with All Triple Mimics for Transition Metals: Synthesis of an Air Stable Organic Mixed
Bonds: OCBBCO, N2BBN2, and [OBBBBO]2−. J. Phys. Chem. A 2009, Valence Compound [M2(C2)+̇ ; M = cyclic(alkyl)(amino)carbene].
113, 11693−11698. Org. Chem. Front. 2014, 1, 351−354.
(195) Hermann, M.; Frenking, G. The Chemical Bond in C2. Chem. - (216) For a discussion of the electronic states of C2, see: Kirby, K.; Liu,
Eur. J. 2016, 22, 4100−4108. B. The Valence States of C2: A Configuration Interaction Study. J.
(196) Shaik, S.; Danovich, D.; Wu, W.; Su, P.; Rzepa, H. S.; Hiberty, P. Chem. Phys. 1979, 70, 893−900.
C. Quadruple Bonding in C2 and Analogous Eight-Valence Electron (217) The original work in reference 213 considered only the (2)1Δg
Species. Nat. Chem. 2012, 4, 195−200. state of C2, and the EDA-NOCV calculations of C2L2 (L = NHCMe,
(197) Shaik, S.; Rzepa, H. S.; Hoffmann, R. One Molecule, Two CAACMe) were carried out with the latter state as reference. Later
Atoms, Three Views, Four Bonds? Angew. Chem., Int. Ed. 2013, 52, calculations carried out by one of us (NH) that are reported here
3020−3033. showed that C2(CAACMe)2 is better described using the 5Δg state of C2
(198) Danovich, D.; Hiberty, P. C.; Wu, W.; Rzepa, H. S.; Shaik, S. as reference.
The Nature of the Fourth Bond in the Ground State of C2: The (218) Melaimi, M.; Jazzar, R.; Soleilhavoup, M.; Bertrand, G. Cyclic
Quadruple Bond Conundrum. Chem. - Eur. J. 2014, 20, 6220−6232. (Alkyl)(amino)carbenes (CAACs): Recent Developments. Angew.
(199) Danovich, D.; Shaik, S.; Rzepa, H. S.; Hoffmann, R. A Response Chem., Int. Ed. 2017, 56, 10046−10068.
to the Critical Comments on “One Molecule, Two Atoms, Three (219) Burford, R. J.; Fryzuk, M. D. Examining the Relationship
Views, Four Bonds? Angew. Chem., Int. Ed. 2013, 52, 5926−5928. between Coordination Mode and Reactivity of Dinitrogen. Nat. Rev.
(200) Zou, W.; Cremer, D. C2 in a Box: Determining Its Intrinsic Chem. 2017, 1, 0026.
Bond Strength for the X1Σg+ Ground State. Chem. - Eur. J. 2016, 22, (220) Wang, Y.; Robinson, G. H. Carbene Stabilization of Highly
4087−4099. Reactive Main-Group Molecules. Inorg. Chem. 2011, 50, 12326−
(201) Piris, M.; Lopez, X.; Ugalde, J. M. The Bond Order of C2 from a 12337.
Strictly N-Representable Natural Orbital Energy Functional Perspec- (221) Kuhn, N.; Al-Sheikh, A. 2,3-Dihydroimidazol-2-ylidenes and
tive. Chem. - Eur. J. 2016, 22, 4109−4115. Their Main Group Element Chemistry. Coord. Chem. Rev. 2005, 249,
(202) Cooper, D. L.; Ponec, R.; Kohout, M. New insights from 829−857.
domain-averaged Fermi holes and bond order analysis into the bonding (222) Wang, Y.; Robinson, G. H. Carbene-stabilized Main Group
conundrum in C2. Mol. Phys. 2016, 114, 1270−1284. Diatomic Allotropes. Dalton Trans 2012, 41, 337−345.
(203) Xu, L. T.; Dunning, T. H., Jr. Insights into the Perplexing Nature (223) Wang, Y.; Robinson, G. H. Unique Homonuclear Multiple
of the Bonding in C2 from Generalized Valence Bond Calculations. J. Bonding in Main Group Compounds. Chem. Commun. 2009, 5201−
Chem. Theory Comput. 2014, 10, 195−201. 5213.
(204) de Sousa, D. W. O.; Nascimento, M. A. C. Is There a Quadruple (224) Wang, Y.; Xie, Y.; Wei, P.; King, R. B.; Schaefer, H. F.; von, R.;
Bond in C2? J. Chem. Theory Comput. 2016, 12, 2234−2241. Schleyer, P.; Robinson, G. H. A Stable Silicon(0) Compound with a
(205) Frenking, G.; Hermann, M. Critical Comments on “One Si=Si Double Bond. Science 2008, 321, 1069−1071.
Molecule, Two Atoms, Three Views, Four Bonds? Angew. Chem., Int. (225) Sidiropoulos, A.; Jones, C.; Stasch, A.; Klein, S.; Frenking, G. N-
Ed. 2013, 52, 5922−5925. Heterocyclic Carbene Stabilized Digermanium(0). Angew. Chem., Int.
(206) Shaik, S.; Danovich, D.; Braida, B.; Hiberty, P. C. The Ed. 2009, 48, 9701−9704.
Quadruple Bonding in C2 Reproduces the Properties of the Molecule. (226) Jones, C.; Sidiropoulos, A.; Holzmann, N.; Frenking, G.; Stasch,
Chem. - Eur. J. 2016, 22, 4116−4128. A. An N-heterocyclic Carbene Adduct of Diatomic Tin,:Sn=Sn: Chem.
(207) For a comment on this work see: Frenking, G.; Hermann, M. Commun. 2012, 48, 9855−9857.
Comment on “The Quadruple Bonding in C2 Reproduces the (227) Wang, Y.; Xie, Y.; Wei, P.; King, R. B.; Schaefer, H. F.; Schleyer,
Properties of the Molecule. Chem. - Eur. J. 2016, 22, 18975−18976. P. v. R.; Robinson, G. H. Carbene-Stabilized Diphosphorus. J. Am.
(208) Shaik, S.; Danovich, D.; Braida, B.; Hiberty, P. C. A Response to Chem. Soc. 2008, 130, 14970−14971.
a Comment by G. Frenking and M. Hermann on: “The Quadruple (228) Abraham, M. Y.; Wang, Y.; Xie, Y.; Wei, P.; Schaefer, H. F.;
Bonding in C2 Reproduces the Properties of the Molecule. Chem.Eur. Schleyer, P. v. R.; Robinson, G. H. Carbene Stabilization of Diarsenic:
J. 2016, 22, 18977−18980. From Hypervalency to Allotropy. Chem. - Eur. J. 2010, 16, 432−435.
(209) This model was suggested by David Cooper at the Symposium (229) Reinmuth, M.; Neuhäuser, C.; Walter, P.; Enders, M.; Kaifer, E.;
The Chemical Bonds at the 21th Century; 14 − 18.6.2015, Xiamen, Himmel, H.-J. The Flexible Coordination Modes of Guanidine Ligands
China. in Zn Alkyl and Halide Complexes: Chances for Catalysis. Eur. J. Inorg.
(210) Karadakov, P. B.; Kirsopp, J. Magnetic Shielding Studies of C2 Chem. 2011, 2011, 83−90.
and C2H2 Support Higher than Triple Bond Multiplicity in C2. Chem. - (230) Appel, R.; Schöllhorn, R. Triphenylphosphineazine Ph3P=N-
Eur. J. 2017, 23, 12949−12954. N=PPh3. Angew. Chem., Int. Ed. Engl. 1964, 3, 805−805.

8840 DOI: 10.1021/acs.chemrev.8b00722


Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

(231) Holzmann, N.; Dange, D.; Jones, C.; Frenking, G. Dinitrogen as (255) Pimentel, G. C. The Bonding of Trihalide and Bifluoride Ions
Double Lewis Acid: Structure and Bonding of Triphenylphosphinazine by the Molecular Orbital Method. J. Chem. Phys. 1951, 19, 446−448.
N2(PPh3)2. Angew. Chem., Int. Ed. 2013, 52, 3004−3008. (256) Coulson, C. A. The Nature of the Bonding in Xenon Fluorides
(232) Wong, M. W.; Nobes, R. H.; Bouma, W. J.; Radom, L. and Related Molecules. J. Chem. Soc. 1964, 1442−1454.
Isoelectronic Analogs of Molecular Nitrogen: Tightly Bound Multiply (257) Kimball, G. E. Directed Valence. J. Chem. Phys. 1940, 8, 188−
Charged Species. J. Chem. Phys. 1989, 91, 2971−2979. 198.
(233) Hurley, A. C. Potential Energy Curves for Doubly Positive (258) The Coulombic attraction between neutral sulfur atom and F6
Diatomic Ions: Part II. Predicted States and Transitions of N22+, O22+, in SF6 using a spherically symmetric charge distribution at S has been
and NO22+. J. Mol. Spectrosc. 1962, 9, 18−29. estimated to provide 22% of the total attraction: Lein, M.; Frenking, G.
(234) Basch, H.; Hoz, S.; Goldberg, M.; Gamss, L. Electronic Chemical Bonding in Octahedral XeF6 and SF6. Aust. J. Chem. 2004, 57,
Structure and Properties of the O22+, SO2+ and S22+ Diatomic Dications. 1191−1195.
Isr. J. Chem. 1991, 31, 335−343. (259) Hirshfeld, F. L.; Rzotkiewicz, S. Electrostatic Binding in the
(235) Nenajdenko, V. G.; Shevchenko, N. E.; Balenkova, E. S.; First-Row AH and A2 Diatomic Molecules. Mol. Phys. 1974, 27, 1319−
Alabugin, I. V. 1,2-Dications in Organic Main Group Systems. Chem. 1343.
Rev. 2003, 103, 229−282. (260) For an early discussion of the bonding in pentacoordinated
(236) Fournier, J.; Fournier, P. G.; Langford, M. L.; Mousselmal, M.; phosphorous compounds, see: Hoffmann, R.; Howell, J. M.;
Robbe, J. M.; Gandara, G. An Experimental and Theoretical Study of Muetterties, E. L. Molecular Orbital Theory of Pentacoordinate
the Doubly Charged Ion O22+. J. Chem. Phys. 1992, 96, 3594−3602. Phosphorus. J. Am. Chem. Soc. 1972, 94, 3047−3058.
(237) Andrada, D. M.; Casals-Sainz, J. L.; Martín Pendás, Á .; (261) Frenking, G. Multiple Bonding of Heavy Main-Group Atoms. In
Frenking, G. Dative and Electron-Sharing Bonding in C2F4. Chem. - Eur. The Chemical Bond. Chemical Bonding Across the Periodic Table;
J. 2018, 24, 9083−9089. Frenking, G., Shaik, S., Eds.; Wiley-VCH: Weinheim, 2014; pp 25−48.
(238) Zou, W.; Kalescky, R.; Kraka, E.; Cremer, D. Relating Normal (262) (a) E = Si: see ref 224; (b) E = Ge: see ref 225; (c) E = Sn: see
Vibrational Modes to Local Vibrational Modes with the Help of an ref 226; (d) E = B: see ref 186; (e) E = P: see ref 227; (f) E = As: see ref
Adiabatic Connection Scheme. J. Chem. Phys. 2012, 137, 084114. 228.
(239) Kalescky, R.; Kraka, E.; Cremer, D. Identification of the (263) Holzmann, N.; Frenking, G. Bonding Situation in Dimeric
Strongest Bonds in Chemistry. J. Phys. Chem. A 2013, 117, 8981−8995. Group 15 Complexes [(NHC)2(E2)] (E = N−Bi). Z. Naturforsch. A
(240) Kraka, E.; Cremer, D. Characterization of CF Bonds with 2014, 69, 385−395.
Multiple-Bond Character: Bond Lengths, Stretching Force Constants, (264) Holzmann, N.; Andrada, D. M.; Frenking, G. Bonding Situation
and Bond Dissociation Energies. ChemPhysChem 2009, 10, 686−698. in Silicon Complexes [(L)2(Si2)] and [(L)2(Si)] with NHC and CAAC
(241) Kalescky, R.; Zou, W.; Kraka, E.; Cremer, D. Quantitative ligands. J. Organomet. Chem. 2015, 792, 139−148.
Assessment of the Multiplicity of Carbon-Halogen Bonds: Carbenium (265) Lee, T. J.; Martin, J. M. L. An Accurate Quartic Force Field,
and Halonium Ions with F, Cl, Br, I. J. Phys. Chem. A 2014, 118, 1948− Fundamental Frequencies, and Binding Energy for The High Energy
1963. Density Material Td N4. Chem. Phys. Lett. 2002, 357, 319−325.
(242) Senekowitsch, J.; ONeil, S.; Meyer, W. On the Bonding in (266) Brassington, N. J.; Edwards, H. G. M.; Long, D. A. The
Doubly Charged Diatomics. Theor. Chim. Acta 1992, 84, 85−93. Vibration-Rotation Raman Spectrum of P4. J. Raman Spectrosc. 1981,
(243) Senekowitsch, J.; Oneil, S. Metastable 3Σ−g Ground State of F2+2 11, 346−348.
and the Bonding in Molecular Dications. J. Chem. Phys. 1991, 95, (267) Jerabek, P.; Frenking, G. Comparative Bonding Analysis of N2
1847−1851. and P2 versus Tetrahedral N4 and P4. Theor. Chem. Acc. 2014, 133,
(244) Hoffmann, R. at the symposium Chemical Bonding in the 21st 1447.
Century, May 28, 2018, Brussels, Belgium. (268) Jerabek, P.; Frenking, G. Erratum to: Comparative Bonding
(245) Kutzelnigg, W. Chemical Bonding in Higher Main Group Analysis of N2 and P2 versusTetrahedral N4 and P4. Theor. Chem. Acc.
Elements. Angew. Chem., Int. Ed. Engl. 1984, 23, 272−295. 2015, 134, 136.
(246) The difference between the radii of the valence s and p orbitals (269) Weidenbruch, M. Some Silicon, Germanium, Tin, and Lead
of main-group atoms between the first and higher octal rows of the Analogues of Carbenes, Alkenes, and Dienes. Eur. J. Inorg. Chem. 1999,
periodic systems was also pointed out in: Pyykkö, P. Interpretation of 1999, 373−381.
secondary periodicity in the periodic system. J. Chem. Res., Synop 1979, (270) Power, P. P. Homonuclear Multiple Bonding in Heavier Main
380−381. Group Elements. J. Chem. Soc., Dalton Trans. 1998, 2939−2951.
(247) Kaupp, M. Chemical bonding of main-group elements. In The (271) Okazaki, R.; Tokitoh, N. Heavy Ketones, the Heavier Element
Chemical Bond: Chemical Bonding Across the Periodic Table; Frenking, Congeners of a Ketone. Acc. Chem. Res. 2000, 33, 625−630.
G., Shaik, S., Eds.; Wiley-VCH: Weinheim, 2014; pp 1 − 24. (272) Tokitoh, N.; Okazaki, R. Recent Advances in the Chemistry of
(248) Kaupp, M. On the Role of Radial Nodes of Atomic Orbitals for Group 14-Group 16 Double Bond Compounds. Adv. Organomet. Chem.
Chemical Bonding and the Periodic Table. J. Comput. Chem. 2007, 28, 2001, 47, 121−166.
320−325. (273) Power, P. P. In Struct. Bonding (Berlin); Roesky, H. W., Atwood,
(249) Kaupp, M.; Schleyer, P. v. R. Ab Initio Study of Structures and D. A., Eds.; Springer-Verlag: Berlin, 2002; Vol. 103, pp 57−84.
Stabilities of Substituted Lead Compounds. Why is Inorganic Lead (274) Weidenbruch, M. Some Recent Advances in the Chemistry of
Chemistry Dominated by PbII but Organolead Chemistry by PbIV? J. Silicon and Its Homologues in Low Coordination States. J. Organomet.
Am. Chem. Soc. 1993, 115, 1061−1073. Chem. 2002, 646, 39−52.
(250) Kaupp, M.; van Wüllen, C.; Franke, R.; Schmitz, F.; Kutzelnigg, (275) Tokitoh, N.; Okazaki, R. In The Chemistry of Organic
W. The Structure of XeF6 and of Compounds Isoelectronic with It. A Germanium, Tin and Lead Compounds; Rappoport, Z., Ed.; John
Challenge to Computational Chemistry and to the Qualitative Theory Wiley and Sons: Chichester, 2002; pp 843−901.
of the Chemical Bond. J. Am. Chem. Soc. 1996, 118, 11939−11950. (276) West, R. Multiple Bonds to Silicon: 20 Years Later. Polyhedron
(251) https://goldbook.iupac.org/html/V/V06588.html. 2002, 21, 467−472.
(252) Rundle, R. E. Electron Deficient Compounds. J. Am. Chem. Soc. (277) Power, P. P. Persistent and Stable Radicals of the Heavier Main
1947, 69, 1327−1331. Group Elements and Related Species. Chem. Rev. 2003, 103, 789−810.
(253) Hach, R. J.; Rundle, R. E. The Structure of Tetramethylammo- (278) Power, P. P. Silicon, Germanium, Tin and Lead Analogues of
nium Pentaiodide. J. Am. Chem. Soc. 1951, 73, 4321−4324. Acetylenes. Chem. Commun. 2003, 0, 2091−2101.
(254) Rundle, R. E. On the Problem Structure of XeF4 and XeF2. J. (279) Gusel’nikov, L. E. Hetero-π-systems from 2 + 2 Cyclo-
Am. Chem. Soc. 1963, 85, 112−113. reversions. Part 1. Gusel’nikov−Flowers Route to Silenes and

8841 DOI: 10.1021/acs.chemrev.8b00722


Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

Origination of the Chemistry of Doubly Bonded Silicon. Coord. Chem. (305) Bogey, M.; Bolvin, H.; Demuyneck, C.; Destombes, J.-L.
Rev. 2003, 244, 149−240. Nonclassical Double-Bridged Structure in Silicon-Containing Mole-
(280) Weidenbruch, M. From a Cyclotrisilane to a Cyclotriplum- cules: Experimental Evidence in Si2H2 from its Submillimeter-Wave
bane: Low Coordination and Multiple Bonding in Group 14 Spectrum. Phys. Rev. Lett. 1991, 66, 413.
Chemistry. Organometallics 2003, 22, 4348−4360. (306) Cordonnier, M.; Bogey, M.; Demuynck, C.; Destombes, J.-L.
(281) Kira, M. Isolable Silylene, Disilenes, Trisilaallene, and Related Nonclassical Structures in Silicon-Containing Molecules: The Mono-
Compounds. J. Organomet. Chem. 2004, 689, 4475−4488. bridged Isomer of Si2H2. J. Chem. Phys. 1992, 97, 7984.
(282) Yoshifuji, M. Recent Developments in the Chemistry of Low- (307) For a review see: Karni, M.; Apeloig, Y.; Kapp, J.; Schleyer, P. v.
Coordinated Organophosphorus Compounds. Pure Appl. Chem. 2005, R. In The Chemistry of Organic Silicon Compounds; Apeloig, Y., Ed.;
77, 2011−2020. Wiley: Chichester, 2001; Vol. 3, p 1.
(283) Ottosson, H.; Steel, P. G. Silylenes, Silenes, and Disilenes: (308) Wang, X.; Andrews, L.; Kushto, G. Infrared Spectra of the Novel
Novel Silicon-Based Reagents for Organic Synthesis? Chem.Eur. J. Ge2H2 and Ge2H4 Species and the Reactive GeH1,2,3 Intermediates in
2006, 12, 1576−1585. Solid Neon, Deuterium and Argon. J. Phys. Chem. A 2002, 106, 5809−
(284) Sekiguchi, A.; Ichinohe, M.; Kinjo, R. The Chemistry of 5816.
Disilyne with a Genuine Si−Si Triple Bond: Synthesis, Structure, and (309) Wang, X.; Andrews, L.; Chertihin, G. V.; Souter, P. F. Infrared
Reactivity. Bull. Chem. Soc. Jpn. 2006, 79, 825−832. Spectra of the Novel Sn2H2 Species and the Reactive SnH1,2,3 and
(285) Kira, M.; Iwamoto, T. Progress in the Chemistry of Stable PbH1,2,3 Intermediates in Solid Neon, Deuterium, and Argon. J. Phys.
Disilenes. Adv. Organomet. Chem. 2006, 54, 73−148. Chem. A 2002, 106, 6302−6308.
(286) Lee, V. Y.; Sekiguchi, A. Stable Silyl, Germyl, and Stannyl (310) Wang, X.; Andrews, L. Infrared Spectra of Group 14 Hydrides in
Cations, Radicals, and Anions: Heavy Versions of Carbocations, Solid Hydrogen: Experimental Observation of PbH4, Pb2H2 and Pb2H4.
Carbon Radicals, and Carbanions. Acc. Chem. Res. 2007, 40, 410−419. J. Am. Chem. Soc. 2003, 125, 6581−6587.
(287) Rivard, E.; Power, P. P. Multiple Bonding in Heavier Element (311) Wiberg, N.; Niedermayer, W.; Fischer, G.; Nöth, H.; Suter, M.
Compounds Stabilized by Bulky Terphenyl Ligands. Inorg. Chem. 2007, Synthesis, Structure and Dehalogenation of the Disilene RClSi = SiClR
46, 10047−10064. [R = (tBu3Si)2MeSi]. Eur. J. Inorg. Chem. 2002, 2002, 1066−1070.
(288) Wang, Y.; Robinson, G. H. Organometallics of the Group 13 M- (312) Wiberg, N.; Vasisht, S. K.; Fischer, G.; Mayer, P. Disilynes. III
M Bond (M= Al, Ga, In) and the Concept of Metalloaromaticity. [1] A Relatively Stable Disilyne RSi≡SiR (R = SiMe(SitBu3)2). Z.
Organometallics 2007, 26, 2−11. Anorg. Allg. Chem. 2004, 630, 1823−1828.
(289) Power, P. P. Bonding and Reactivity of Heavier Group 14 (313) Sekiguchi, A.; Kinjo, R.; Ichinohe, M. A Stable Compound
Element Alkyne Analogues. Organometallics 2007, 26, 4362−4372. Containing a Silicon-Silicon Triple Bond. Science 2004, 305, 1755−
(290) Sasamori, T.; Tokitoh, N. Doubly Bonded Systems between 1757.
Heavier Group 15 Elements. Dalton Trans 2008, 0, 1395−1408. (314) Phillips, A. D.; Wright, R. J.; Olmstead, M. M.; Power, P. P.
(291) Ottosson, H.; Eklof, A. M. Silenes: Connectors between
Synthesis and Characterization of 2,6-Dipp2-H3C6SnSnC6H3-2,6-
Classical Alkenes and Nonclassical Heavy Alkenes. Coord. Chem. Rev.
Dipp2 (Dipp = C6H3-2,6-Pri2): A Tin Analogue of an Alkyne. J. Am.
2008, 252, 1287−1314.
Chem. Soc. 2002, 124, 5930−5931.
(292) Scheschkewitz, D. Anionic Reagents with Silicon-Containing
(315) Pu, L.; Twamley, B.; Power, P. P. Synthesis and Character-
Double Bonds. Chem. - Eur. J. 2009, 15, 2476−2485.
ization of 2,6-Trip2H3C6PbPbC6H3-2,6-Trip2 (Trip = C6H2-2,4,6-i-
(293) Wang, Y.; Robinson, G. H. Unique Homonuclear Multiple
Pr3): A Stable Heavier Group 14 Element Analogue of an Alkyne. J.
Bonding in Main Group Compounds. Chem. Commun. 2009, 0, 5201−
Am. Chem. Soc. 2000, 122, 3524−3525.
5213.
(316) Chen, Y.; Hartmann, M.; Diedenhofen, M.; Frenking, G.
(294) Mizuhata, Y.; Sasamori, T.; Tokitoh, N. Stable Heavier Carbene
Analogues. Chem. Rev. 2009, 109, 3479−3511. Turning a Transition State into a MinimumThe Nature of the
(295) Kira, M. An Isolable Dialkylsilylene and its Derivatives. A Step Bonding in Diplumbylene Compounds RPbPbR (R = H, Ar). Angew.
toward Comprehension of Heavy Unsaturated Bonds. Chem. Commun. Chem., Int. Ed. 2001, 40, 2051−2055.
2010, 46, 2893−2903. (317) Dyker, C. A.; Lavallo, V.; Donnadieu, B.; Bertrand, G. Synthesis
(296) Jones, C. Bulky Guanidinates for the Stabilization of Low of an Extremely Bent Acyclic Allene (A “Carbodicarbene”): A Strong
Oxidation State Metallacycles. Coord. Chem. Rev. 2010, 254, 1273− Donor Ligand. Angew. Chem., Int. Ed. 2008, 47, 3206−3209.
1289. (318) Fürstner, A.; Alcarazo, M.; Goddard, R.; Lehmann, C. W.
(297) Fischer, R. C.; Power, P. P. π-Bonding and the Lone Pair Effect Coordination Chemistry of Ene-1,1-diamines and a Prototype
in Multiple Bonds Involving Heavier Main Group Elements: Develop- “Carbodicarbene. Angew. Chem., Int. Ed. 2008, 47, 3210−3214.
ments in the NewMillennium. Chem. Rev. 2010, 110, 3877−3923. (319) Pranckevicius, C.; Fan, L.; Stephan, D. W. Cyclic Bent Allene
(298) Trinquier, G.; Malrieu, J. P. Nonclassical Distortions at Multiple Hydrido-Carbonyl Complexes of Ruthenium: Highly Active Catalysts
Bonds. J. Am. Chem. Soc. 1987, 109, 5303−5315. for Hydrogenation of Olefins. J. Am. Chem. Soc. 2015, 137, 5582−5589.
(299) Malrieu, J. P.; Trinquier, G. Trans-bending at Double Bonds. (320) Hsu, Y.-C.; Shen, J.-S.; Lin, B.-C.; Chen, W.-C.; Chan, Y.-T.;
Occurrence and Extent. J. Am. Chem. Soc. 1989, 111, 5916−5921. Ching, W.-M.; Yap, G. P. A.; Hsu, C.-P.; Ong, T.-G. Synthesis and
(300) Carter, E. A.; Goddard, W. A. Relation between Singlet-Triplet Isolation of an Acyclic Tridentate Bis(pyridine)carbodicarbene and
Gaps and Bond Energies. J. Phys. Chem. 1986, 90, 998−1001. Studies on Its Structural Implications and Reactivities. Angew. Chem.,
(301) For a discussion of the bonding model, see: Driess, M.; Int. Ed. 2015, 54, 2420−2424.
Grützmacher, H. Main GroupElement Analogues of Carbenes, Olefins, (321) Goldfogel, M. J.; Roberts, C. C.; Meek, S. J. Intermolecular
and Small Rings. Angew. Chem., Int. Ed. Engl. 1996, 35, 828−856. Hydroamination of 1, 3-dienes Catalyzed by bis (phosphine)
(302) Wilson, R. J.; Lichtenberger, N.; Weinert, B.; Dehnen, S. Carbodicarbene−Rhodium Complexes. J. Am. Chem. Soc. 2014, 136,
Intermetalloid and Heterometallic Clusters Combining p-Block 6227−6230.
(Semi)Metals with d- or f-block Metals. Chem. Rev. 2019, 119, (322) Roberts, C. C.; Matías, D. M.; Goldfogel, M. J.; Meek, S. J. Lewis
DOI: 10.1021/acs.chemrev.8b00658 Acid Activation of Carbodicarbene Catalysts for Rh-Catalyzed
(303) Frenking, G., Hermann, M. Gilbert Lewis and the Model of Hydroarylation of Dienes. J. Am. Chem. Soc. 2015, 137, 6488−6491.
Dative Bonding. In The Chemical Bond I. Structure and Bonding; Mingos, (323) Chen, W.-C.; Shen, J.-S.; Jurca, T.; Peng, C.-J.; Lin, y.-H.; Wang,
D., Ed.; Springer: Cham, 2016; Vol. 169. Y.-P.; Shih, W.-C.; Yap, G. P. A.; Ong, T.-G. Expanding the Ligand
(304) Lein, M.; Krapp, A.; Frenking, G. Why Do the Heavy-Atom Framework Diversity of Carbodicarbenes and Direct Detection of
Analogues of Acetylene E2H2 (E = Si−Pb) Exhibit Unusual Structures? Boron Activation in the Methylation of Amines with CO2. Angew.
J. Am. Chem. Soc. 2005, 127, 6290−6299. Chem., Int. Ed. 2015, 54, 15207−15212.

8842 DOI: 10.1021/acs.chemrev.8b00722


Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

(324) Chen, W.-C.; Shih, W.-C.; Jurca, T.; Zhao, L.; Andrada, D. M.; (344) Gagliardi, L.; Pyykko, P. Cesium and Barium as Honorary d
Peng, C.-J.; Chang, C.-C.; Liu, S.-k.; Wang, Y.-P.; Wen, Y.-S.; Yap, G. P. Elements: CsN7Ba as an Example. Theor. Chem. Acc. 2003, 110, 205−
A.; Hsu, C.-P.; Frenking, G.; Ong, T.-G. Carbodicarbenes: Unexpected 210.
π-Accepting Ability during Reactivity with Small Molecules. J. Am. (345) Jin, J.; Yang, T.; Xin, K.; Wang, G.; Wang, X.; Zhou, M.;
Chem. Soc. 2017, 139, 12830−12836. Frenking, G. Octacarbonyl Anion Complexes of Group 3 Transition
(325) Takagi, N.; Shimizu, T.; Frenking, G. Divalent Silicon(0) Metals [TM(CO)8]− (TM = Sc, Y, La) and the 18-Electron Rule.
Compounds. Chem. - Eur. J. 2009, 15, 3448−3456. Angew. Chem., Int. Ed. 2018, 57, 6236−6241.
(326) Takagi, N.; Shimizu, T.; Frenking, G. Divalent E(0) (346) Dyall, K. G.; Faegri, K., Jr. Introduction to Relativistic Quantum
Compounds (E = Si−Sn). Chem. - Eur. J. 2009, 15, 8593−8604. Chemistry; Oxford University Press: Oxford, 2007.
(327) Mondal, K. C.; Roesky, H. W.; Schwarzer, M. C.; Frenking, G.; (347) Wolf, A.; Reiher, M. Relativistic Quantum Chemistry: The
Niepötter, B.; Wolf, H.; Herbst-Irmer, R.; Stalke, D. A Stable Singlet Fundamental Theory of Molecular Science; Wiley-VCH: Weinheim,
Biradicaloid Siladicarbene: (L:)2Si. Angew. Chem., Int. Ed. 2013, 52, 2009.
2963−2967. (348) Schwerdtfeger, P. Relativistic Electronic Structure Theory. Part 1:
(328) Xiong, Y.; Yao, S.; Inoue, S.; Epping, J. D.; Driess, M. A Cyclic Fundamentals; Elsevier: Amsterdam, 2002.
Silylone (“Siladicarbene”) with an Electron-Rich Silicon(0) Atom. (349) Pyykkö, P. The Physics behind Chemistry and the Periodic
Angew. Chem., Int. Ed. 2013, 52, 7147−7150. Table. Chem. Rev. 2012, 112, 371−384.
(329) Kuwabara, T.; Nakada, M.; Hamada, J.; Guo, J. D.; Nagase, S.; (350) Autschbach, J. Perspective: Relativistic effects. J. Chem. Phys.
Saito, M. (η4-Butadiene)Sn(0) Complexes: A New Approach for Zero- 2012, 136, 150902.
Valent p-Block Elements Utilizing a Butadiene as a 4π-Electron Donor. (351) Peng, D.; Reiher, M. Exact Decoupling of the Relativistic Fock
J. Am. Chem. Soc. 2016, 138, 11378−11382. Operator. Theor. Chem. Acc. 2012, 131, 1081.
(330) Kinjo, R.; Donnadieu, B.; Celik, M. A.; Frenking, G.; Bertrand, (352) Liu, W. Advances in Relativistic Molecular Quantum
G. Synthesis and Characterization of a Neutral Tricoordinate Mechanics. Phys. Rep. 2014, 537, 59−89.
Organoboron Isoelectronic with Amines. Science 2011, 333, 610−613. (353) Cremer, D.; Zou, W.; Filatov, M. Dirac-Exact Relativistic
(331) Kong, L.; Li, Y.; Ganguly, R.; Vidovic, D.; Kinjo, R. Isolation of a Methods: The Normalized Elimination of the Small Component
Bis(oxazol-2-ylidene)−Phenylborylene Adduct and its Reactivity as a Method. WIREs Comput. Mol. Sci. 2014, 4, 436−467.
Boron-Centered Nucleophile. Angew. Chem., Int. Ed. 2014, 53, 9280− (354) Ilias, M.; Kellö, V.; Urban, M. Relativistic Effects in Atomic and
9283. Molecular Properties. Acta Phys. Slov. 2010, 60, 259−391.
(332) Braunschweig, H.; Dewhurst, R. D.; Hupp, F.; Nutz, M.; (355) Dirac, P. A. M. The Quantum Theory of the Electron. Proc. R.
Radacki, K.; Tate, C. W.; Vargas, A.; Ye, Q. Multiple Complexation of Soc. London, Ser. A 1928, 117, 610−624.
CO and Related Ligands to a Main-Group Element. Nature 2015, 522, (356) Anderson, C. D. The Positive Electron. Phys. Rev. 1933, 43, 491.
327−330. (357) Liu, W. Perspectives of Relativistic Quantum Chemistry: The
(333) Bernhardi, I.; Drews, T.; Seppelt, K. Isolation and Structure of Negative Energy Cat Smiles. Phys. Chem. Chem. Phys. 2012, 14, 35−48.
the OCNCO+ Ion. Angew. Chem., Int. Ed. 1999, 38, 2232−2233. (358) Saue, T. Relativistic Hamiltonians for Chemistry: A Primer.
(334) Mondal, K. C.; Roy, S.; Dittrich, B.; Andrada, D. M.; Frenking, ChemPhysChem 2011, 12, 3077−3094.
G.; Roesky, H. W. A Triatomic Silicon(0) Cluster Stabilized by a Cyclic (359) Dyall, K. G. Interfacing relativistic and nonrelativistic methods.
Alkyl(amino) Carbene. Angew. Chem,. Int. Ed. 2016, 55, 3158−3161. I. Normalized elimination of the small component in the modified
(335) Jin, J.; Wang, G.; Zhou, M.; Andrada, D. M.; Hermann, M.; Dirac equation. J. Chem. Phys. 1997, 106, 9618.
Frenking, G. The [B3(NN)3]+ and [B3(CO)3]+ Complexes Featuring (360) Dyall, K. G. A systematic sequence of relativistic approx-
the Smallest π-Aromatic Species B3+. Angew. Chem., Int. Ed. 2016, 55, imations. J. Comput. Chem. 2002, 23, 786.
2078−2082. (361) Iliaš, M.; Saue, T. An Infinite-Order Two-Component
(336) Ghadwal, R. S.; Roesky, H. W.; Merkel, S.; Henn, J.; Stalke, D. Relativistic Hamiltonian by a Simple One-Step Transformation. J.
Lewis Base Stabilized Dichlorosilylene. Angew. Chem., Int. Ed. 2009, 48, Chem. Phys. 2007, 126, 064102.
5683−5686. (362) Sikkema, J.; Visscher, L.; Saue, T.; Iliaš, M. The Molecular
(337) Mondal, K. C.; Roesky, H. W.; Schwarzer, M. C.; Frenking, G.; Mean-Field Approach For Correlated Relativistic Calculations. J. Chem.
Tkach, I.; Wolf, H.; Kratzert, D.; Herbst-Irmer, R.; Niepötter, B.; Stalke, Phys. 2009, 131, 124116.
D. Conversion of a Singlet Silylene to a Stable Biradical. Angew. Chem., (363) Nakajima, T.; Hirao, K. The Douglas-Kroll-Hess Approach.
Int. Ed. 2013, 52, 1801−1805. Chem. Rev. 2012, 112, 385−402.
(338) Reinmuth, M.; Neuhäuser, C.; Walter, P.; Enders, M.; Kaifer, E.; (364) van Lenthe, E.; Baerends, E. J.; Snijders, J. G. Relativistic Total
Himmel, H.-J. The Flexible Coordination Modes of Guanidine Ligands Energy Using Regular Approximations. J. Chem. Phys. 1994, 101,
in Zn Alkyl and Halide Complexes: Chances for Catalysis. Eur. J. Inorg. 9783−9792.
Chem. 2011, 2011, 83−90. (365) Schwerdtfeger, P. The Pseudopotential Approximation in
(339) von Szentpaly, L.; Schwerdtfeger, P. Which double-octet ABC Electronic Structure Theory. ChemPhysChem 2011, 12, 3143−3155.
molecules are bent? CI calculations on CaF2, and a softness criterion to (366) Dolg, M.; Cao, X. Relativistic Pseudopotentials: Their
predict bending. Chem. Phys. Lett. 1990, 170, 555−560. Development and Scope of Applications. Chem. Rev. 2012, 112,
(340) Kaupp, M.; Schleyer, P. v. R.; Stoll, H.; Preuss, H. The question 403−480.
of bending of the alkaline earth dihalides MX2 (M= beryllium, (367) Hellmann, H. A. New Approximation Method in the Problem of
magnesium, calcium, strontium, barium; X= fluorine, chlorine, Many Electrons. J. Chem. Phys. 1935, 3, 61.
bromine, iodine). An ab initio pseudopotential study. J. Am. Chem. (368) Pyykkö, P. Relativistic Effects in Structural Chemistry. Chem.
Soc. 1991, 113, 6012−6020. Rev. 1988, 88, 563−594.
(341) Dewar, M. J. S. A Review of π Complex Theory. Bull. Soc. Chim. (369) Desclaux, J. P.; Pyykkö , P. Dirac-Fock One-Centre Cl
Fr. 1951, 18, C79. calculations. The Molecules CuH, AgH and AuH Including p-type
(342) Chatt, J.; Duncanson, L. A. 586. Olefin Co-ordination Symmetry Functions. Chem. Phys. Lett. 1976, 39, 300−303.
Compounds. Part III. Infra-red Spectra and Structure: Attempted (370) Pyykkö, P. Relativistic Quantum Chemistry. Adv. Quantum
Preparation of Acetylene Complexes. J. Chem. Soc. 1953, 2939−2947. Chem. 1978, 11, 353−409.
(343) Wu, X.; Zhao, L.; Jiang, D.; Fernández, I.; Berger, R.; Zhou, M.; (371) Pitzer, K. S. Relativistic Effects on Chemical Properties. Acc.
Frenking, G. Barium as Honorary Transition Metal in Action: Chem. Res. 1979, 12, 271−276.
Experimental and Theoretical Study of [Ba(CO)]+ and [Ba(CO)]−. (372) Schwerdtfeger, P. Relativistic Electronic Structure Theory. Part 2:
Angew. Chem., Int. Ed. 2018, 57, 3974−3980. Applications; Elsevier: Amsterdam, 2004.

8843 DOI: 10.1021/acs.chemrev.8b00722


Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

(373) Hess, B. A. Relativistic Effects in Heavy-Element Chemistry. (395) Fossgaard, O.; Gropen, O.; Eliav, E.; Saue, T. Bonding in the
Ber. Bunsenges. Phys. Chem. 1997, 101, 1−10. Homologous Series CsAu, CsAg, and CsCu Studied at the 4-
(374) Pyykkö, P. Relativistic Effects in Chemistry: More Common Component Density Functional Theory and Coupled Cluster Levels.
Than You Thought. Annu. Rev. Phys. Chem. 2012, 63, 45−64. J. Chem. Phys. 2003, 119, 9355−9363.
(375) Schwarz, W. H. E.; Schwerdtfeger, P.; Snijders, J. G.; Baerends, (396) Belpassi, L.; Tarantelli, F.; Sgamellotti, A.; Quiney, H. M. The
E. J. Relativistic Atomic Orbital Contractions and Expansions: Electronic Structure of Alkali Aurides. A Four-Component Dirac−
Magnitudes and Explanations. J. Phys. B: At., Mol. Opt. Phys. 1990, Kohn−Sham Study. J. Phys. Chem. A 2006, 110, 4543−4554.
23, 3225−3240. (397) Pyykkö, P. Relativity, Gold, Closed-Shell Interactions, and
(376) Schwarz, W. H. E.; van Wezenbeek, E. M.; Baerends, E. J.; CsAu·NH3. Angew. Chem., Int. Ed. 2002, 41, 3573−3578.
Snijders, J. G. The Origin of Relativistic Effects of Atomic Orbitals. J. (398) Schwerdtfeger, P. Relativistic and Electron Correlation
Phys. B: At. Mol. Opt. Phys. 1989, 22, 1515−1530. Contributions in Atomic and Molecular Properties. Benchmark
(377) Burke, V. M.; Grant, I. P. The effect of relativity on atomic wave Calculations on Au and Au2. Chem. Phys. Lett. 1991, 183, 457−463.
functions. Proc. Phys. Soc., London 1967, 90, 297−314. (399) Jansen, M. The Chemistry of Gold as an Anion. Chem. Soc. Rev.
(378) Autschbach, J.; Siekierski, S.; Seth, M.; Schwerdtfeger, P.; 2008, 37, 1826−1835.
Schwarz, W. H. E. Dependence of Relativistic Effects on Electronic (400) Lagowski, J. J. Liquid Ammonia. Synth. React. Synth. React.
Configuration in the Neutral Atoms of d- And f-Block Elements. J. Inorg., Met.-Org., Nano-Met. Chem. 2007, 37, 115−153.
Comput. Chem. 2002, 23, 804−813. (401) Mudring, A.-V.; Jansen, M.; Daniels, J.; Kramer, S.; Mehring,
(379) Moore, C. E. Atomic Energy Levels (Natl. Bur. Stand. U.S. Circ. M.; Prates Ramalho, J. P.; Romero, A. H.; Parinello, M. Cesiumauride
No. 467, U.S. GPO Washington, D.C., 1958). Ammonia (1/1), CsAu.NH3: A Crystalline Analogue to Alkali Metals
(380) Buenker, R. J.; Alekseyev, A. B.; Liebermann, H.-P.; Lingott, R.; Dissolved in Ammonia? Angew. Chem., Int. Ed. 2002, 41, 120−124.
Hirsch, G. Comparison of Spin-Orbit Configuration Interaction (402) Wesendrup, R.; Schwerdtfeger, P. Extremely Strong s2 − s2
Methods Employing Relativistic Effective Core Potentials for the Closed-Shell Interactions. Angew. Chem., Int. Ed. 2000, 39, 907−910.
Calculation of Zero-Field Splittings of Heavy Atoms with a 2p0 Ground (403) Sidgwick, N. V. The Electronic Theory of Valency; Clarendon:
State. J. Chem. Phys. 1998, 108, 3400−3408. Oxford, 1927; pp 178−181.
(381) Desclaux, J. P. Relativistic Dirac-Fock Expectation Values for (404) Sidgwick, N. V. Some Physical Properties of the Covalent Link in
Atoms with Z = 1 to Z = 120. At. Data Nucl. Data Tables 1973, 12, 311− Chemistry; Cornell University Press: Ithaca, NY, 1933.
406. (405) Sidgwick, N. V.; Wardlaw, W.; Whytlaw-Gray, R. Inorganic
(382) Schwerdtfeger, P.; Lein, M. Theoretical Chemistry of Gold. In Chemistry. Annu. Rep. Prog. Chem. 1933, 30, 82−132.
Gold Chemistry. Current trends and future directions; Mohr, F., Ed.; (406) Schwerdtfeger, P.; Heath, G. A.; Dolg, M.; Bennett, M. A. Low
Valencies and Periodic Trends in Heavy Element Chemistry. A
Wiley-VCH: Weinheim, 2009; pp183−247.
Theoretical Study of Relativistic and Correlation Effects in Group 13
(383) Schwarz, W. H. E. An Introduction to Relativistic Quantum
and Period 6 Hydrides and Fluorides. J. Am. Chem. Soc. 1992, 114,
Chemistry. In Relativistic Methods for Chemists; Barysz, M., Ishikawa,
7518−7527.
M., Eds.; Springer: Dordrecht, 2010; pp 1−62.
(407) Schwerdtfeger, P.; Ischtwan, J. Theoretical Investigations on
(384) Jerabek, P.; Schuetrumpf, B.; Schwerdtfeger, P.; Nazarewicz, W.
Thallium Halides: Relativistic and Electron Correlation Effects in TlX
Electron and Nucleon Localization Functions in Superheavy Elements.
and TlX3 Compounds (X = F, Cl, Br, and I). J. Comput. Chem. 1993, 14,
Phys. Rev. Lett. 2018, 120, 053001.
913−921.
(385) Opalka, D.; Segado, M.; Poluyanov, L. V.; Domcke, W.
(408) Tebbe, K. F.; Georgy, U. Die Kristallstrukturen von
Relativistic Jahn-Teller Effect in Tetrahedral Systems. Phys. Rev. A: At., Rubidiiumtriodid und Thalliumtriiodid. Acta Crystallogr., Sect. C:
Mol., Opt. Phys. 2010, 81, 042501. Cryst. Struct. Commun. 1986, C42, 1675−1678.
(386) Bae, C.; Han, Y.-K.; Lee, Y. S. Spin−Orbit and Relativistic (409) Thanthiriwatte, K. S.; Vasiliu, M.; Battey, S. R.; Lu, Q.;
Effects on Structures and Stabilities of Group 17 Fluorides EF3 (E = I, Peterson, K. A.; Andrews, L.; Dixon, D. A. Gas Phase Properties of MX2
At, and Element 117): Relativity Induced Stability for the D3h Structure and MX4 (X = F, Cl) for M = Group 4, Group 14, Cerium, and
of (117)F3. J. Phys. Chem. A 2003, 107, 852−858. Thorium. J. Phys. Chem. A 2015, 119, 5790−5803.
(387) Wang, S.; Liu, W.; Schwarz, W. H. E. On Relativity, Bonding, (410) Ahuja, R.; Blomqvist, A.; Larsson, P.; Pyykko, P.; Zaleski-
and Valence Electron Distribution. J. Phys. Chem. A 2002, 106, 795− Ejgierd, P. Relativity and the Lead-Acid Battery. Phys. Rev. Lett. 2011,
803. 106, 018301.
(388) Balasubramanian, K. Relativity and Chemical Bonding. J. Phys. (411) Ba̧stŭg, T.; Rashid, K.; Sepp, W.-D.; Kolb, D.; Fricke, B. All-
Chem. 1989, 93, 6585−6596. Electron Xα Self-Consistent-Field Calculations of Relativistic Effects in
(389) Thayer, J. S. Relativistic Effects and the Chemistry of Heavy the Molecular Properties of Tl2, Pb2, and Bi2 Molecules. Phys. Rev. A
Main Group Elements. In Relativistic Methods for Chemists; Barysz, M., 1997, 55, 1760−1764.
Ishikawa, Y., Eds.; Springer; Dordrecht, 2010; pp63−98. (412) Lee, H.-S.; Han, Y.-K.; Kim, M. C.; Bae, C.; Lee, Y. S. Spin-Orbit
(390) Lim, I. S.; Schwerdtfeger, P. Four-Component and Scalar Effects Calculated by Two-Component Coupled-Cluster Methods:
Relativistic Douglas-Kroll Calculations for Static Dipole Polarizabilities Test Calculations on AuH, Au2, TlH and Tl2. Chem. Phys. Lett. 1998,
of the Alkaline-Earth Elements and Their Ions from Can to Ran (n = 0, + 293, 97−102.
1, + 2). Phys. Rev. A: At., Mol., Opt. Phys. 2004, 70, 062501. (413) Mayer, M.; Krüger, S.; Rösch, N. A Two-Component Variant of
(391) Türler, A.; Pershina, V. Advances in the Production and the Douglas−Kroll Relativistic Linear Combination of Gaussian-Type
Chemistry of the Heaviest Elements. Chem. Rev. 2013, 113, 1237− Orbitals Density-Functional Method: Spin−Orbit Effects in Atoms and
1312. Diatomics. J. Chem. Phys. 2001, 115, 4411−4423.
(392) Schwerdtfeger, P.; Pašteka, L. F.; Punnett, A.; Bowman, P. O. (414) Balasubramanian, K. Spectroscopic Constants and Potential-
Relativistic and Quantum Electrodynamic Effects in Superheavy Energy Curves of Heavy P-Block Dimers and Trimers. Chem. Rev. 1990,
Elements. Nucl. Phys. A 2015, 944, 551−577. 90, 93−167.
(393) Schwerdtfeger, P.; Seth, M. Relativistic Effects of the (415) Froben, F. W.; Schulze, W.; Kloss, H. Raman Spectra of Matrix-
Superheavy Elements. In Encyclopedia of Computational Chemistry; Isolated Group IIIA Dimers: Ga2, In2, Tl2. Chem. Phys. Lett. 1983, 99,
Schleyer, P. v. R.; Schreiner, P. R.; Allinger, N. L.; Clark, T.; Gasteiger, 500−502.
J.; Kollman, P. A.; Schaefer, H. F., III, Eds.; John Wiley & Sons, Ltd, (416) Christiansen, P. A.; Pitzer, K. S. Electronic Structure And
1998; Vol.4, pp 2480−2499. Dissociation Curves for the Ground States of Tl2 and Tl2+ from
(394) Pyykkö, P.; Desclaux, J. P. Relativity and the Periodic System of Relativistic Effective Potential Calculations. J. Chem. Phys. 1981, 74,
Elements. Acc. Chem. Res. 1979, 12, 276−281. 1162−1165.

8844 DOI: 10.1021/acs.chemrev.8b00722


Chem. Rev. 2019, 119, 8781−8845
Chemical Reviews Review

(417) Han, Y.-K.; Hirao, K. On the Ground-State Spectroscopic (439) Schwerdtfeger, P. On the anomaly of the metal-carbon bond
Constants of Tl2. J. Chem. Phys. 2000, 112, 9353−9355. strength in (CH3)2M compounds of the heavy elements M = Au−, Hg,
(418) Weston, L.; Pownall, B. T.; Mair, F. S.; McDouall, J. J. W. Tl+ and Pb2+. Relativistic effects in metal-ligand force constants. J. Am.
Thallophilic Tl(I)−Tl(I) Contacts Mediated by Tl−aryl Interactions. Chem. Soc. 1990, 112, 2818−2820.
A Computational Study. Dalton Trans. 2016, 45, 8433−8439. (440) Schwerdtfeger, P.; Boyd, P. D. W.; Brienne, S.; McFeaters, J. S.;
(419) Schumann, H.; Janiak, C.; Pickardt, J.; Bö r ner, U. Dolg, M.; Liao, M. S.; Schwarz, W. H. E. The Mercury - Mercury Bond
Pentabenzylcyclopentadienylthallium(I): Synthesis and Structure of a in Inorganic and Organometallic Compounds. A Theoretical Study.
“Dimeric” Organothallium Compound with Tl-Tl Interaction. Angew. Inorg. Chim. Acta 1993, 213, 233−246.
Chem., Int. Ed. Engl. 1987, 26, 789−790. (441) Autschbach, J. Perspective: Relativistic effects. J. Chem. Phys.
(420) Schwerdtfeger, P. Metal-Metal Bonds in Thallium (I)-Thallium 2012, 136, 150902.
(I) Compounds: Fact or Fiction? Inorg. Chem. 1991, 30, 1660−1663. (442) Almlöf, J.; Gropen, O. Relativistic Effects in Chemistry. Rev.
(421) Hermann, A.; Furthmüller, J.; Gäggeler, H.; Schwerdtfeger, P. Comput. Chem. 2007, 8, 203−244.
Spin-Orbit Effects In Structural And Electronic Properties For The (443) Schwarz, H. Relativistic Effects in Gas-Phase Ion Chemistry: An
Solid-State Of The Group 14 Elements From Carbon To Superheavy Experimentalist’s View. Angew. Chem., Int. Ed. 2003, 42, 4442−4454.
Element. 114. Phys. Rev. B: Condens. Matter Mater. Phys. 2010, 82, (444) Plato. Rouse, W. H. D., Ed. The Republic Book VII; Penguin
155116. Group Inc.: New York, 2008; pp 365−401.
(422) Kraka, E.; Zou, W.; Freindorf, M.; Cremer, D. Energetics and (445) Frenking, G. Covalent Bonding and Charge Shift Bonds.
Mechanism of the Hydrogenation of XHn for Group IV to Group VII Comment on ‘The Carbon−Nitrogen Bonds in Ammonium Com-
Elements X. J. Chem. Theory Comput. 2012, 8, 4931−4943. pounds Are Charge Shift Bonds’. Chem. - Eur. J. 2017, 23, 18320−
(423) Filatov, M.; Zou, W.; Cremer, D. Spin-orbit coupling 18324.
calculations with the two-component normalized elimination of the (446) Chen, P.; Gershoni-Poranne, R. Response to ‘Covalent Bonding
small component method. J. Chem. Phys. 2013, 139, 014106. and Charge Shift Bonds. Comment on ‘The Carbon−Nitrogen Bonds
(424) Giuliani, S. A.; Matheson, Z.; Nazarewicz, W.; Olsen, E.; in Ammonium Compounds Are Charge Shift Bonds’. Chem. - Eur. J.
Reihardt, P. − G.; Sadhukhan, J.; Schuetrumpf, B.; Schunck, N.; 2017, 23, 18325−18329.
Schwerdtfeger, P. Colloquium: Superheavy Elements: Oganesson and (447) Schmidbaur, H. Réplique: A New Concept for Bonding in
Beyond. Rev. Mod. Phys. 2019, 91, 011001. Carbodiphosphoranes? Angew. Chem., Int. Ed. 2007, 46, 2984−2985.
(425) Schädel, M. Chemistry of Superheavy Elements. Angew. Chem., (448) Frenking, G.; Neumüller, B.; Petz, W.; Tonner, R.; Ö xler, F.
Int. Ed. 2006, 45, 368−401. Reply to ‘Réplique: A New Concept for Bonding in Carbodiphosphor-
(426) Seth, M.; Schwerdtfeger, P.; Faegri, K. The Chemistry of the anes? Angew. Chem., Int. Ed. 2007, 46, 2986−2987.
Superheavy Elements III. Theoretical Studies on Element 113
Compounds. J. Chem. Phys. 1999, 111, 6422−6433.
(427) Seth, M.; Faegri, K.; Schwerdtfeger, P. The Stability of the
Oxidation State + 4 in Group 14 Compounds from Carbon to Element
114. Angew. Chem., Int. Ed. 1998, 37, 2493−2496.
(428) Eliav, E.; Fritzsche, S.; Kaldor, U. Electronic Structure Theory
of the Superheavy Elements. Nucl. Phys. A 2015, 944, 518−550.
(429) Pitzer, K. S. Are Elements 112, 114, and 118 Relatively Inert
Gases? J. Chem. Phys. 1975, 63, 1032−1033.
(430) Mitin, A. V.; van Wüllen, C. Two-component Relativistic
Density-Functional Calculations of the Dimers of the Halogens from
Bromine Through Element 117 Using Effective Core Potential and All-
Electron Methods. J. Chem. Phys. 2006, 124, 064305.
(431) Schwerdtfeger, P. Toward an Accurate Description of Solid-
State Properties of Superheavy Elements. A Case Study for the Element
Og (Z = 118). EPJ Web Conf. 2016, 131, 07004.
(432) Eliav, E.; Kaldor, U.; Ishikawa, Y.; Pyykkö, P. Element 118: The
First Rare Gas with an Electron Affinity. Phys. Rev. Lett. 1996, 77,
5350−5352.
(433) Goidenko, I.; Labzowsky, L.; Eliav, E.; Kaldor, U.; Pyykkö, P.
QED Corrections to the Binding Energy of the Eka-Radon (Z = 118)
Negative Ion. Phys. Rev. A: At., Mol., Opt. Phys. 2003, 67, 020102.
(434) Gong, S.; Wu, W.; Wang, F. Q.; Liu, J.; Zhao, Y.; Shen, Y.; Wang,
S.; Sun, Q.; Wang, Q. Classifying superheavy elements by machine
learning. Phys. Rev. A 2019, 99, 022110.
(435) Nash, C. S.; Bursten, B. E. Spin-Orbit Coupling versus the
VSEPR Method: On the Possibility of a Nonplanar Structure for the
Super-Heavy Noble Gas Tetrafluoride (118)F4. Angew. Chem., Int. Ed.
1999, 38, 151−154.
(436) Heinemann, C.; Schwarz, H.; Koch, W.; et al. Relativistic effects
in the cationic platinum carbene PtCH2+. J. Chem. Phys. 1996, 104,
4642.
(437) (b) Rakowitz, F.; Marian, C. B.; Schimmelpfennig, B. Ground
and excited states of PtCH2+: assessment of the no-pair Douglas−Kroll
ab initio model potential method. Phys. Chem. Chem. Phys. 2000, 2,
2481−2488.
(438) Demissie, T. B.; Garabato, B. D.; Ruud, K.; Kozlowski, P. M.
Mercury Methylation by Cobalt Corrinoids: Relativistic Effects Dictate
the Reaction Mechanism. Angew. Chem., Int. Ed. 2016, 55, 11503−
11506.

8845 DOI: 10.1021/acs.chemrev.8b00722


Chem. Rev. 2019, 119, 8781−8845

You might also like