Anda di halaman 1dari 8

review article

Published online: 23 septemBER 2009|doi: 10.1038/nchem.359

Ultrashort metalmetal distances and extreme bond orders


Frank R. Wagner1*, Awal Noor2 and Rhett Kempe2*
Chemical bonding is at the very heart of chemistry. Although main-group-element EE bond orders range up to triple bonds, higher formal bond orders are known between transition metals. Here we review recent developments related to the synthesis of formally quintuply bonded transition metals in coordination compounds, and their theoretical description. The quadruple bond fascinated chemists for about 40 years. Recently, a stable molecule containing a formal quintuple bond initiated a renaissance in synthesizing and understanding bonds with high bond orders. Ultrashort metalmetal distances as low as 1.73 are one of the outcomes. First results indicate that the relevance of these bimetallic platforms to synthetic chemistry can be addressed through quintuple-bond reactivity studies. The theoretical description of the bonding situation in molecules with extreme bond orders has only just begun.

hemical bonding, in particular the nature of a chemical bond, the electronic structure and its reactivity, is of fundamental interest 1. Reactivity expresses the significance of reaction pathways (it is a projection of reaction rates) and the electronic structure allows us to understand why some of the reaction pathways a bond can take are relevant while others are not. Chemical bonding is the basis of chemistry, the science of synthesis. It is more than 90years ago that Lewis introduced the idea of electron pairing and sharing between neighbouring atoms2, in the first model of covalent bonding (see ref. 3 for a summary of Lewiss work). Remarkably, this simple model is still the basis of teaching in school and academia, and lies behind the bonding dash we draw between two atoms to describe the structure of a compound. A little more than ten years later, Heitler and London provided the first theoretical evidence supporting Lewiss model by showing that the energy of the HH bond is due to the resonance between the electrons as they exchange positions between the two atoms4,5. The question arises as to how many of such covalent bonds can be formed between two atoms (of the same kind). A maximum of three has been found between main-group elements, for instance in acetylene and dinitrogen, and a greater number, up to six, could be expected for bonds involving transition metals, owing to the participation of not just s and porbitals but s and dorbitals. Stable molecules with fourfold bonding have been known for a little more than 40 years and have been investigated experimentally and theoretically in detail since then6. In 2005, Power and co-workers provided the first experimental evidence of quintuple bonding in a stable molecule7. Here we summarize experimental and theoretical work related to quintuple bonding initiated by this work.

Bond orders and quantum mechanical calculations

Because bond order is a chemical concept, and not an observable in the quantum mechanical sense, there does not exist a unique definition of bond multiplicity in quantum chemistry. Thus, it is characteristic of the situation that in the recent papers on CrCr quintuple bonding, different methods of studying the bonding and, in particular, evaluating the bond order have been used. There follows a general, brief overview of the different concepts used in the analysis of the bond order of the molecules highlighted in this Review.
1

There is a common idea that the covalent bond is the result of electron sharing between two atoms8, as already proposed by Lewis2 in 1916. However, concerning the role of spin pairing itself, the direct interaction of the spins, in the sense of an interaction between magnetic dipoles, is entirely negligible9. Thus, the Lewis two-electron covalent bond is essentially the cumulative result of the effects of each electron being shared individually between two atoms (tempered, of course, by the effect of the interelectronic repulsion)8. From this it becomes clear that the primary goal of a definition of bond order is to give a measure of the number of electrons shared between two atoms. For this purpose, a definition of an atom within a polyatomic unit must be made, and unfortunately this is not possible in a unique and unambiguous quantum mechanical way (see, for example, ref.10). The common concepts used in defining an atom within a polyatomic unit can be divided into Hilbert-space and position-space concepts. In Hilbert space, an atom is defined using its basis set and the engagement of it in the wavefunction. In position space, an atom is defined as an entity on the basis of some space-partitioning scheme. Two principally different schemes are used: non-overlapping, space-filling atomic units, the most prominent ones being the basins of electron density in Baders quantum theory of atoms in molecules (QTAIM)11; and fuzzy atoms with overlapping regions, the most prominent ones being the Hirshfeld atoms12. Both have been shown to possess a general physical meaning 10. Quantum chemical calculations are typically done in Hilbert space. Historically, there are two types of wavefunction representation, the molecular-orbital and valence-bond types, each of which influences how chemical bonding is considered in a specific way. Although at present the representations of the wavefunction can be at least approximately transformed into one another, they give rise to different types of bonding analysis in Hilbert space that are moreor-less specific to the wavefunction method13. The advantage of these procedures is that chemical bonding analysis can be done in terms of the immediate basic quantities used to generate the wavefunction, that is, basis sets, molecular orbitals and valence-bond structures. Notably, the position-space representation of chemical bonding, which is based on the one- and two-particle density matrices, is essentially independent of the type of wavefunction used. These matrices contain the essential information about the physical particles under

Max-Plank-Institut fr Chemische Physik fester Stoffe, 01187 Dresden, Germany, 2Anorganische Chemie II, Universitt Bayreuth, 95440 Bayreuth, Germany. *e-mail: frank.wagner@cpfs.mpg.de; kempe@uni-bayreuth.de
nature chemistry | VOL 1 | OCTOBER 2009 | www.nature.com/naturechemistry

529

2009 Macmillan Publishers Limited. All rights reserved.

review article
consideration, that is, the electrons: the one-particle density matrix (the diagonal part of which is the electron density) describes the distribution of electrons and the two-particle density matrix describes the distribution of electron pairs. Ideally, the different methods of position-space analysis of chemical bonding each focus on specific aspects of the electronic behaviour without introducing new artificial objects (unicorns14). In this respect, they are not competing with each other, but synergistically construct a view on the chemical bond complementary to the various Hilbert-space views. Methods for theoretical bond-order analysis. In the very early days of molecular-orbital theory, even before it had been named, the first definition of the number of bonds between two atoms in a symmetric diatomic molecule was given by Herzberg 15: in its original form, it is equal to half the difference between the number of bonding electrons and the number of loosening electrons in the molecule. For conceptual reasons, this counting scheme was adopted shortly after also by Mulliken16, but he considered more fundamental a continuous conception of chemical bonding in which a nonintegral bonding power of either sign is attributable to every outer electron. The Mulliken population analysis (see below) developed 20years later contained such a quantity. Although Herzberg wrote in terms of electrons in his original paper 15, it was becoming more and more recognized that the objects obtained from certain electronic-structure methods were one-electron wavefunctions, which Mulliken chose to term orbitals17. From then on, in the framework of the developing molecular-orbital theory, bonding, non-bonding and antibonding orbitals were increasingly used to describe the bonding scenario in molecules. The Herzberg scheme of determining the bond order from molecular-orbital occupations is now taught in general chemistry textbooks and the resulting bond order is sometimes called the Herzberg number. The Herzberg number, defined on the basis of a monodeterminantal wavefunction, is now accepted as the molecular-orbital definition of the formal bond order. Also, for explicitly correlated wavefunctions, the antibonding molecular orbitals that were empty in the initial HartreeFock wavefunction become partially occupied. In the resulting natural orbitals, this happens at the expense of the occupation of the initially fully occupied bonding orbitals. Applying the Herzberg definition also in these situations generally leads to fractional bond orders. This possible generalization of the Herzberg definition was described and applied to a metalmetal multiple-bonding situation in 198718. Recently 19, the idea was revived and proposed as the definition of the effective bond order (EBO), which has been applied to characterize the bonding in diatomic molecules and in bimetallic complexes with formal metalmetal multiple bonding (see above). The Herzberg scheme is restricted to either diatomic molecules or to polyatomic molecules with strongly localized orbitals between a pair of atoms. The first definition of bond order in the framework of molecular-orbital theory applicable to delocalized orbitals was given by Coulson in terms of the Hckel model20. The application of the new definition to conjugated hydrocarbons resulted in fractional CC bond orders, for example 1.67 for benzene and 1.5 for graphite. Although not mentioned in ref.20, this yields total valences of 4.33 and 4.5 for the respective carbon atoms, which were certainly considered less chemical than the bond orders given by Pauling21 several years before. In the competing valence bond method of Heitler, London, Slater and Pauling, the electronic structure was considered to be a superposition of wavefunctions of Lewis-type structures with perfectly paired electrons. For quantitative purposes, the inclusion of ionic structures even for classical covalent bonds was soon discovered to be necessary to obtain reliable results. On the conceptual side, fractional bond orders were considered to arise as a result of resonating bonds. Three years before Coulsons20 molecular-orbital definition of the bond order, the resonance between the two (benzene) and three (graphite)
530

Nature chemistry doi: 10.1038/nchem.359


Kekul-type structures led Pauling 21 to assign a CC bond order of 1.5 to benzene and 1.33 to graphite, which yields a total valence of four for both types of carbon species. In this framework, Pauling set up an empirical function from four basic values expressing the dependence of the (experimentally known) CC interatomic distance on the assumed amount of single-bonddouble-bond resonance, that is, on the amount of double-bond character. From this function, the interatomic CC distances in other organic molecules were predicted and compared with experimental data. The idea of there being a simple relationship between the bond distance and the bond order is still in use today, also for heavier elements, despite the existence of various counter-examples. One interesting counter-example concerning TcTc multiple bonding in Tc2Cl8 units with variable charge is discussed in ref.6. On the methodological side, there is a connection between the old resonance concepts of chemical bonding and the modern idea of natural resonance theory 22 (NRT) presented in 1998 as an extension of the then-established natural-bond-orbital analysis23. The basic quantity of this method is the (possibly correlated) one-particle density matrix, which is considered to be constructed as a weighted superposition of one-particle density matrices from Lewis-type structures already including bond polarity. In practice, the generation of the structures and their weights from the given one-particle density matrix (typically obtained from a molecular-orbital calculation) occurs automatically by a complex process. For monodeterminantal wavefunctions, bond orders very close to the corresponding Herzberg numbers are obtained for prototype molecules H2, N2 and F2. Interestingly, a decomposition of the total bond order in terms of only additive covalent and electrovalent (ionic) contributions has been defined. In molecular-orbital theory, the Mulliken population analysis developed in the 1950s for wavefunctions calculated using linear combination of atomic orbitals (LCAO) has had a big impact (and still has) on the conceptual aspects of chemical bonding. It defines a decomposition of the molecular orbitals electronic population into net atomic populations and diatomic overlap populations24,25. Summed up over all occupied orbitals, the latter can be either positive or negative. As an indicator of covalent bonding between two atoms, Mulliken overlap populations characterize the accumulation of charge density in the region between the chemically bonded atoms, but they do not represent a bond order. The bond-order index was defined independently by Giambiagi et al.26 and Mayer 27 for HartreeFock-like electronicstructure methods. It was initially defined as a generalization of the Wiberg index 28, which has found widespread use in semi-empirical electronic-structure theory using an orthonormal basis set. With respect to the partitioning of the two-centre overlap density it is related to the Mulliken population analysis, and the whole method is therefore sometimes called MullikenMayer population analysis (MMA). It has been proven that under certain conditions the bondorder index of a homonuclear diatomic molecule exactly equals the Herzberg number 29. In actual applications, the values normally reproduce the classical bond orders for C2H6, C2H4, C2H2, N2 and F2, for example, very well. One big conceptual and practical drawback is its strong dependence on the basis set (inherited from the Mulliken type of partitioning), which affords the use of well-balanced, nottoo-diffuse atomic basis sets in practical applications. Despite this known drawback, the method is conceptually very interesting. The bond-order index between two atoms has been shown to be equal to the diatomic contribution to the integral of the exchange density, which is a part of the same-spin pair density. It measures the degree to which the fluctuations of the atomic populations from their average values in both atoms are correlated with each other 30. This means that the bond-order index represents a measure of the electronic delocalization between two atoms in Hilbert space. There is an important connection between the Hilbert-space MMA method and the position-space analysis of chemical bonding. The integrals of the exchange density within a single atomic domain
nature chemistry | VOL 1 | OCTOBER 2009 | www.nature.com/naturechemistry

2009 Macmillan Publishers Limited. All rights reserved.

Nature chemistry doi: 10.1038/nchem.359


and between a pair of atomic domains in position space had already been shown31 to measure the extent of electron localization in the single domain and the extent of delocalization between the pair of domains, respectively. The decision to choose the QTAIM basins as the basis for the position-space partitioning of the exchange-density integrals gave rise to the topological definition of a bond order 32 that represents the position-space counterpart of the MMA bond index. In contrast to its Hilbert-space analogue, the topological definition does not explicitly suffer from unbalanced and diffuse atomic basis sets. Unfortunately, the initially proposed extension of the topological bond-order index to the case of correlated wavefunctions lessens its physical content 31,33, and the question arises of whether this price is worth paying. To describe the situation more completely, as a continuation of the original work31 a so-called delocalization index between two QTAIM atoms was defined in a physically transparent way using the general exchange-correlation density 34. Both electronsharing indices, the topological bond order and the delocalization index, are identical at the monodeterminantal level, but they differ for correlated wavefunctions35. The calculation of the delocalization index at the correlated level is quite complicated and time consuming, and only a few studies have been published so far. The results for monodeterminantal wavefunctions are quite promising. For symmetric diatomic molecules such as, for example, H2, N2 and F2, the two definitions yield identical values close to the Herzberg number. For unsymmetrical ones such as, for example, LiH, the resulting values are much smaller than one, which is consistent with the significant ionic component of the bonding. Additionally, for polyatomic molecules it is found that the values between atoms A and B is significantly influenced by the other atoms connected to A or B, which leads to a reduction of the values relative to the Herzberg number 36. These observations point to the chemical interpretation (in addition to the clear physical interpretation of the delocalization index) of both of these electron-sharing indices being effective covalent bond orders.

review article
twice the energy (45eV)40. However, in the 1970s the * transition was uniquely identified as the 14,700cm1 (1.82eV) band and the low oscillator strength was explained to be due to poor -orbital overlap44. A similar energy for the * transition was theoretically obtained in a 2003 computational study 45 that employed an explicitly correlated treatment of the electronic structure in the ground and excited states using the complete-active-space method with additional second-order perturbative treatment of dynamical correlation (CASPT2), an advanced technique in transition-metal computational chemistry. The spectral features were obtained with good accuracy. ReRe bonding has been characterized by means of the effective bond order described above. An EBO of 3.2 has been obtained from the sum of the partial bond orders 0.92(), 1.74() and 0.54(). The strongly reduced -bond contribution (0.54) is consistent with the small * energy difference mentioned above, and does not invalidate the formal picture of the quadruple bonding. Owing to the established weakness of the bond, another reason for the eclipsed arrangement of the chlorine atoms, on which Cotton and co-workers put so much emphasis, has been sought. As the result of density functional theory (DFT) calculations in combination with coupled cluster calculations, hyperconjugation has been proposed46 to be the true reason for the observed rotomeric preference. The shortest metalmetal bond in coordination compounds for decades. Bond lengths of main-group elements become significantly shorter as bond order increases1. This trend should also occur in bonds involving transition-metal homobimetallic complexes. The shortest metalmetal bond in a stable molecule has been of interest to natural scientists, in particular chemists, for decades. The element chromium has an important role in the search for such a bond. Owing to their electron configurations, chromium and the higher homologues of group-6 metals can form metalmetal bonds of high formal bond orders. Within group6, chromium has the smallest ion radius and is ideally suited to form exceptionally short metalmetal bonds. For nearly 30 years, an aryl Cr(ii) compound, structurally characterized in ref.47 and first prepared by Hein and Tille more than 40years ago48, was thought to have the shortest experimentally
a
CrCr, 1.830(4) CrCr, 1.828(2)

Historical background

The first quadruple bond. About 120years after the first report 37 on a compound which is now known to contain a formal metalmetal quadruple bond, namely Cr(ii)2(O2CCH3)4(OH2)2, the existence of metalmetal quadruple bonding is supported by strong experimental and theoretical evidence. In 1964, on the basis of a careful chemical characterization of several synthesized dirhenium octahalide compounds38, the experimental crystal structure39 of K2[Re2Cl8]2H2O and simplified molecular-orbital calculations40, Cotton and co-workers presented the first strong evidence for the notion of metalmetal 242 quadruple bonding in [Re2Cl8]2 molecular units41. The -bond interpretation was strongly suggested by the observed eclipsed arrangement of the chlorine ligands and the simultaneous unusually short ReRe distance (2.24 ). This type of compound had in fact been obtained several years before42, and the experimental crystal structure with the correct Re2Cl8 units had been published43. However, owing to the erroneous assignment of the Re(ii) oxidation state during the characterization studies41, the bimetallic units had been formulated as [Re2Cl8]4, and additional H+ ions had been supposed to be present in the crystal structure. Mainly, though, the authors, like many of their colleagues, were not yet mentally prepared for the notion of metal metal multiple bonding. It was therefore Cotton and co-workers, working on similar compounds and being aware of these previous publications, who put the pieces together in the right way and created the prototype picture of a metalmetal quadruple bond. The whole story has been told in several places (see, for example, ref.6). To characterize the strength of the bond, the optical absorption spectrum of the alkylammonium compound (NR4)2[Re2Cl8] was analysed to find the energy of the * absorption band. Owing to its observed low oscillator strength, the 14,000cm1 (1.7eV) band was ruled out and the transition was attributed with more than
nature chemistry | VOL 1 | OCTOBER 2009 | www.nature.com/naturechemistry

2
O

Figure 1 | Quadruple bonding in chromium homobimetallic complexes. a, Molecular structures of the two coordination compounds that for decades held the record for the shortest metalmetal distance in stable molecules. b, Reversible cleavage of an unsupported CrCr multiple bond by pyridine (red,Cr; orange,C; blue,O; green,N; purple,Li; brown,Br).
531

2009 Macmillan Publishers Limited. All rights reserved.

review article

Nature chemistry doi: 10.1038/nchem.359


such as Cr2 trapped in inert matrices at low temperature54,55 generated by vaporization of the metals56 or as short-living intermediates through pulsed photolysis of chromium hexacarbonyl57, were the only existing evidence for molecules with formal bond orders higher than four 58. These species are ligand free and all valence electrons can be used to form multiple bonds. Designing a compound with bond orders higher than four involves the reduction of the coordination number of covalently bonded ligands to allow as many electrons as possible to participate in metalmetal bonding. Sterically demanding ligands such as the monovalent terphenyl ligand C6H3-2,6-[C6H3-2,6-(isopropyl)2]2 are able to provide these features59. The first quintuple bonding in a stable compound was observed by Power and co-workers in a chromium complex stabilized by this ligand (Fig.2)7. Those authors defined quintuple bonding as follows: The description quintuple bond is intended to indicate that five electron pairs play a role in holding the metal atoms together. It does not imply that the bond order is five or that the bonding is very strong, because the ground state of the molecule necessarily involves mixing of other higher-energy configurations with less bonding character. This gives lower, usually noninteger, bond orders. The dark-red crystalline compound studied in ref.7 is very air and moisture sensitive. The weak temperature-independent paramagnetism is consistent with an S=0 ground state and strongly coupled d 5d 5 bonding electrons. Surprisingly, the bimetallic compound is characterized by a metalmetal bond of 1.8351(4), which is long in comparison with the shortest formal quadruple CrCr bonds known at the time (Fig.1a)47,49. The secondary arene interaction has an important role in this compound. Quantum mechanical calculations, using both CASPT2 and DFT methods, were undertaken for chromium, iron and cobalt (model) complexes to provide a multireference description of the metal bond and to determine the extent of secondary metalarene interaction in 1 (ref.60). The studies indicate that the arene interaction in the chromium complexes causes only a slight weakening of the quintuple bond and that in the iron and cobalt complexes strong arene coordination precludes significant metalmetal bonding. We note that the quintuple-bond character in 1 has been predicted61 to be theoretically controversial because of the trans-bent structure of the CCrCrC entity, which differs from the translinear arrangement expected for a quintuple bond. Trans-bent geometries in quintuply bonded molecules have been proposed62 to originate in the preference of a strong bond from sdz2 hybridization. Bending does not destroy the bond: the hybridization scheme proposed even leads to the creation of a more favourable bond through sd 4 hybridization as well as a pure d orbital. Even more generally, a subsequent computational DFT study 63 on bending isomers of simple model compounds RCrCrR with various monodentate ligands reported a remarkable persistence of the qualitative quintuple-bond picture for the various bending geometries representing minima of the corresponding potential hypersurface. More specifically, at the level of CASPT2 calculations on a model compound PhCrCrPh (Ph, phenyl), only a tiny energy difference between the trans-linear and the trans-bent structure (but an appreciable energy barrier) was obtained64. The preference for the trans-bent structure in the experimental compound 1 was ascribed to secondary interactions of chromium with the true ligand. For the CrCr bonding in the model compound, an EBO value of 3.5 was calculated. Despite this, it has been emphasized that the formal bond order is still 5. A bond energy of 76 kcal mol1 was computed, which is about twice as large as for a Cr2 molecule with a formal sextuple bond. This counterintuitive result has been mainly attributed to the destabilizing 4s4s interactions in the dimer. From supplementary DFT calculations, certain bond-weakening interactions between chromium and the flanking aryl groups have been detected. It has been concluded
nature chemistry | VOL 1 | OCTOBER 2009 | www.nature.com/naturechemistry

Figure 2 | Molecular structure of the first stable compound in which quintuple bonding was observed. The aryl chromium dimer 1 has a CrCr distance of 1.8351(4) (colour code as in Fig.1). The metalmetal bond is supported by a metal arene interaction (thin line).

obtained metalmetal distance in a stable molecule, namely 1.830(4) (Fig.1a, left). Hein and Tille also proposed the dimeric structures for this compound, which were verified by structure determination. Interestingly, a little more than a month before the submission of ref. 47, tetrakis(2-methoxy-5-methylphenyl)dichromium (Fig. 1a, right), a homobimetallic Cr(ii) compound with a CrCr distance of essentially the same length, 1.828(2), was submitted for publication49. It appeared in print one issue after ref.47. Both publications are recognized by the community equally. Owing to the better availability of X-ray crystal structure analysis in inorganic laboratories, a great interest in synthesizing homobimetallic complexes with short metalmetal contacts evolved6. A critical review 50 of the role of the bridging ligands in alkoxo, amido and aryl Cr(ii) homobimetallics with regard to the formation of short CrCr quadruple bonds concluded that divalent chromium show a surprisingly passive behaviour compared with the dominance of the ligand. The Cr2 unit in such complexes seems to respond to the ligand by overlapping orbitals and coupling electrons. It seems almost chemically nonsensical for extremely short metalmetal distances to suggest the existence of stronger bonds or a higher degree of metal metal bond multiplicity. Very important in this regard are Cr(ii) atoms multiply bonded without a bridging ligand, that is, chromium homobimetallics with unsupported metalmetal bonds. Examples are rare. A metalmetal distance of 2.096(1) was observed for a tetramethyldibenzotetraaza[14]annulene derivative (Fig. 1b, left)51. This compound undergoes reversible metalmetal bond cleavage upon the addition of pyridine, forming a bispyridine adduct (Fig.1b). It was concluded that the energy of such CrCr multiple bonds are in the same range as the sum of the energies of the generated pyridine chromium bonds52. Dissolution of the pyridine adduct in tetrahydrofuran leads back to the dimer. These findings are indicative of the weakness of multiple bonding between Cr(ii) atoms. Very recently, it was shown53 that homobimetallic chromium complexes can have metalmetal distances of less than 1.80 . This indicates that formally quadruply bonded complexes can have much shorter metal metal distances than was thought to be the case for many years. The key to this compound is an organometallic synthesis in which AlMe3 acts as an N-ligand acceptor and a methyl donator.

The first quintuple bond in a stable molecule

Despite the fact that transition metals can formally form bonds with six shared electrons, only quadruply bonded coordination compounds had been isolated until 2005. Transition metal dimers,
532

2009 Macmillan Publishers Limited. All rights reserved.

Nature chemistry doi: 10.1038/nchem.359


that the CrCr bond length found in 1 is not the shortest possible for this type of system, allowing for exciting new possibilities.

review article
for trans-bent HCrCrH (ref. 62). In an independent bonding analysis of the same compound using the QTAIM methodology at the HartreeFock level of theory, a CrCr bond order of 3.6 was obtained67 on the basis of the delocalization index. The difference between the NRT and QTAIM results has been attributed to the physically different definitions used by the two methods of analysis. On the basis of a multiconfigurational CASPT2 wavefunction, a bonding analysis of this compound yielded a CrCr EBO of 3.4 (ref.68), which is identical to the one obtained for 1 (ref.60) at the same level of theory. Interestingly, the authors of ref.68 encountered some difficulties with the EBO formalism owing to the delocalization of one chromium d orbital, and the reported EBO was obtained only after an additional localization procedure. Thus, three different types of bond-order analysis have been reported for this compound, two at the monodeterminantal level of theory but fully including delocalization effects, and one at the multiconfigurational level but with a weakness with respect to the full separation of metalmetal interactions from metalligand interactions. It seems that, with both delocalization and correlation effects being significant, the method of analysis has an important role as well. This makes the analysis of chemical bonding of these types of compounds a real challenge. Clearly, NRT and QTAIM analysis at the multiconfigurational level of theory for this and similar compounds would be very interesting. As a continuation of ref. 7, Power and co-workers synthesized derivatives of 1 with varying substitution patterns, finding metal metal distances of between 1.8077(7) and 1.8351(4) . Packing forces rather than electronic effects were thought to control the different distances69. Inspired by 2 and the potential of three-atom bridging ligands to establish short metalmetal distances6, introduced as the Hein Cotton concept 70, two groups independently synthesized compounds stabilized by such ligands, and respectively observed metalmetal distances of 1.75 and 1.74, for instance, 3 and 4 (Fig.3b,c)70,71,72. The NN distances within the ligands listed in Fig.3 demonstrate the importance of the ligand in terms of establishing short metal metal bonds. These NN distances clearly correlate with the CrCr bond lengths. The examples shown in Fig.3b,c represent the class of ultrashort metalmetal-bonded homobimetallics. Calculations indicate quintuple bonding. The CrCr bond lengths of these compounds are intermediate between the quadruply bonded Cr(ii) Cr(ii) complexes and the gas-phase Cr2 molecule with its formal sextuple bond. Interestingly, in one of the compounds71 chromium is three-coordinated with respect to the ligand, which is difficult to understand in the context of formal quintuple bonding, as the number of metal dorbitals involved in metalligand bonding is considered to be minimized60. A comparison of the CrN bond lengths
c
NN, 2.24 CrCr, 1.74

Ultrashort metalmetal bonds

A way of comparing the shortness of bonds across the periodic table, and thus establishing what a short bond is, is the formal shortness ratio (FSR)6. The FSR is a dimensionless number given by the ratio of the atomatom distance, d, of a bond and the sum of the radii1 of the two atoms involved, rA+rB: FSR=d/(rA+rB). The advantage of this formalism is its interelement applicability. The FSR is a useful tool for comparing formal quintuple bonding and short metal metal distances within the group-6 metals, for instance. Because here we mainly discuss the distances in chromium homobimetallic complexes, we write in terms of the distances and not the FSRs. Quintuple bonds could naively be expected to be shorter than quadruple bonds, considering general trends that have been observed for increasing bond orders in main-group chemistry 1. However, the first formal quintuple bond observed in a stable molecule was longer than the shortest CrCr distances observed in formally quadruply bonded Cr(ii) homobimetallic complexes at that time. The CrCr bond length of laser-evaporated Cr2 in the gas phase is about 1.68 (ref.56). Spectroscopic studies of Cr2 generated from pulsed photolysis of Cr(CO)6 are indicative of a distance of 1.71 (ref.57), and the calculated minimum of the ground-state potential is comparable in distance65. The Cr2 molecule can be regarded as formally sextuply bonded. Interestingly, the potential curve is rather flat around the minimum65 and should, if it is similarly flat in multiply bonded Cr(i) dimers, allow drastic shortening by rational ligand design. A quintuply bonded complex should reach a CrCr distance between those observed for 1 and Cr2, namely between 1.84 and 1.68. Consequently, efforts have been made to find formally quintuply bonded complexes that have metalmetal bond lengths within this range. The first report in this regard66 considered an N-ligandstabilized homobimetallic chromium complex (CrCr bond length, 1.8028(9)). The compound, 2 (Fig.3a), was synthesized by addition of an excess of sodium and a Cr(iii) chloride to a sterically demanding diazadiene, affording a chromium monochloride that was further reduced using potassium graphite. The problem with the diazadiene used, and other imine ligands, is its redox ambiguity, which makes it difficult to judge the oxidation state of the chromium atoms66. The DFT electronic-structure calculations reported in ref.66 showed that one of the five occupied CrCr bonding orbitals, a d orbital, is highly delocalized over the ligand. NRT analysis was used to investigate the consequences of the delocalization on the metalmetal bond order. The NRT-based bond order of 4.3 indicates a rather significant effect, to be compared with the NRT-based bond order of 4.6 obtained
a
NN, 2.84 CrCr, 1.80

NN, 2.26 CrCr, 1.75

Figure 3 | Homobimetallic chromium compounds in which unusually short metalmetal distances were observed. ac, Molecular structures of 2, 3 and 4, respectively. Complexes 3 and 4 represent the class of compounds that contain ultrashort metalmetal bonds. Quintuple bonding and metalmetal distances nearly as short as for the transient Cr2 molecule, which could be considered formally sextuply bonded, have been observed in them. Colour code as in Fig.1.
nature chemistry | VOL 1 | OCTOBER 2009 | www.nature.com/naturechemistry

533

2009 Macmillan Publishers Limited. All rights reserved.

review article
a b

Nature chemistry doi: 10.1038/nchem.359


c
N

N Cr N

N Cr N

N Cr N

N Cr N

N Cr N N

N Cr N

Figure 4 | The role of the ligand in stabilizing ultrashort metalmetal bonds. a, In aminopyridinates, the wing-upwing-down arrangement may cause interligand repulsion limiting the compression of the two metals by the ligands. b, In amidinates, a two-wings-up arrangement allows the wings to be put further down, aligning the N-centred lone pairs to provide shorter metalmetal bonds. c, In guanidinates, the steric pressure on top of the ligand initiates a process that pushes the wings down, resulting in a further shortening of the CrCr distance.

in a series of structurally analogous chromium amidinates, in which the distances are roughly equal, is indicative that the CrN bond lengths are independent of the steric demand of the ligand72. For the aminopyridinate70, chemical bonding analysis at the DFT level has been done using position-space quantities, the electron localizability indicator 73,74 and the delocalization index. Even at the monodeterminantal level, indications of a CrCr bond order significantly reduced relative to the Herzberg number owing to weak interatomic electron sharing of one of the bonds in particular are reported. The other bond can be seen to be of the more favourable sdtype already discussed62. The calculated bond order of 4.2 based on the delocalization index between the QTAIM chromium atoms supports the notion of a formal quintuple bond. The majority of the contributions (62%) are found to originate in the chromium third atomic shell region in position space, which is consistent with earlier results obtained for Mo2(formamidinate)4 displaying a formal quadruple bond75. We note that a significantly lower value of the delocalization index, of 3.6, was obtained for the CrCr bond in 2 in a later study 67, which may indicate a substantial difference in CrCr bonding. By comparing the two ligand families used to stabilize ultrashort CrCr bonds, namely aminopyridinates and amidinates (Fig.4a,b), the two-wings-up arrangement observed for amidinates (Fig.4b) apparently causes much less interligand repulsion within the bimetallic complex than the wing-upwing-down arrangement observed for aminopyridinates (Fig.4a). This allows the generation of a closer NCN pincer and/or an alignment of the ligand orbitals (lone pairs) that bind with chromium towards each other. Both structural consequences may result in a shorter CrCr distance. As a result, steric pressure generated through the introduction of a bulky substituent at the bridging carbon atom should give rise to a further reduced distance between the metal atoms (Fig. 4c). Guanidinates appear to be ideally suited for this purpose. The system is delocalized and thus prefers a planar arrangement of the three Natoms and their residues (Fig.4c). This planar arrangement can be embedded between two bulky 2,6-alkyl groups of the phenyl wings. Consequently, the substituents of the non-metalbonded N atom push both wings further down. A ligand introduced recently, (CH3)2N-C(N-2,6-diisopropylphenyl)2, seems to be ideal in this regard76 and was used successfully to shorten the CrCr bond length to 1.73 (ref. 77). The molecular structure of this compound, 5, is shown in Fig. 5. The space-filling model (Fig.5b) indicates the steric pressure of the methyl groups (brown carbon atom) towards the wing. The CrCr distance of 5 is 0.1 shorter than the distances observed in complexes that held the 30-year record for the shortest metalmetal distances in a stable molecule (Fig.1a). The studies discussed here indicate that the shortening of the metalmetal
534

bond of ultrashort chromium homobimetallics is primarily determined by the ligand. This has also been one of the conclusions of a very recently published article78. CrCr bonding competes with Crligand bonding, yielding lower CrCr bond orders and in principle larger CrCr distances for substantial ligand -bond participation. The CrCr distance effect may be hidden by the bracketing effect of the ligand. A stimulating model calculation at the CASPT2 level on the hypothetical complex FCrCrF, in which ligand delocalization is maximally excluded, yielded a CrCr distance of 1.65 . This is similar to that obtained for Cr2, which sets the benchmark for rational ligand design.

Outlook for other metals and quintuple bond reactivity

Dimers of the 4f and 5f metals could in principle form bond orders even higher than the transition metals. The crucial point is the itinerancy of the f states. For this reason, not the lanthanides, with their atomic-like localized 4f states, but the actinides are considered the interesting candidates. On the experimental side, evidence for these types of species comes from matrix isolation and gas-phase detection79 (U2). Electronic structures of the U2 dimer have been calculated80 at the CASPT2 level of theory including either scalar relativistic effects or even spinorbit coupling. Ten of the 12 valence electrons are found to be chemically active, displaying an exotic bonding pattern with a classical bonding part consisting of one (7sg) and two (6du) doubly occupied bonding orbitals, and a ferromagnetic bonding 81 part with four singly occupied bonding orbitals (6dg, 6dg and 5fu, 5fg). In total, a formal bond order of 3+4/2=5 is obtained. The remaining two electrons are found to be situated in localized 5f
a
CrCr, 1.7293(12)

Figure 5 | Molecular structure of the coordination compounds with the shortest metalmetal bond so far observed. a, Ellipsoid plot of 5. b, Spacefilling model indicating the steric pressure provided by the methyl moieties (brown carbon atom) towards the aryl wing. Colour code as in Fig.1.
nature chemistry | VOL 1 | OCTOBER 2009 | www.nature.com/naturechemistry

2009 Macmillan Publishers Limited. All rights reserved.

Nature chemistry doi: 10.1038/nchem.359


orbitals on each Uatom. The calculations predict all six electrons to be ferromagnetically coupled. As a continuation of this work in a broader and more refined study 19 of Ac2, Th2, Pa2 and U2, the concept of the EBO has been additionally applied to characterize chemical bonding. EBO values of 1.7, 3.7, 4.5 and 4.2 have been obtained for Ac2, Th2, Pa2 and U2, respectively. The Pa2 molecule displays the highest EBO, the highest computed bond energy (4.0eV) and the shortest distance (2.37). Despite being close to Pa2 in EBO value, U2 has a much lower bond energy (1.2 eV), whereas Th2, which has a lower EBO, is found to have much higher bond energy (3.3eV). There is no direct relation between the EBO and the bond energy, because the latter is a complicated energy difference that depends on variable electronic properties of the actual molecule and the isolated atoms. The study has not been continued to heavier homologues. Owing to increasing 7s promotion energies and the increasing localization of the 5f electrons, bond energies and bond orders are expected to be generally smaller than for the early actinides. Thus, the question arises of what is the highest covalent-bond bond order between any atoms of the periodic system. On the basis of CASPT2 calculations (at least at scalar relativistic level) it has been argued82 that the maximum bond order between two transition metal atoms can be six, with two, two and twobonds. In contrast to the main-group elements, high effective bond orders have been found to be more favourable for the heavier homologues. Scalar relativistic effects give rise to more itinerant (n1)dorbitals, leading to higher effective EBO contributions from the bonds and to more contracted nsorbitals, yielding more optimal -bonding contributions. Thus, although for Cr2 an EBO of 3.5 and a dissociation energy of only 1.65eV (calculated) have been obtained, EBO values of 5.2 have been computed for Mo2 and W2 molecules with dissociation energies of 4.41 (calculated) and 5.37 (calculated), respectively. Likewise, to achieve high EBOs for formal quintuple bonds, it has been proposed that the molybdenum and tungsten homologues of the Cr(i) compounds be synthesized. Another candidate for a strong quintuple bond has been predicted to be Nb2, because the niobium atom is already fully polarized (d 4s 1 configuration) in its ground state. The heavier homologue, Ta2, is less favourable owing to the higher promotion energy of the tantalum atom (d 3s 2 configuration) to a valence state with five unpaired electrons. Because stable, formally quintuply bonded species are now available, the investigation of the reactivity of these diamagnetic Cr2 platforms can be started. The first results are available and, for instance, the carboalumination of a CrCr quintuple bond has been observed (Fig.6)83. Carbometallation and especially carboalumination reactions of CC double and triple bonds are a well-established synthetic protocol in organometallic chemistry and organic synthesis. Analogous reactivity patterns of CrCr quintuple bonds indicate that such quintuple bonds are not as exotic as was assumed beforehand. However, the peculiarites of these reactions reflect the specific nature of the high metalmetal bond orders. The reactions of 1 with N2O and adamantanyl azide resulted in the formation of compounds that have no metalmetal bonding 84. These studies are the final step in the journey from establishing stable quintuple bonds through an understanding of their electronic structure(s) to investigating their reactivity and chemistry to find potential applications. At this early stage, it is rather difficult to judge the impact coordination compounds containing formal quintuple bonds may have, or to foresee possible applications, as only a few reactivity studies are available. They show the potential for small-molecule activation, particularly on a diatomic platform that can provide from two to eight (in principle even ten) electrons. In addition, one could also expect a potential for homogeneous catalysis, because a monoligand-stabilized Cr(i) species can be provided which is not coordinatively saturated by other ligands, for instance by solvent molecules. Selective ethylene tri- and tetramerization might be examples. What we can
nature chemistry | VOL 1 | OCTOBER 2009 | www.nature.com/naturechemistry

review article
0H$O & & & $O0H
Al1

0H &
C1 Cr2

&U

&U

Cr1

&

& 0H & &

0H$O

Figure 6 | Carboalumination of a CrCr quintuple bond. This carboalumination proceeds in analogy to carboalumination reactions observed for CC double and triple bonds, and thus indicates similarities between these classical bonds and quintuple bonding. The blue bonds indicate the unprecedented binding of the (H3C)2Al moiety to the two chromium atoms. Aluminium, white, plus colour code as in Fig.1.

really do with these high-bond-order platforms may become clear if more examples with other metals are available. In fact, during the proofing process for this article, the first examples of molybdenum molybdenum quintuple bonds have appeared85. Understanding chemical bonding better is another important aspect. Quintuple bonding is only a small part of understanding chemical bonds, and we must now seek to place it into a common conceptual framework encompassing all types of chemical bonds.

References

1. Pauling, L. The Nature of the Chemical Bond 3rd edn (VCH, 1973). 2. Lewis, G.N. The atom and the molecule. J.Am. Chem. Soc. 38, 762785 (1916). 3. Shaik, S. The Lewis legacy: the chemical bonda territory and heartland of chemistry. J.Comput. Chem. 28, 5161 (2007). 4. Heitler, W. & London, F. Interaction of neutral atoms and homopolar binding according to the quantum mechanics. Z.Phys. 44, 455472 (1927). 5. London, F. The quantum theory of monopolar valence numbers. Z.Phys. 46, 455477 (1928). 6. Cotton, F.A., Murillo, L.A. & Walton, R.A. Multiple Bonds Between Metal Atoms 3rd edn (Springer, 2005). 7. Nguyen, T. et al. Synthesis of a stable compound with fivefold bonding between two chromium(I) centers. Science 310, 844847 (2005). 8. Ruedenberg, K. & Schmidt, M.W. Why does electron sharing lead to covalent bonding? A variational analysis. J.Comput. Chem. 28, 391410 (2007). 9. Kutzelnigg, W. in Theoretical Models of Chemical Bonding, Part 2: The Concept of the Chemical Bond (ed. Maksi, Z.B.) 143 (Springer, 1990). 10. Parr, R.G., Ayers, P.W. & Nalewajski, R.F. What is an atom in a molecule? J.Phys. Chem. A 109, 39573959 (2005). 11. Bader, R.F.W. Atoms in Molecules: A Quantum Theory (Clarendon, 1995). 12. Hirshfeld, F.L. Bonded-atom fragments for describing molecular charge densities. Theor. Chem. Acc. 44, 129138 (1977). 13. Hoffmann, R., Shaik, S. & Hiberty, P.C. A conversation on VB vs. MO theory: a never-ending rivalry? Acc. Chem. Res. 36, 750756 (2003). 14. Frenking, G. & Krapp, A. Unicorns in the world of chemical bonding models. J.Comput. Chem. 28, 1524 (2007). 15. Herzberg, G. Zum Aufbau der zweiatomigen Molekle. Z.Phys. 57, 601630 (1929). 16. Mulliken, R.S. Bonding power of electrons and theory of valence. Chem. Rev. 9, 347388 (1931). 17. Mulliken, R.S. Electronic structure of polyatomic molecules and valence. II. General considerations. Phys. Rev. 41, 4971 (1932). 18. Hall, M.B. Problems in the theoretical description of metalmetal multiple bonds or how I learned to hate the electron correlation problem. Polyhedron 6, 679684 (1987). 19. Roos, B.O., Malmqvist, P.-A. & Gagliardi, L. Exploring the actinideactinide bond in Ac2, Th2, Pa2, and U2. J.Am. Chem. Soc. 128, 1700017006 (2006). 20. Coulson, C.A. The electronic structure of some polyenes and aromatic molecules VII. Bonds of fractional order by the molecular orbital method. Proc. R.Soc. Lond. A 169, 413428 (1939). 21. Pauling, L., Brockway, L.O. & Beach, J.Y. The dependence of interatomic distance on single-bond-double-bond resonance. J.Am. Chem. Soc. 57, 27052709 (1935).
535

2009 Macmillan Publishers Limited. All rights reserved.

review article
22. Glendening, E.D. & Weinhold, F. Natural resonance theory: I. General formalism. J.Comput. Chem. 19, 593609 (1998); Natural resonance theory: II. Natural bond order and valency. J.Comput. Chem.19, 610627 (1998); Natural resonance theory: III. Chemical applications. J.Comput. Chem. 19, 628646 (1998). 23. Weinhold, F. & Landis, C. Valence and Bonding: A Natural Bond Orbital Donor-Acceptor Perspective 1st edn (Cambridge Univ. Press, 2005). 24. McWeeny, R. Charge densities in conjugated systems. J.Chem. Phys. 19, 16141615 (1951). 25. Mulliken, R.S. Electronic population analysis on LCAO-MO molecular wave functions. I. J.Chem. Phys. 23, 18331840 (1955). 26. Giambiagi, M., de Giambiagi, M.S., Grempel, D.R. & Heynmann, C.D. No. 3 Sur la definition dun indice de liaison (TEV) pour des bases non orthogonales. Proprietes et applications. J.Chim. Phys. 72, 1522 (1975). 27. Mayer, I. Charge, bond order and valence in the SCF theory. Chem. Phys. Lett. 97, 270274 (1983). 28. Wiberg, K.B. Application of the PopleSantrySegal CNDO method to the cyclopropylcarbinyl and cyclobutyl cation and to bicyclobutane. Tetrahedron 24, 10831096 (1968). 29. Mayer, I. Bond orders and valence indices: a personal account. J.Comput. Chem. 28, 204221 (2007). 30. De Giambiagi, M.S., Giambiagi, M. & Jorge, F.E. Bond index: relation to second order density matrix and charge fluctuations. Theor. Chim. Acta 68, 337341 (1985). 31. Bader, R.F.W. & Stephens, M.E. Spatial localization of the electron pair and number distributions in molecules. J.Am. Chem. Soc. 97, 73917399 (1975). 32. ngyn, J.G., Loos, M. & Mayer, I. Covalent bond orders and atomic valence indices in the topological theory of atoms in molecules. J.Phys. Chem. 98, 52445248 (1994). 33. Matito, E., Mayer, I. & Sol, M. General discussion. Faraday Discuss. 135, 367401 (2007). 34. Fradera, X., Austen, M.A. & Bader, R.F.W. The Lewis model and beyond. J.Phys. Chem. A 103, 304314 (1999). 35. Matito, E., Sol, M., Salvador, P. & Duran, M. Electron sharing indexes at the correlated level. Application to aromaticity calculations. Faraday Discuss. 135, 325345 (2007). 36. Fradera, X., Poater, J., Simon, S., Duran, M. & Sol, M. Electron-pairing analysis from localization and delocalization indices in the framework of the atoms-in-molecules theory. Theor. Chem. Acc. 108, 214224 (2002). 37. Peligot, E. Sur un nouvel oxide de chrome. C.R.Acad. Sci. 19, 609618 (1844); Recherches sur le chrome. Ann. Chim. Phys. 12, 528579 (1844). 38. Cotton, F.A., Curtis, N.F., Johnson, B.F.G. & Robinson, W.R. Compounds containing dirhenium(III) octahalide anions. Inorg. Chem. 4, 326330 (1965). 39. Cotton, F.A. & Harris, C.B. The crystal and molecular structure of dipotassium octachlorodirhenate(III) dihydrate, K2[Re2Cl8]2H2O. Inorg. Chem. 4, 330333 (1965). 40. Cotton, F.A. Metal-metal bonding in [Re2X8]2- ions and other metal atom clusters. Inorg. Chem. 4, 334336 (1965). 41. Cotton, F.A. et al. Mononuclear and polynuclear chemistry of rhenium(III): its pronounced homophilicity. Science 145, 13051307 (1964). 42. Kotelnikov, A.S. & Tronev, V.G. Study of the complex compounds of divalent rhenium. J.Inorg. Chem. USSR 3, 10081027 (1958). 43. Kuznetzov, B.G. & Kozmin, P.A. The determination of the structure of (PyH)HReCl4. Zh. Strukt. Khim. 4, 5562 (1963). 44. Trogler, W.C. & Gray, H.B. Electronic spectra and photochemistry of complexes containing quadruple metalmetal bonds. Acc. Chem. Res. 11, 232239 (1978). 45. Gagliardi, L. & Roos, B.O. The electronic spectrum of Re2Cl82-: a theoretical study. Inorg. Chem. 42, 15991603 (2003). 46. Krapp, A., Lein, M. & Frenking, G. The strength of the -, - and -bonds in Re2Cl82-. Theor. Chem. Acc. 120, 313320 (2008). 47. Cotton, F.A. & Koch, S.A. Rational preparation and structural study of a dichromium o-oxophenyl compound: the shortest metal-to-metal bond yet observed. Inorg. Chem. 17, 20212024 (1978). 48. Hein, F. & Tille, D. Zur Existenz substituierter Chromphenylverbindungen. Z.Anorg. Allg. Chem. 329, 7282 (1964). 49. Cotton, F.A., Koch, S.A. & Millar, M. Tetrakis(2-methoxy-5-methylphenyl) dichromium. Inorg. Chem. 17, 20842086 (1978). 50. Edema, J.H.H. & Gambarotta, S. Short and supershort CrCr distances: a vanishing borderline between metalmetal bonds, magnetic coupling and ligand artefacts. Comments. Inorg. Chem. 11, 195214 (1991). 51. Edema, J.H.H., Gambarotta, S., Van der Sluis, P., Smeets, W.J.J. & Spek, A.L. Preparation and X-ray structure of (tetramethyldibenzotetraaza[14]annulene) chromium dimer [(tmtaa)Cr]2. A multiply bonded complex of dichromium(II) without bridging ligands. Inorg. Chem. 28, 37823784 (1989). 52. Hao, S., Edema, J.H.H., Gambarotta, S. & Bensimon, C. Reversible cleavage of the chromiumchromium multiple bond in [(TAA)Cr]2 (TAA = tetramethyldibenzotetraaza[14]annulene). Inorg. Chem. 31, 26762678 (1992). 53. Horvath, S., Gorelsky, S.I., Gambarotta, S. & Korobkov, I. Breaking the 1.80 barrier of the CrCr multiple bond between CrII atoms. Angew. Chem. Int. Ed. 47, 99379940 (2008). 54. Knding, E.P., Moskovits, M. & Ozin, G.A. Matrix synthesis and characterisation of dichromium. Nature 254, 503504 (1975).
536

Nature chemistry doi: 10.1038/nchem.359


55. Klotzbcher, W. & Ozin, G.A. Niniobium, Nb2, and dimolybdenum, Mo2. Syntheses, ultraviolet-visible spectra and molecular orbital investigations of diniobium and dimolybdenum. Spectral and bonding comparison with divanadium (V2) and dichromium (Cr2). Inorg. Chem. 16, 984987 (1977). 56. Bondybey, V.E. & English, J.H. Electronic structure and vibrational frequency of diatomic chromium (Cr2). Chem. Phys. Lett. 94, 443447 (1983). 57. Efremov, Yu. M., Samoilova, A.N. & Gurvich, L.V. The = 4600 band in a spectrum by pulsed photolysis of chromium carbonyl. Opt. Spektrosc. 36, 654657 (1974). 58. Morse, M.D. Clusters of transition metal atoms. Chem. Rev. 86, 10491109 (1986). 59. Schiemenz, B. & Power, P.P. Synthesis and structure of a unique monomeric -bonded aryllithium compound stabilized by a weak Li-benzene interaction. Angew. Chem. Int. Edn Engl. 35, 21502152 (1996). 60. La Macchia, G., Gagliardi, L., Power, P.P. & Brynda, M. Large differences in secondary metalarene interactions in the transition-metal dimers ArMMAr (Ar = Terphenyl; M = Cr, Fe, or Co): implications for CrCr quintuple bonding. J.Am. Chem. Soc. 130, 51045114 (2008). 61. Frenking, G. Building a quintuple bond. Science 310, 796797 (2005). 62. Landis, C.R. & Weinhold, F. Origin of trans-bent geometries in maximally bonded transition metal and main group molecules. J.Am. Chem. Soc. 128, 73357345 (2006). 63. Merino, G., Donald, K.J., DAcchioli, J.S. & Hoffmann, R. The many ways to have a quintuple bond. J.Am. Chem. Soc. 129, 1529515302 (2007). 64. Brynda, M., Gagliardi, L., Widmark, P.-O., Power, P.P. & Roos, B.O. A quantum chemical study of the quintuple bond between two chromium centers in [PhCrCrPh]: trans-bent versus linear geometry. Angew. Chem. Int. Ed. 45, 38043807 (2006). 65. Roos, B.O. The ground state potential for the chromium dimer revisited. Collect. Czech. Chem. Commun. 68, 265274 (2003). 66. Kreisel, K.A., Yap, G.P.A., Dmitrenko, O., Landis, C.R. & Theopold, K.H. The shortest metalmetal bond yet: molecular and electronic structure of a dinuclear chromium diazadiene complex. J.Am. Chem. Soc. 129, 1416214163 (2007). 67. DuPr, D. Multiple bonding in the chromium dimer supported by two diazadiene ligands. J.Phys. Chem. A 113, 15591563 (2009). 68. La Macchia, G., Aquilante, F., Veryazov, V., Roos, B.O. & Gagliardi, L. Bond length and bond order in one of the shortest CrCr bonds. Inorg. Chem. 47, 1145511457 (2008). 69. Wolf, R. et al. Substituent effects in formally quintuple bonded ArCrCrAr compounds (Ar = terphenyl) and related species. Inorg. Chem. 46, 1127711290 (2007). 70. Noor, A., Wagner, F.R. & Kempe, R. Metalmetal distances at the limit: a coordination compound with an ultrashort chromiumchromium bond. Angew. Chem. Int. Ed. 47, 72467249 (2008). 71. Tsai, Y.-C. et al. Remarkably short metalmetal bonds: a lantern-type quintuply bonded dichromium(I) complex. Angew. Chem. Int. Ed. 47, 72507253 (2008). 72. Hsu, C.-W. et al. Quintuply-bonded dichromium(I) complexes featuring metal metal bond lengths of 1.74 . Angew. Chem. Int. Ed. 47, 99339936 (2008). 73. Kohout, M. Bonding indicators from electron pair density functionals. Faraday Discuss. 135, 4354 (2007). 74. Wagner, F.R., Bezugly, V., Kohout, M. & Grin, Yu. Charge decomposition of the electron localizability indicator - a bridge between the direct and Hilbert space representation of the chemical bond. Chem. Eur. J. 13, 57245741 (2007). 75. Llusar, R., Beltrn, A., Andrs, J., Fuster, F. & Silvi, B. Topological analysis of multiple metalmetal bonds in dimers of the M2(formamidinate)4 type with M = Nb, Mo, Tc, Ru, Rh, and Pd. J.Phys. Chem. A 105, 94609466 (2001). 76. Ge, S., Meetsma, A. & Hessen, B. Highly efficient hydrosilylation of alkenes by organoyttrium catalysts with sterically demanding amidinate and guanidinate ligands. Organometallics 27, 31313135 (2008). 77. Noor, A. et al. Metal-metal distances at the limit: Cr-Cr 1.73 - the importance of the ligand and its fine tuning. Z.Anorg. Allg. Chem. 635, 11491152 (2009). 78. Brynda, M., Gagliardi, L. & Roos, B.O. Analysing the chromiumchromium multiple bonds using multiconfigurational quantum chemistry. Chem. Phys. Lett. 471, 110 (2009). 79. Gorokhov, L.N., Emelyanov, A.M. & Khodeev, Y.S. Mass-spectroscopic investigation of stability of gaseous molecules of U2O2 and U2. High Temp. 12, 11561158 (1974). 80. Gagliardi, L. & Roos, B.O. Quantum chemical calculations show that the uranium molecule U2 has a quintuple bond. Nature 433, 848851 (2005). 81. Glukhovtsev, M.N. & v Schleyer, P.R. Polyatomic molecules without electronpair bonds: high-spin trigonal, tetrahedral, and octahedral lithium clusters. Isr. J.Chem. 33, 455466 (1993). 82. Roos, B.O., Borin, A.C. & Gagliardi, L. Reaching the maximum multiplicity of the covalent bond. Angew. Chem. Int. Ed. 46, 14691472 (2007). 83. Noor, A., Glatz, G., Mller, R., Kaupp, M., Demeshko, S. & Kempe, R. Carboalumination of a chromiumchromium quintuple bond. Nature Chem. 1, 322325 (2009). 84. Ni, C., Ellis, B.D., Long, G.J., Power, P.P. Reactions of ArCrCrAr with N2O or N3(1-Ad) : complete cleavage of the CrCr quintuple interaction. Chem. Commun. 23322334 (2009). 85. Tsai, Y.-C. et al. Journey from MoMo quadruple bonds to quintuple bonds. J. Am. Chem. Soc. 131, 1253412535 (2009).
nature chemistry | VOL 1 | OCTOBER 2009 | www.nature.com/naturechemistry

2009 Macmillan Publishers Limited. All rights reserved.

Anda mungkin juga menyukai