COMMENT Beyond the bond More than ever before, new techniques show the bond to be a convenient fiction, albeit one that holds the field of chemistry together, finds Philip Ball. ot so long ago, the chemistry stu- fresh ways to probe and quantify chemi- the dynamical dimension. Molecules have dent’s standard text on the theory of cal bonds2. Traditional measures such as traditionally been regarded, if not as static, OZ CHAST CHAST OZ chemical bonding was Charles Coul- crystallographic interatomic distances and then as having clear architectural frame- R Nson’s Valence (1952). Absent from it was this dissociation energies have been supple- works that are merely shaken and rotated by ONS BY BY ONS theoretical chemist’s real view on the sticks mented by spectroscopic techniques for thermal motions. The bonds get stretched I that generations of students have drawn to determining vibrational frequencies, meth- and bent, but they still have an equilibrium link atoms into molecules. “A chemical bond ods such as nuclear magnetic resonance to length and strength that seems to justify is not a real thing: it does not exist: no one has measure shifts in the electronic environment their being pictured as lines and stalks. Now, ILLUSTRAT ever seen it, no one ever can. It is a figment of of atoms and their magnetic interactions, thanks to ultrafast spectroscopies, such time- our own imagination,” he later wrote1. measurements of force constants (bond average values may not always accurately There is a good reason for postponing this stiffness) and a host of quantum-chemical characterize either structure or reactivity. awkward truth. The bond is literally the glue tools for calculating such aspects as electron What is measured in a bond depends not just that makes the entire discipline cohere; to distributions or localization. on how but on when it is measured. consider it an objective reality is necessary An indication of the shifting ground is the Some chemists argue that as a result of this, for any kind of discourse on chemistry. The recent decision to redefine the nature of the the very existence (or not) of a bond depends discipline is in fact riddled with such conven- hydrogen bond in the light of experimental on how the problem is probed. Others are ient (and contested) fictions, such as electro- results that alter the traditional electrostatic committed to absolute criteria. This differ- negativity, oxidation state, tautomerism description. ence of opinion goes to the heart of what and acidity. The nature of the chemical bond is now chemistry is about: can all be reduced to Disputes about the correct description further complicated by the introduction of quantum physics, or are fuzzy rules of thumb of bonding have ruffled chemists’ feathers essential? More pressingly, the issue of how since the concept of molecular structure 20��: YEAR OF CHEMISTRY best to describe a chemical bonding pattern first emerged in the mid-nineteenth century. Celebrating the central science has tangible implications for a wide range Now they are proliferating, as new theoreti- nature.com/chemistry20�� of problems in chemistry, from molecules cal and experimental techniques present in which atoms are coerced out of their 26 | NATURE | VOL 469 | 6 JANUARY 2011 © 2011 Macmillan Publishers Limited. All rights reserved COMMENT usual bonding geometry to the symmetric as some combination of the wavefunctions der Waals attraction, which is why helium hydrogen bond (in which a bound hydrogen of individual electrons. That’s also the basis is a liquid at very low temperatures; but they atom is shared equally between two other of the Hartree–Fock method for calculating are not generally thought to be chemically atoms), and new variations on old themes the wavefunction and energy of the lowest- bonded as a result. such as aromaticity (special patterns of ‘delo- energy state (ground state) of a molecular Besides, the ‘bonded or not’ question calized’ bonds, like those in benzene)3. system — a method that became practical becomes context-dependent once atoms Just about every area of chemistry har- in the 1950s, when computers made it pos- are embedded in a molecule, where they bours its own bonding conundrums. Most sible to solve the equations numerically. may be forced close together merely by illustrate that we have a far from exhaustive But separating the wavefunction into one- the atoms around them, and where there understanding of the ways in which quan- electron components is a fiction, as they is inevitably some arbitrariness in decid- tum rules will permit atoms to unite — and all influence one another: the behaviour of ing which electrons that in consequence, the synthetic inventive- one electron depends on what all the oth- “Chemists’ ‘belong’ to which ness of chemists suffers from a limited view ers are doing. The difference between true inventiveness atoms. The result- of the possibilities. ground-state energy and that calculated ing ambiguities were using the Hartree–Fock approach is called suffers from illustrated recently CARVING UP ELECTRONS the correlation energy. More recent com- a limited when three experts We can all agree on one thing: chemical putational methods can capture most of view of the failed to agree about bonding has something to do with elec- the correlation energy, but none can give possibilities.” whether two sulphur trons. Two atoms stick together because an exact solution. As a result, describing atoms in an organo- of the arrangement of electrons around the quantum chemical bond remains a mat- metallic compound are linked by a bond5. their nuclei. In the nineteenth century, it ter of taste: all descriptions are, in effect, The argument involved different interpre- was commonly thought that this attraction approximate ways of carving up the electron tations of quantum-chemistry calculations, was electrostatic: that atoms in molecules distribution. tussles over the best criteria for identifying are positively or negatively ionized. That If that were the limit of the bond’s a bond, and evidence of precedent from left the puzzle of how identical atoms can ambiguity, there would be little to argue comparable compounds. form diatomic molecules such as H2 and O2. about. It is not. There is, for example, the All this is merely a reminder that a mol- American chemist G. N. Lewis proposed that matter of when to regard two atoms as being ecule is ultimately a set of nuclei embedded bonding can instead result from the sharing bonded at all. Pauling’s somewhat tautologi- in a continuous electron cloud that stabilizes of electrons to create filled shells of eight, cal definition rather gave the game away: a a particular configuration, which balls and visualized as the corners of a cube. group of atoms is to be considered bonded sticks can sometimes idealize and some- In the 1920s and 1930s another Ameri- when it is “convenient for the chemist to times not. But disputes about the nature of can chemist, Linus Pauling, showed how consider it as an independent molecular spe- the chemical bond are not simply semantic. this interaction could be formulated in the cies”. Pauling admitted some of the conse- It matters, for example, whether we regard language of quantum mechanics as the over- quent ambiguities: for example, although his a very strong multiple bond as quintuple or lap of electron orbitals (see Nature 468, 1036; sextuple, even if this is a categorization that 2010). In essence, if two atomic orbitals each only textbooks, and not nature, recognize containing a single electron can overlap, a — not least, because textbook concepts are bond is formed. Pauling generalized earlier what provide a discipline with intellectual work on the quantum description of hydro- coherence and consistency. gen to write an approximate equation for Besides, how we talk about bonds can the wavefunction created by orbital overlap. determine our ability to rationalize real This became known as the valence-bond chemical behaviour. For example, the differ- (VB) description. ent descriptions of the bonds in what are now But an approximation is all it is. Around called non-classical ions of hydrocarbons — the same time, Robert Mulliken and Frie- whose relative merits were furiously debated drich Hund proposed another approximate in the 1950s and 1960s — have direct implica- wavefunction, which led to an alternative way tions for the way these species are predicted to formulate bonds: not as overlaps between to react. Whether to consider the bonding specific orbitals on separate atoms, but as non-classical, involving electrons spread over electron orbitals that extend over many atoms, more than two atomic nuclei, or tautomeric, called molecular orbitals (MOs). The relative involving rapid fluctuations between con- merits of the VB and MO descriptions were ventional two-atom bonds, had immediate debated furiously for several decades, with consequences for organic chemistry. no love lost between the protagonists: Mul- definition in general excludes the weak van Perhaps one might seek a distinction liken’s much-repeated maxim, “I believe the der Waals attraction that occurs between all between bonded and not-bonded in terms of chemical bond is not so simple as some people atoms, occasionally — as in the association how the force between two atoms varies with seem to think”, was possibly a jibe at Pauling. of two oxygen molecules into the O4 cluster their separation? Yes, there is an exponential By the 1960s, for all Pauling’s salesmanship, — even this force can be strong enough to be fall-off for a covalent bond such as that in MO theory was generally agreed to be more regarded as a chemical bond. H2, and a power-law decay for van der Waals convenient for most purposes. But the debate It’s no use, either, to suggest (as Coulson attraction.