Metal Carbonyl Complexes and Structures

Binary metal carbonyls remain industrially and synthetically central more than a century after Mond's discovery. The Mond process is still used at industrial scale for nickel refining, exploiting exactly the reversible formation/decomposition equilibrium Mond identified. Fe(CO)₅ is manufactured on a large scale and thermally decomposed to produce ultra-pure, extremely fine "carbonyl iron" powder used in soft-magnetic composite cores for inductors and electromagnetic-interference shielding, and historically in powder-metallurgy applications. Ni(CO)₄, despite its extreme toxicity (it is a potent pulmonary and systemic poison, precisely because of its volatility and ability to deliver Ni deep into tissue before decomposing), remains central to industrial nickel purification and to chemical vapor deposition of nickel films in electronics fabrication. In synthetic organometallic and organic chemistry, mononuclear and derived carbonyl complexes are workhorse stoichiometric and catalytic reagents. Fe(CO)₅ and its photochemically or thermally generated fragments (Fe(CO)₄, a coordinatively unsaturated 16-electron species) are used in alkene isomerization and in stoichiometric carbonylative transformations. The Collman reagent, Na₂[Fe(CO)₄] discussed in Example 2, is a textbook nucleophilic organometallic reagent for converting alkyl halides and tosylates into aldehydes via acyl-iron intermediates, then reductively cleaving the C–Fe bond. Co₂(CO)₈, formed because cobalt cannot achieve 18 electrons mononuclearly, is the direct precursor to HCo(CO)₄ and to the active catalytic species in industrial hydroformylation (the oxo process), which converts terminal alkenes and syngas (CO/H₂) into aldehydes on a multi-million-tonne annual scale, a process whose mechanism is a showcase of 18-electron bookkeeping at every catalytic intermediate. Mn₂(CO)₁₀ and Re₂(CO)₁₀ are precursors to a wide family of manganese- and rhenium-carbonyl photochemistry and radiopharmaceutical labeling chemistry (⁹⁹ᵐTc and ¹⁸⁶/¹⁸⁸Re carbonyl complexes are explored as diagnostic and therapeutic radiotracers, exploiting the same electron-counting logic that governs the parent binary carbonyls). W(CO)₆ and Mo(CO)₆ are common, air-stable, commercially available starting materials for generating reactive fragments (by thermal or photochemical CO loss) used throughout synthetic inorganic and catalytic chemistry, from olefin metathesis catalyst precursors to CVD sources for depositing refractory metal thin films in semiconductor manufacturing. In every one of these applications, the same 18-electron logic developed here for the parent binary carbonyls is the first tool a chemist reaches for in predicting what a new derivative, fragment, or catalytic intermediate should look like.

:::quiz[What geometry is predicted for the 18-electron, tetrahedral anion [Co(CO)₄]⁻, and why?]

• Trigonal bipyramidal, because Co is group 9
• Tetrahedral, because four identical ligands minimize mutual repulsion at the vertices of a tetrahedron, with no stereoactive lone pairs at the d10 metal center ✓
• Square planar, because Co(−I) is isoelectronic with Pt(II)
• Octahedral, because 18-electron complexes are always octahedral

Ask the tutor: Why does Fe₂(CO)₉ have three bridging carbonyls and an Fe–Fe bond instead of simply being a five-coordinate 18-electron monomer like Fe(CO)₅?

Carbon monoxide is an unusual ligand. Free CO has a very short, strong C≡O triple bond (bond order 3 in the simplest MO picture, with a filled 5σ HOMO that is weakly antibonding and largely carbon-centered, and empty 2π orbitals antibonding between C and O). When CO binds a metal, it does so <strong>end-on through carbon</strong>, donating the 5σ lone pair into an empty metal d-orbital (or hybrid) of correct symmetry — that is the classical Lewis-acid/Lewis-base σ-donation you already know from Werner-type coordination chemistry. But CO is also a good π-acid: the metal's filled d-orbitals of correct symmetry donate electron density back into the ligand's empty π orbitals. This "synergic" push–pull is the subject of its own dedicated treatment; here we need only the electron-counting consequence: regardless of the detailed bonding picture, CO is treated as a neutral two-electron donor ligand in the standard convention, contributing its 5σ lone pair without changing the formal oxidation state of the metal. Because CO is a neutral donor (unlike, say, Cl⁻, which is anionic and forces you to adjust oxidation state), binary carbonyls M(CO)ₙ almost always have the metal at oxidation state 0. There is no counter-ion, no anionic ligand, nothing to balance — a neutral molecule made of a neutral metal atom and neutral CO ligands is automatically at M(0). This is the first and most important bookkeeping fact: for a homoleptic (all-CO) neutral mononuclear carbonyl, the metal oxidation state is 0, and its d-electron count equals the group number (in the modern convention where group number = number of valence electrons of the neutral atom, appropriate for M(0)). The 18-electron rule states that a stable, closed-shell transition-metal complex will accumulate a total of 18 valence electrons around the metal — the sum of the metal's d-electrons (at its actual oxidation state) and the electrons donated by all the ligands. Structurally this total electron count fills a set of nine valence orbitals: one metal ns orbital, three (n)p orbitals, and five (n−1)d orbitals — nine orbitals, eighteen electrons, exactly analogous to how eight electrons fill four orbitals (one s + three p) at a main-group atom obeying the octet rule. For a neutral homoleptic carbonyl this collapses to a beautifully simple relationship: if the metal in group $g$ (contributing $g$ electrons at M(0)) binds $n$ CO ligands (each donating 2 electrons),

so $n = (18-g)/2$. This single line predicts every mononuclear binary carbonyl stoichiometry in the periodic table, and its failures (odd $g$, giving non-integer or radical-forced $n$) predict exactly which metals cannot form simple mononuclear carbonyls and must instead dimerize. Once $n$ is fixed, geometry follows a VSEPR-like logic even though there are no lone pairs on the metal to worry about (all "extra" d-electrons in an 18-electron count are typically in filled, non-bonding, or metal–ligand π-backbonding orbitals, not stereochemically active lone pairs the way main-group lone pairs are). With d-orbitals available for π-backbonding to every ligand, and no lone-pair repulsion complications, the geometries seen are simply those that minimize CO–CO steric/electronic repulsion for a given coordination number: linear for n = 2 (formally, though no stable mononuclear dicarbonyls of this type are common at n=2 for neutral binary species — but the principle appears in substituted complexes), trigonal for n = 3, tetrahedral for n = 4, trigonal bipyramidal for n = 5, and octahedral for n = 6. These are precisely the shapes VSEPR would assign to a central atom with that many bonding domains and no lone pairs.

The three panels above show the pattern in miniature: as the group number of the metal rises by one, the number of CO ligands needed to reach 18 electrons falls by exactly half of one — but because $n$ must be an integer, in practice $n$ falls by one every two groups (6 → 5 → 4 as group goes 6 → 8 → 10, skipping the odd groups 7 and 9 which give non-integer or radical solutions). This "every other group" pattern is precisely why V(CO)₆, Mn(CO)₅, and Co(CO)₄· are not stable 18-electron mononuclear species, while Cr(CO)₆, Fe(CO)₅, and Ni(CO)₄ are the textbook carbonyls.

The master relationship for the neutral-atom ("ionic") method of electron counting, applied to a homoleptic neutral binary carbonyl M(CO)ₙ, is

where:

• $\text{EC}$ = total valence electron count at the metal center
• $g$ = group number of the metal (equal to the number of valence electrons of the neutral atom, since oxidation state is 0 in a homoleptic neutral carbonyl)
• $n$ = number of CO ligands, each donating 2 electrons as a neutral σ-donor

Reading: this equation says the metal's own valence electrons and the lone pairs donated by every CO ligand simply add up; setting the sum to 18 and solving for $n$ tells you exactly how many carbonyls a given metal needs. Solving for the carbonyl stoichiometry that satisfies the 18-electron rule,

where:

• $n$ must come out to a non-negative integer for a stable mononuclear 18-electron binary carbonyl to exist
• $g$ ranges over the groups where carbonyls are known to form (chiefly 5–10)
• non-integer $n$ signals that mononuclear closed-shell carbonyls are geometrically/electronically forbidden for that metal, forcing either a 17-electron radical or M–M bond formation in a dimer

Reading: because $18-g$ must be even for $n$ to be an integer, $g$ must be even — this is the deep reason only even-group metals (6, 8, 10) give stable simple mononuclear carbonyls, while odd-group metals (5, 7, 9) do not. For polynuclear species where metal–metal bonds supply the missing electrons, the electron count at each metal is

where:

• $n_{\text{CO,terminal}}$ = terminal CO ligands bound only to that metal (2e⁻ each)
• $n_{\text{CO,bridging,shared}}$ = the fractional contribution from bridging carbonyls, conventionally counted as each bridging CO donating 1 electron to each of the two metals it bridges (so 2 electrons total, split 1+1) in the covalent/neutral counting convention
• $n_{M-M}$ = number of metal–metal bonds from that metal (each single M–M bond contributes 1 electron to each partner's count)

Reading: every M–M single bond "lends" one electron to each metal's count, exactly as sharing a covalent bond between two carbon atoms gives each carbon credit for both electrons in that bond — this is precisely how Mn₂(CO)₁₀ and Co₂(CO)₈ solve the odd-group problem.

Example 1 (foundational): Predict the formula and geometry of the tungsten binary carbonyl. Tungsten is in group 6, so as a neutral atom in a homoleptic zero-valent carbonyl it contributes $g = 6$ valence electrons. Solve for the carbonyl stoichiometry:

So the predicted formula is W(CO)₆. Six identical CO ligands minimize mutual repulsion by adopting an octahedral arrangement (Step 2 of the second derivation above), so W(CO)₆ is predicted to be octahedral, Oₕ. This matches the real, well-characterized compound: W(CO)₆ is a white, air-stable, sublimable solid, isostructural with Cr(CO)₆ and Mo(CO)₆, with W–C bond lengths around 2.06 Å and all-90°/180° C–W–C angles confirmed by X-ray diffraction. As a sanity check on the electron count itself: EC $= 6 + 2(6) = 18$. ✓.

Example 2 (applied): Rationalize why Fe(CO)₅ reacts with two equivalents of Na/Hg (sodium amalgam) to give Na₂[Fe(CO)₄], and why the product is tetrahedral, not trigonal bipyramidal. Sodium amalgam is a strong one-electron reductant. Adding two electrons to neutral Fe(CO)₅ (18 electrons, EC $= 8 + 10 = 18$) while simultaneously losing one CO ligand — a common reaction pathway for 18-electron carbonyls under reduction, since a full 18-electron 5-coordinate complex cannot simply accept extra electrons without violating the rule (that would give a 20-electron species, which is disallowed for a first-row metal without very unusual ligands) — gives the anion [Fe(CO)₄]²⁻. Recompute the electron count for this new species directly. The iron center is now formally Fe(−II) (each of the 2 extra electrons corresponds to a unit of negative charge localized formally on the metal in the ionic counting convention), so the d-electron count at the metal is $g - (\text{oxidation state}) = 8 - (-2) = 10$ d-electrons contributed by the metal itself, plus four CO ligands at 2 electrons each:

So [Fe(CO)₄]²⁻ is indeed an 18-electron species — consistent with its known existence and stability as the isolable disodium salt Na₂[Fe(CO)₄] (Collman's reagent), an important stoichiometric reagent in organic synthesis for converting alkyl halides into aldehydes and ketones. With four ligands around the metal and no stereochemically active lone pairs, the geometry-minimization argument (Step 4 above) predicts a tetrahedral anion, matching the observed tetrahedral [Fe(CO)₄]²⁻ structure — the same reasoning that gives Ni(CO)₄ its shape, since both are 4-coordinate, 18-electron, d10-configured centers (Ni(0) is d10, and Fe(−II) is likewise formally d10). This is a clean illustration that geometry tracks ligand count, not the identity or oxidation state of the metal, once the electron count is fixed at 18.

Example 3 (exam/competition level): A first-row transition metal carbonyl anion [M(CO)₄]⁻ is found to be a stable, diamagnetic, tetrahedral 18-electron species. Identify M, and then predict the structure and electron count of the neutral radical M(CO)₄· obtained by one-electron oxidation, commenting on its expected stability relative to the parent anion. Set up the electron-count equation for the anion, with unknown group number $g$ and one unit of negative charge (contributing one extra electron to the ionic-convention count):

where the $+1$ at the end accounts for the single negative charge (one extra electron localized formally at the metal in the ionic counting scheme, exactly as in Example 2 but here for a monoanion rather than dianion). Solving,

Group 9 in the first row is cobalt. So $M=\text{Co}$, and the species is $[\text{Co(CO)}_4]^-$ — a real, well-known 18-electron, tetrahedral, diamagnetic anion (isolable as, e.g., the Na⁺ or K⁺ salt, and a classic nucleophile in organometallic substitution chemistry, being isoelectronic with Ni(CO)₄ since Co(−I) is d10 just as Ni(0) is d10). Now remove one electron (one-electron chemical or electrochemical oxidation) to give neutral Co(CO)₄·. The electron count drops by exactly 1:

This is a 17-electron radical — an odd-electron species that, by the 18-electron rule's own logic, is expected to be coordinatively and electronically unsaturated and highly reactive: it should seek to reach 18 electrons either by dimerizing through formation of a Co–Co bond (each monomer contributing one electron to a shared covalent bond, restoring $17+1=18$ at each center) or by picking up an additional 2-electron donor ligand. Indeed, this is exactly the logic behind the existence of Co₂(CO)₈ as the thermodynamically stable binary cobalt carbonyl rather than a mononuclear Co(CO)₄ radical: two "Co(CO)₄·" 17-electron fragments combine, each losing one CO along the way in some structural isomers, to give the observed dimer with a Co–Co bond (and, in the bridged isomer, two bridging carbonyls) that restores the 18-electron count at each metal. The one-electron oxidation product of the stable anion is thus predicted — correctly — to be thermodynamically unstable as a monomer and to dimerize spontaneously, which is precisely the behavior established experimentally for the Co(CO)₄· radical fragment as a transient intermediate in cobalt carbonyl chemistry.

Ludwig Mond was a German-born industrial chemist who spent most of his career in England, where he co-founded the alkali manufacturer Brunner Mond (a forerunner of ICI) and made a fortune from the Solway soda-ash process before turning to nickel refining. In 1890, working with his assistants Carl Langer and Friedrich Quincke, Mond discovered that impure nickel exposed to carbon monoxide gas at modest temperature spontaneously formed a volatile liquid — Ni(CO)₄ — which decomposed back to pure metallic nickel and CO gas on heating to around 180 °C. This reversible reaction became the basis of the Mond process, an industrial method for purifying nickel ore that is still used today: crude nickel is treated with CO at roughly 50–60 °C to form the volatile carbonyl, which is distilled away from involatile impurities and then thermally decomposed at higher temperature to deposit ultra-pure nickel metal. Mond's discovery is historically the very first binary metal carbonyl ever isolated, and it opened an entirely new branch of chemistry in which metals in low, even zero, oxidation states could be stabilized by π-acid ligands — a discovery whose electronic rationale (the synergic bonding picture and the 18-electron rule) would only be fully worked out decades later, chiefly through the efforts of chemists including Walter Hieber (who systematically studied Fe, Co, and Mn carbonyls in the 1920s–1930s) and, later, the broader development of ligand-field and molecular-orbital theory applied to organometallics.