Synergic Bonding in Metal Carbonyls

The synergic bonding model is not a historical curiosity; it is the everyday working language of organometallic and catalytic chemistry. In homogeneous catalysis, the electron density delivered to a metal center by ancillary ligands (phosphines, N-heterocyclic carbenes) is routinely quantified using the Tolman electronic parameter, which is defined using the ν(CO) shift of Ni(CO)₃L complexes — a direct, industrial-scale application of the exact bond-order argument developed above, used to rank ligands for hydroformylation, cross-coupling, and olefin polymerization catalyst design. In bioinorganic chemistry, backbonding explains why CO, CN⁻, and NO bind so tightly (and dangerously, in the case of CO poisoning) to the reduced iron centers of hemoglobin and myoglobin — the same π-acceptor logic that stabilizes Fe(CO)₅ stabilizes the toxic Fe(II)–CO adduct in heme proteins, and distinguishing physiological O₂ binding from pathological CO binding is partly a question of relative backbonding and steric discrimination by the distal histidine. In materials and surface chemistry, CO chemisorption on transition-metal surfaces (a cornerstone of heterogeneous catalysis, e.g., in automotive catalytic converters and Fischer–Tropsch synthesis) is analyzed with the same synergic donor/acceptor framework extended from molecular orbitals to surface d-bands (the Blyholder model, a direct surface-science descendant of Dewar–Chatt–Duncanson). In spectroscopic instrumentation, ν(CO) tracking by FT-IR remains the single most common, fastest diagnostic for monitoring the progress and electronic character of organometallic reactions in a standard synthetic laboratory, feeding directly into the broader toolkit developed in Infrared Spectroscopy of Carbonyl Ligands. Finally, the entire conceptual apparatus generalizes without modification to other π-acceptor ligands — isocyanides, phosphines, phosphites, N₂, NO⁺, alkenes, alkynes, and N-heterocyclic carbenes — organized and compared systematically in Isoelectronic and Related Pi-Acceptor Ligands, making the synergic model one of the most exported ideas in all of inorganic chemistry.

Ask the tutor: How does the synergic bonding model change when CO is replaced by a stronger π-acid like NO⁺ or a weaker one like a phosphine, and how do we quantify that difference experimentally?

Carbon monoxide's molecular orbital diagram is the starting point for everything that follows, because the identity of the frontier orbitals dictates both what CO can accept and what it can donate. Combining carbon 2s/2p and oxygen 2s/2p atomic orbitals under C∞v symmetry produces (in order of increasing energy, ignoring core 1s orbitals) a set of σ and π molecular orbitals: 3σ (bonding, largely O 2s/2p character mixed with C 2s), 1π (a degenerate bonding pair from C 2p and O 2p, perpendicular to the internuclear axis), 4σ — the HOMO, and 2π — the degenerate LUMO pair, followed at higher energy by the strongly antibonding 5σ. Because O is more electronegative, oxygen atomic orbitals sit lower in energy; the bonding combinations (3σ, 1π) are pulled toward O, while the antibonding partners (4σ HOMO, 2π LUMO) are pushed toward and concentrated on carbon. The 4σ HOMO, despite formally being an antibonding combination of s/p hybrids along the axis, is only weakly antibonding overall (it also has substantial nonbonding lone-pair character localized on carbon), which is why CO's own bond order remains 3 and its bond is so strong (1072 kJ/mol, the strongest known in a neutral diatomic). This same 4σ orbital, however, is the orbital CO uses to bind metals: it points outward from carbon, away from oxygen, exactly the right shape and location to overlap with an empty metal orbital along the M–C axis. The 2π LUMOs are the acceptor orbitals. They are antibonding with respect to C–O, degenerate, oriented perpendicular to the C–O axis, and — crucially — have larger coefficients on carbon than on oxygen (a consequence of the same electronegativity argument that shaped 4σ). This carbon-heavy character means the π orbitals overlap well with metal d orbitals that are also directed toward carbon, i.e., the filled dπ orbitals (dxz, dyz in the standard octahedral frame with the M–C bond along z) whose lobes lie in planes containing the M–C axis. Populating 2π by donation from the metal necessarily weakens the C–O bond (since 2π is C–O antibonding) while it simultaneously strengthens the M–C bond (since the same orbital overlap is M–C bonding in character). Put the two pieces together and you get the synergic bonding picture: CO acts as a σ donor through 4σ into an empty metal orbital of appropriate symmetry (dz², dsp hybrids, or simply an empty s/p/d hybrid depending on the metal's electron count and geometry), and simultaneously as a π acceptor, receiving density from filled metal dπ orbitals into its 2π pair. Because both processes occur along/around the same M–C axis and reinforce rather than compete with each other electronically, the resulting bond is unusually strong for what is nominally a "weak" ligand donating through a marginally antibonding HOMO.

The second piece of the physical picture is how these CO orbitals fit into a metal's d-orbital manifold once coordinated, using the octahedral frame as the reference case (as in Cr(CO)₆ or the M(CO)₆ ions). Take the M–C bond along z. The empty metal orbital that accepts the CO lone pair is a σ-symmetry combination — in a filled-shell d⁶ octahedral complex this is often described using an eg-derived hybrid or, more simply, the metal 4s/4p character mixed with dz² that points toward the ligand. The filled metal orbitals of π symmetry with respect to that same axis are dxz and dyz (the t2g set in Oh symmetry), which are non-bonding toward σ-only ligands but become strongly stabilized when they can donate into CO's 2π*. This is exactly the situation that makes CO special among simple ligands: unlike NH₃ or H₂O, which have no low-lying empty orbitals of π symmetry, CO offers the metal's "extra," otherwise non-bonding t2g electrons somewhere productive to go.

A third, quantitative dimension of the picture concerns bond order and vibrational frequency. Free CO has a formal bond order of 3 (one σ + two π bonds, using the simple Lewis/MO correlation), a bond length of 112.8 pm, and a stretching frequency of 2143 cm⁻¹. Any electron density placed in 2π* — regardless of its source — lowers the C–O bond order below 3, lengthens the bond, and lowers the force constant, hence lowers ν(CO). σ donation alone (removing density from a weakly antibonding orbital) has only a small, secondary effect on ν(CO); the dominant, chemically diagnostic effect is π backbonding. This is why infrared spectroscopy of the carbonyl stretch is used throughout organometallic chemistry as a direct, quantitative probe of the electron density available at the metal for backbonding — more backbonding correlates with more electron-rich metals (more negative charge, more reducing ancillary ligands, lower oxidation states), and shows up as lower ν(CO).

The essential relationships governing synergic bonding can be organized around three linked ideas: orbital symmetry matching, the bond-order/force-constant connection, and the electronic feedback (synergy) itself.

Symmetry matching condition. For a metal dπ orbital to backbond into CO's 2π*, the two must transform as the same irreducible representation of the local M–C–O symmetry (C∞v, reducible to whatever point group the whole complex has). In an octahedral ML₅(CO) fragment with the M–C axis as z,

where:

• $d_{xz}, d_{yz}$ are the filled, non-bonding (in a purely σ-donor picture) metal orbitals in the t2g set
• $2\pi^*_x, 2\pi^*_y$ are the two orthogonal, degenerate antibonding π* orbitals of CO
• the equality of representations is the symmetry requirement for nonzero orbital overlap and hence nonzero backbonding

Reading: this is simply the group-theoretic statement that "only orbitals of matching symmetry and similar energy mix." It tells you immediately why π-symmetry metal orbitals (not the σ-symmetry eg set) are the ones responsible for backbonding, and why ligands with no accessible low-lying π* orbitals (like NH₃) cannot participate in this second half of the synergic cycle at all.

Bond order / force constant relationship. Within the harmonic approximation, the C–O stretching frequency is set by the force constant and reduced mass:

where:

• $\tilde{\nu}_{\text{CO}}$ is the observed IR wavenumber (cm⁻¹)
• $c$ is the speed of light
• $k_{\text{CO}}$ is the C–O bond force constant (N/m), which tracks bond order
• $\mu_{\text{CO}} = \dfrac{m_C m_O}{m_C + m_O}$ is the reduced mass of the C–O oscillator (fixed for all ¹²C¹⁶O species)

Reading: since $\mu_{\text{CO}}$ is essentially constant across all metal carbonyls (isotopic substitution aside), $\tilde{\nu}_{\text{CO}}$ is a direct, monotonic proxy for $k_{\text{CO}}$, which in turn tracks C–O bond order. Populating 2π* through backbonding lowers the bond order below 3, lowers $k_{\text{CO}}$, and lowers $\tilde{\nu}_{\text{CO}}$ below free CO's 2143 cm⁻¹. This equation is the bridge between the qualitative MO argument and a number you can measure on a benchtop FT-IR spectrometer.

Synergic feedback (qualitative charge-balance statement). Let $q_\sigma$ be the fractional electron density donated from CO's 4σ into the metal, and $q_\pi$ the fractional electron density back-donated from the metal into CO's 2π*. The net charge transferred to the metal is

where:

• $q_\sigma > 0$ always (CO is always a net σ donor in stable complexes)
• $q_\pi > 0$ whenever the metal has filled dπ orbitals of the right symmetry and energy
• $\Delta q_M$ is the net charge accumulated on the metal from this ligand's bonding

Reading: if $q_\pi$ were zero (no backbonding, as would be the case for a d⁰ metal with no filled dπ electrons), σ donation alone would pile positive-compensating electron density onto the metal every time a CO binds, making each successive CO bind more weakly (electrostatically unfavorable) — this is why simple hard, high-oxidation-state, d⁰ or d⁰-like metal centers form few or no stable homoleptic carbonyls. Backbonding drains this density back out through the π* channel, keeping the metal near-neutral and letting σ donation continue efficiently — the literal meaning of "synergic." Both $q_\sigma$ and $q_\pi$ respond to the same driving force (more backbonding lowers the energy gap resisting further σ donation, and vice versa), which is the mechanistic content of the feedback loop.

Example 1 (Foundational): Assigning bond order and predicting the direction of the ν(CO) shift in Cr(CO)₆. Chromium in Cr(CO)₆ is Cr(0), configuration $3d^6$ in an octahedral field, low-spin $t_{2g}^6 e_g^0$. All six t2g electrons are available, in principle, for backbonding into the six CO ligands' 2π orbitals (with the appropriate symmetry-adapted linear combinations required by Oh symmetry). This is a d⁶, electron-rich, neutral, zero-oxidation-state metal center — exactly the profile that maximizes backbonding. Predicted consequence: substantial population of 2π, so the C–O bond order should sit measurably below 3, and ν(CO) should be well below the free-CO value of 2143 cm⁻¹. Experimental value: Cr(CO)₆ shows its main IR-active $t_{1u}$ C–O stretch at 2000 cm⁻¹ (in the gas/solution IR spectrum; solid-state and different symmetry components vary slightly but remain close to this value). Check the direction and magnitude: $2143 - 2000 = 143~\text{cm}^{-1}$ downshift, entirely consistent with strong backbonding from a d⁶ neutral, zero-oxidation-state metal with full access to its t2g electrons. This is the qualitative benchmark reference point against which more/less electron-rich carbonyls are compared.

Example 2 (Applied): Ranking ν(CO) across an isoelectronic series and extracting the electronic trend. Consider the isoelectronic series $[\text{Mn(CO)}_6]^+$, $\text{Cr(CO)}_6$, and $[\text{V(CO)}_6]^-$ — all d⁶, octahedral, 18-electron, differing only by nuclear charge and overall ionic charge (+1, 0, −1 respectively). Nuclear charge increases along Mn⁺ < Cr⁰ < ... wait — arrange correctly by charge: V(−1) is most electron-rich, Cr(0) is intermediate, Mn(+1) is most electron-poor. Reasoning: increasing positive charge on the metal contracts the d orbitals and lowers their energy, both of which reduce the metal's ability to backbond (contracted d orbitals overlap less well with the diffuse, higher-energy CO π, and lowered d-orbital energy increases the energy gap to 2π, reducing the driving force for donation). Less backbonding means less population of the C–O antibonding 2π, hence a higher C–O bond order and a <strong>higher</strong> ν(CO). Predicted order: $\tilde{\nu}_{\text{CO}}\big([\text{V(CO)}_6]^-\big) < \tilde{\nu}_{\text{CO}}\big(\text{Cr(CO)}_6\big) < \tilde{\nu}_{\text{CO}}\big([\text{Mn(CO)}_6]^+\big)$ Experimental values confirm this: $[\text{V(CO)}_6]^-$, ν(CO) ≈ 1859 cm⁻¹; Cr(CO)₆, ν(CO) ≈ 2000 cm⁻¹; $[\text{Mn(CO)}_6]^+$, ν(CO) ≈ 2090 cm⁻¹. Numerical check of trend consistency: going from V(−1) to Cr(0), $\Delta\tilde\nu = 2000 - 1859 = 141~\text{cm}^{-1}$; going from Cr(0) to Mn(+1), $\Delta\tilde\nu = 2090 - 2000 = 90~\text{cm}^{-1}$. Both increments are positive and of comparable (though not identical) magnitude, confirming a smooth, monotonic dependence of backbonding capacity on formal metal charge across an isoelectronic, isostructural series — exactly the trend the synergic model predicts, with the somewhat smaller second increment reflecting the diminishing marginal effect of adding further positive charge (each additional unit of positive charge produces a progressively smaller relative* contraction of orbitals already fairly contracted).

Example 3 (Exam/competition level): Estimating relative M–C and C–O bond orders in a mixed-ligand carbonyl using ν(CO) data, and predicting structural consequences. Consider $\text{Mn(CO)}_5\text{Br}$, an octahedral $d^6$ Mn(I) complex, versus $[\text{Mn(CO)}_6]^+$ (also d⁶, Mn(I)... check oxidation state: in $[\text{Mn(CO)}_6]^+$ the overall charge is +1 with six neutral CO ligands, so Mn is formally +1, i.e., Mn(I), d⁶, matching Mn(CO)₅Br where Br⁻ is a one-electron-pair (X-type, −1) ligand and Mn is again formally Mn(I), d⁶). Both complexes have the same formal metal oxidation state (Mn(I), d⁶) and the same overall ionic charge is different: $[\text{Mn(CO)}_6]^+$ carries a +1 overall charge on the whole ion, while Mn(CO)₅Br is a neutral molecule (charge balanced internally by Br⁻). The question: which complex has the lower ν(CO), and why, given identical formal metal oxidation state? Step 1 — Identify the operative difference. Formal oxidation state counts electrons the same way in both cases, but it does not measure the actual electron density at the metal nucleus, which is what governs backbonding. In Mn(CO)₅Br, one CO has been replaced by Br⁻, a strong σ-donor, weak π-acceptor (halide lone pairs can donate into empty metal orbitals or even weakly π-donate into partially filled sets, but Br⁻ has no low-lying π acceptor orbital of its own). Br⁻ therefore does not compete with the remaining five CO ligands for the metal's backbonding electron density. Step 2 — Compare total positive charge experienced at the metal. In $[\text{Mn(CO)}_6]^+$, the whole +1 charge is spread over a metal center whose only ligands are six weak-donor, strong π-acceptor CO groups — none of which can supply extra σ density to compensate the overall cationic charge beyond their intrinsic (and by MO argument, modest) 4σ donation. In Mn(CO)₅Br, the anionic bromide ligand supplies substantial extra electron density directly to the metal (both through its own strong σ donation and by being a formally anionic ligand that need not "make room" for backbonding electron traffic to five other identical acceptors), leaving the metal significantly more electron-rich overall, i.e., more able to backbond into each remaining CO. Step 3 — Predict and check ν(CO). More electron density at Mn in Mn(CO)₅Br means more backbonding per CO ligand than in $[\text{Mn(CO)}_6]^+$, so ν(CO) for Mn(CO)₅Br should be substantially lower than for $[\text{Mn(CO)}_6]^+$ (2090 cm⁻¹), despite both being formally Mn(I), d⁶. Experimental comparison: Mn(CO)₅Br shows C–O stretches clustered around 2138, 2073, 2011 cm⁻¹ (several IR-active modes due to lower, $C_{4v}$, symmetry splitting the former $t_{1u}$ set); the lowest and most diagnostic band sits well below 2090 cm⁻¹, confirming markedly greater backbonding than in the cationic hexacarbonyl, even though the formal* oxidation state count (a bookkeeping device) is identical in both species. Step 4 — Structural consequence. Greater backbonding in Mn(CO)₅Br also means shorter, stronger M–C bonds and, correspondingly, slightly longer C–O bonds relative to $[\text{Mn(CO)}_6]^+$ — a prediction directly testable by X-ray or electron diffraction bond-length data, illustrating that ν(CO) is tracking real, measurable differences in bond order and bond length, not merely a spectroscopic curiosity.