Ionic and Covalent Electron Counting Methods

Take ferrocene, Fe(C₅H₅)₂, and ask a simple question: how many electrons sit around the iron? Depending on which bookkeeping convention you use, you might say the iron is Fe(0). This would pair it with two neutral 5-electron cyclopentadienyl radicals. Alternatively, you might say it is Fe(II). This would pair it with two anionic 6-electron cyclopentadienide ligands. Both conventions predict the same total, 18 electrons. Both correctly explain why ferrocene is a stable, isolable, diamagnetic sandwich compound. This apparent contradiction — two different oxidation states for the same real molecule — is not a paradox to be resolved. It is a feature to be understood. Oxidation state and electron count are bookkeeping devices, not experimental observables. Organometallic chemists have built two self-consistent formalisms for doing that bookkeeping. This matters enormously in practice. The 18-Electron Rule is the single most useful predictive tool in organometallic chemistry. It rationalizes the stoichiometry of Fe(CO)₅, Cr(CO)₆, and Ni(CO)₄. It predicts why Mn(CO)₅ dimerizes to Mn₂(CO)₁₀. It tells you which carbonyl anion, [Co(CO)₄]⁻ or [Mn(CO)₅]⁻, is the stable one. But the rule is only as good as the electron count feeding into it. There are exactly two standard ways to generate that count: the covalent (neutral) method and the ionic (donor-pair) method. Students who mix the two get wrong answers. This happens when they assign a ligand's electron donation as if it were anionic, while also assigning the metal its neutral-atom electron count. The result is wrong oxidation states, wrong d-electron counts, and wrong predictions about reactivity. The big picture is this: every ligand in a complex can be described in either of two electron-donation conventions. Every metal, correspondingly, can be described in either of two reference states. But the sum — metal electrons plus ligand electrons — is invariant under the choice of convention. This holds provided the bookkeeping is applied consistently and completely. This invariance is not an accident. It follows from the fact that both methods start from the same total number of electrons in the real molecule. They merely redistribute them formally between metal and ligands. This is done using a "virtual heterolysis" (ionic) or "virtual homolysis" (covalent) of each metal–ligand bond. This article builds both methods from the ground up. It works through the arithmetic in full for a spread of real complexes — CO, cyclopentadienyl, alkyl, alkene, nitrosyl, and alkylidene ligands. It shows exactly where students go wrong (the NO problem, the allyl-hapticity problem, the "same ligand, different donor number" problem). It closes by connecting electron counting to the ideas that follow from it directly. These are ligand classification into L, X, and Z types, formal oxidation state, d-electron configuration, and coordination number. All of these are different views of the same bookkeeping.

Key takeaways

Two self-consistent conventions exist for counting valence electrons at a metal: the covalent (neutral atom) method and the ionic (donor-pair) method.
Covalent method: split every M–L bond homolytically; ligands donate as neutral radicals (CO donates 2e, Cl donates 1e, Cp donates 5e, H donates 1e).
Ionic method: split every M–L bond heterolytically; ligands donate as closed-shell ions (Cl⁻ donates 2e, Cp⁻ donates 6e, H⁻ donates 2e), and the metal is assigned a formal oxidation state accordingly.
Both methods must give the same total electron count for the same real molecule — this is the built-in consistency check.
The ionic method directly yields the metal's formal oxidation state and dⁿ configuration; the covalent method does not, and should not be used to assign oxidation state.
Key idea: electron counting is a bookkeeping convention applied to a fixed total; covalent and ionic methods redistribute the same electrons differently between metal and ligand but must always reconcile to one grand total.

Where this lives today

Electron counting by both methods is the daily working tool of synthetic organometallic and catalytic chemists. In homogeneous catalysis, mechanistic cycles for hydrogenation, hydroformylation, cross-coupling, and olefin polymerization are drawn as sequences of 16- and 18-electron intermediates, and the ionic method is what lets a chemist track the metal's oxidation state through oxidative addition, reductive elimination, migratory insertion, and β-hydride elimination steps — each of which changes electron count and/or oxidation state in a diagnostic, predictable way (oxidative addition: +2 electrons, +2 oxidation state; reductive elimination: −2 electrons, −2 oxidation state; migratory insertion: no change in either; ligand dissociation: −2 electrons for an L-type ligand leaving, oxidation state unchanged). In industrial catalysis, Wilkinson's catalyst [RhCl(PPh₃)₃] (16e, Rh(I), d⁸) and Vaska's complex illustrate exactly the unsaturation logic developed above, and the same bookkeeping underlies the design of Grubbs and Schrock olefin-metathesis catalysts, where alkylidene (=CR₂) ligands are formally dianionic Z-type/X₂-type donors in the ionic scheme, giving high, well-defined oxidation states (Mo(VI), Ru(II)) essential to correctly predicting d-electron count and spin state. In bioinorganic chemistry, iron–sulfur clusters and nitrogenase's FeMo-cofactor are analyzed with the same ionic-vs-covalent duality to debate formal oxidation states of iron and molybdenum centers whose true electronic structure is highly covalent and delocalized — a modern echo of the ferrocene ambiguity that opened this article. In materials chemistry, metal–organic frameworks and single-atom catalysts are characterized using the same electron-counting logic to rationalize coordination geometry and predict open coordination sites for substrate binding. And in the classroom and in computational cheminformatics (SMILES/ligand-typing software for reaction databases), the L/X/Z (covalent bond classification) notation is the standard machine-readable encoding of exactly the dual bookkeeping presented here.

:::quiz[For [Mn(CO)₅]⁻, what oxidation state does the ionic method assign to manganese?]

Mn(0)
Mn(+1)
Mn(−1) ✓
Mn(+5)

Ask the tutor: How do you assign the ionic donor number and oxidation-state contribution for an η²-bound alkyne or a bridging carbonyl ligand shared between two metals?

The physical picture

A transition-metal complex is held together by a set of metal–ligand bonds, each of which is, physically, a two-electron covalent interaction of variable polarity — from nearly covalent (M–alkyl) to strongly polarized (M–halide) to almost purely dative (M←CO). Because the polarity of each bond is a matter of degree, not of kind, there is no unique, physically forced way to decide "how many electrons the ligand brought" versus "how many the metal brought." What is fixed is only the total number of electrons in the bonding + nonbonding valence shell around the metal — this total is what the 18-electron rule tests, and it is invariant to how you partition it. The covalent method partitions bonds by homolysis: cut every M–L bond down the middle, giving one electron to the metal and one electron to the ligand fragment, exactly as you would when computing bond dissociation energies or radical stabilities in organic chemistry. Under this partition the metal is left in oxidation state 0, contributing its full periodic-table valence-electron count (equal to its group number, using the modern IUPAC numbering where Sc–Zn is group 3–12), and each ligand is described as the neutral radical fragment it would be if the bond broke homolytically. A methyl ligand contributes 1 electron (as •CH₃ does), CO contributes 2 (as :C≡O: does, donating its lone pair while remaining formally neutral), Cl contributes 1 (as •Cl does), and cyclopentadienyl contributes 5 (as the cyclopentadienyl radical does). The ionic method partitions bonds by heterolysis: cut every M–L bond so that both electrons go to the more electronegative fragment, exactly as you would when assigning formal oxidation states in Werner-type coordination chemistry. Under this partition, anionic ligands (Cl⁻, CH₃⁻, Cp⁻, H⁻) take both bonding electrons and are counted as closed-shell anions; neutral two-electron donors (CO, PR₃, NH₃, alkenes) are unaffected because they were already neutral dative donors; and the metal is left with a positive formal charge equal to the sum of the negative charges removed from the ligands. This positive charge is, by definition, the oxidation state of the metal. The metal's electron count is then read off not from the neutral atom but from the appropriate d-electron configuration for that oxidation state, i.e., (group number) − (oxidation state), because removing electrons as "ligand charge" formally removes them from the metal's own count.

Figure

Two ways to split the same Fe–Cp bond: homolytic (covalent) vs heterolytic (ionic)

Both routes land on 18. That reconciliation is not a coincidence to be marveled at every time — it is the mathematical guarantee that makes the two methods interchangeable, and it is exactly the check you should run on every unfamiliar complex.

Figure

Ligand donor numbers under the two conventions

The pattern in that table is the entire content of the "translation dictionary" between the two methods: for a purely σ-donor X-type ligand (one that is a radical when neutral: H, Cl, CH₃, Cp when treated monohapto-like, aryl), the ionic count is always one electron more than the covalent count, because the ionic method also removes one electron from the metal (raising its oxidation state by one) and gives both to the ligand. For a purely dative L-type ligand (CO, PR₃, NH₃, alkenes, arenes as neutral 6e donors) the two counts are identical, because there was never any charge to redistribute — the ligand was neutral and closed-shell (or effectively so, donating a lone pair or π system) under both conventions.

The core ideas

The master statement, valid for any mononuclear complex $[\mathrm{ML}_n]^{q}$ with overall charge $q$ , is:

N_{\text{total}} = g_M \;-\; q \;+\; \sum_i d_i^{\text{(cov)}}

where:

$N_{\text{total}}$ is the total valence electron count at the metal (the number tested against 18),
$g_M$ is the metal's group number (periodic table group, e.g. Fe = 8, Mn = 7, Ni = 10),
$q$ is the overall charge on the complex ion (0 for a neutral complex),
$d_i^{\text{(cov)}}$ is the covalent (neutral-radical) donor number of ligand $i$ (H = 1, CO = 2, Cp = 5, alkene = 2, etc.),
the sum runs over all ligands bound to the metal.

This is the covalent method formula: metal contributes its neutral group-number electron count, corrected for overall ionic charge, and every ligand is counted as a neutral radical fragment. The equivalent ionic method formula is:

N_{\text{total}} = \big(g_M - \mathrm{OS}\big) \;+\; \sum_i d_i^{\text{(ion)}}

where:

$\mathrm{OS}$ is the formal oxidation state of the metal, obtained by summing the charges assigned to all ligands (as their closed-shell ionic forms) and subtracting from the overall complex charge $q$ : $\mathrm{OS} = q - \sum_i c_i$ , with $c_i$ the ionic charge assigned to ligand $i$ ,
$g_M - \mathrm{OS}$ is the metal's dⁿ configuration electron count (this is exactly the number of d electrons formally left on the metal),
$d_i^{\text{(ion)}}$ is the ionic (donor-pair) donor number of ligand $i$ (H⁻ = 2, Cl⁻ = 2, Cp⁻ = 6, neutral L-type ligands unchanged),
the sum again runs over all ligands.

Reading: the covalent formula never mentions oxidation state at all — it is a "radical bookkeeping" that is fast and mechanical, ideal for getting the 18-electron count quickly. The ionic formula is slightly more work (you must first decide each ligand's ionic charge and sum them to get OS) but it is the formula that simultaneously delivers the oxidation state and the dⁿ configuration, which is why it is preferred whenever you need to reason about redox chemistry, ligand-field splitting, or spin state. Both formulas must give the same $N_{\text{total}}$ for the same molecule; that equality is the built-in self-check.

Derivation AWhy the two totals must always agree

Goal. Show algebraically that the covalent and ionic electron counts are identical for any complex, given consistent ligand-charge assignment — i.e., prove $N_{\text{total}}^{\text{cov}} = N_{\text{total}}^{\text{ion}}$ always.

Step 1 — relate the two donor numbers ligand by ligand. For any ligand $i$ that is assigned ionic charge $c_i$ (negative for anionic ligands like Cl⁻, zero for neutral L-type ligands like CO), the ionic donor count and covalent donor count are related by

d_i^{\text{(ion)}} = d_i^{\text{(cov)}} + |c_i|

because converting a neutral radical fragment (covalent picture) into a closed-shell anion (ionic picture) requires giving it $|c_i|$ additional electrons, taken from the metal. For a monoanionic ligand ( $c_i = -1$ , e.g. Cl, CH₃, Cp when it counts as X-type-like in charge terms), $|c_i| = 1$ , matching the "+1" pattern seen in the table above. For a neutral ligand ( $c_i = 0$ , e.g. CO, PR₃) the two donor numbers are equal.

Step 2 — sum over all ligands. Summing Step 1 over the whole ligand set:

\sum_i d_i^{\text{(ion)}} = \sum_i d_i^{\text{(cov)}} + \sum_i |c_i|

Since every $c_i \le 0$ (ligands are assigned negative or zero ionic charge by convention — never positive, except for the rare formally cationic NO⁺-type ligand handled separately), $\sum_i |c_i| = -\sum_i c_i$ .

Step 3 — relate metal electron counts. The oxidation state is defined as $\mathrm{OS} = q - \sum_i c_i$ , so $-\sum_i c_i = \mathrm{OS} - q$ . The metal's ionic-picture electron count is $g_M - \mathrm{OS}$ , while its covalent-picture count is $g_M - q$ (a neutral complex, $q=0$ , simply gives the neutral atom; a charged complex shifts the metal's assigned electron count by the overall charge in the covalent scheme, since ligands remain neutral radicals and any overall ionic charge must be carried by the metal fragment). Subtracting,

(g_M - q) - (g_M - \mathrm{OS}) = \mathrm{OS} - q = -\sum_i c_i

which is exactly $\sum_i |c_i|$ from Step 2.

Step 4 — add metal and ligand contributions. Now form the full totals:

N^{\text{cov}} = (g_M - q) + \sum_i d_i^{\text{(cov)}}, \qquad N^{\text{ion}} = (g_M-\mathrm{OS}) + \sum_i d_i^{\text{(ion)}}

Substitute Step 2 into $N^{\text{ion}}$ :

N^{\text{ion}} = (g_M - \mathrm{OS}) + \sum_i d_i^{\text{(cov)}} + \sum_i|c_i|

and use Step 3, $\sum_i |c_i| = (g_M - q) - (g_M-\mathrm{OS})$ , to substitute:

N^{\text{ion}} = (g_M-\mathrm{OS}) + \sum_i d_i^{\text{(cov)}} + (g_M - q) - (g_M - \mathrm{OS}) = (g_M - q) + \sum_i d_i^{\text{(cov)}} = N^{\text{cov}}

The oxidation-state terms cancel exactly, leaving $N^{\text{ion}} = N^{\text{cov}}$ for every complex, regardless of how many anionic ligands are present or what the overall charge is. This is the algebraic content of "both methods give the same 18-electron count" — it is forced by the definitions, not an empirical coincidence.

:::derivation[Worked general case: an asymmetric mixed-ligand complex, [Mo(η⁵-C₅H₅)(CO)₃CH₃]] Goal. Apply both counting formulas in full to a genuinely mixed complex — one η⁵-Cp, three CO, one methyl on molybdenum — and confirm the totals agree, illustrating every term of the two master formulas with non-trivial (non-symmetric) ligand content.

Step 1 — covalent method setup. Molybdenum is in group 6, so $g_{Mo} = 6$ . The complex is neutral, $q = 0$ . List covalent (neutral-radical) donor numbers: Cp (as •C₅H₅ radical) donates 5e; each CO donates 2e (three of them, $3\times 2 = 6$ e); CH₃ (as •CH₃ radical) donates 1e.

Step 2 — sum the covalent count.

N^{\text{cov}} = g_{Mo} - q + \big[d_{\text{Cp}}^{\text{cov}} + 3\,d_{\text{CO}}^{\text{cov}} + d_{\text{CH}_3}^{\text{cov}}\big] = 6 - 0 + \big[5 + 3(2) + 1\big]

N^{\text{cov}} = 6 + [5+6+1] = 6+12 = 18

Step 3 — ionic method setup. Assign ionic charges: Cp is treated as Cp⁻ (charge $c=-1$ , donor number 6e), CO remains neutral L-type ( $c=0$ , donor number 2e each), CH₃ is treated as CH₃⁻ (charge $c=-1$ , donor number 2e). Sum the ligand charges: $\sum_i c_i = (-1) + 0 + 0 + 0 + (-1) = -2$ .

Step 4 — compute oxidation state. Using $\mathrm{OS} = q - \sum_i c_i$ :

\mathrm{OS} = 0 - (-2) = +2

So this is a Mo(II) complex. The metal's ionic-picture electron count (its dⁿ configuration) is $g_{Mo} - \mathrm{OS} = 6-2 = 4$ , i.e. a d⁴ metal center.

Step 5 — sum the ionic count.

N^{\text{ion}} = (g_{Mo}-\mathrm{OS}) + \big[d_{\text{Cp}}^{\text{ion}} + 3\,d_{\text{CO}}^{\text{ion}} + d_{\text{CH}_3}^{\text{ion}}\big] = 4 + [6 + 3(2) + 2]

N^{\text{ion}} = 4 + [6+6+2] = 4+14 = 18

Both routes give exactly 18 electrons, confirming Cp(CO)₃MoCH₃ is an 18-electron, coordinatively saturated, Mo(II) d⁴ complex, consistent with its known stability as a robust, isolable half-sandwich alkyl. The ionic route additionally hands you, for free, the oxidation state (+2) and d-count (d⁴) that the covalent route cannot directly supply.

Worked examples

Example 1 (foundational): Cr(CO)₆, the textbook octahedral carbonyl. Chromium is group 6, $g_{Cr}=6$ . The complex is neutral. All six ligands are CO, a purely neutral, L-type, 2-electron σ-donor/π-acceptor in both conventions (CO has no charge to redistribute — carbon monoxide is not a radical or an ion in either bookkeeping). Covalent method:

N^{\text{cov}} = 6 - 0 + 6(2) = 6+12 = 18

Ionic method: no ligand carries ionic charge, so $\sum c_i = 0$ , giving $\mathrm{OS} = 0 - 0 = 0$ , i.e. Cr(0), $d^{6}$ ( $g_{Cr}-\mathrm{OS}=6-0=6$ ).

N^{\text{ion}} = (6-0) + 6(2) = 6+12=18

Both give 18, Cr(0) d⁶, matching the well-known octahedral, diamagnetic, thermally robust Cr(CO)₆. As a limiting-case check: setting all ligand charges to zero collapses the ionic formula onto the covalent formula term-by-term, since $g_M-\mathrm{OS}=g_M-q$ when $\sum c_i=0$ — exactly what happens here, and exactly what the general derivation predicts for an all-neutral-ligand complex.

Example 2 (applied): [Mn(CO)₅]⁻, the manganese carbonyl anion used in stoichiometric organometallic synthesis. Manganese is group 7, $g_{Mn}=7$ . The overall ionic charge is $q=-1$ (this whole species is an anion). Covalent method: five CO ligands, each 2e ( $5\times2=10$ e). The complex charge correction uses $q=-1$ :

N^{\text{cov}} = g_{Mn} - q + \sum d_i^{\text{cov}} = 7 - (-1) + 10 = 7+1+10 = 18

Ionic method: the five CO ligands are neutral, so $\sum_i c_i = 0$ from the ligands; the entire $-1$ charge of the complex must then reside formally on the metal as the extra ligand-like anionic charge, i.e. the "5th CO plus the odd electron" picture is folded into $\mathrm{OS}=q-\sum c_i = -1-0=-1$ . This is Mn(−I), an unusually low, formally negative oxidation state — physically sensible because [Mn(CO)₅]⁻ really is isoelectronic with neutral Fe(CO)₅ and behaves as a potent nucleophile/reducing agent in organometallic substitution chemistry.

N^{\text{ion}} = (g_{Mn}-\mathrm{OS}) + \sum d_i^{\text{ion}} = (7-(-1)) + 5(2) = 8+10 = 18

Both give 18 (an 18-electron, closed-shell anion, exactly why it is isolable as, e.g., the Na⁺ or PPN⁺ salt and does not dimerize the way neutral 17-electron Mn(CO)₅• radical does). This example is deliberately asymmetric in overall charge (nonzero $q=-1$ ) to show that the charge correction term is essential and must be tracked with its correct sign in both formulas — a common source of student error is forgetting the $-q$ (equivalently $+|q|$ for an anion) term in the covalent formula.

Example 3 (exam/competition level): Vaska's complex, trans-[IrCl(CO)(PPh₃)₂], and the effect of oxidative addition of H₂. Iridium is group 9, $g_{Ir}=9$ . Vaska's complex is square planar, neutral overall ( $q=0$ ), with ligands Cl (X-type), CO (L-type), and two PPh�3 (L-type). Before oxidative addition. Covalent method: Cl• contributes 1e, CO contributes 2e, each PPh₃ contributes 2e ( $2\times2=4$ e).

N^{\text{cov}} = 9 - 0 + [1 + 2 + 4] = 9+7 = 16

Ionic method: Cl⁻ contributes charge $c=-1$ and donor number 2e; CO and PPh₃ are neutral. $\sum c_i = -1$ , so $\mathrm{OS}=0-(-1)=+1$ , i.e. Ir(I), $d^{8}$ ( $g_{Ir}-\mathrm{OS}=9-1=8$ ).

N^{\text{ion}} = (9-1) + [2+2+4] = 8+8 = 16

Both agree: Vaska's complex is a 16-electron, square-planar, Ir(I) d⁸ complex — the textbook signature of a coordinatively unsaturated d⁸ center primed for oxidative addition, exactly analogous to Rh(I) and Pt(II) d⁸ square-planar catalytic intermediates. After oxidative addition of H₂. The product is octahedral $[IrCl(H)_2(CO)(PPh_3)_2]$ : two new hydride ligands appear, and the metal geometry becomes six-coordinate. Covalently, H2 splits homolytically into two •H fragments, each donating 1e (total 2e added), while the metal's own electron count and formal charge in the neutral-radical convention do not change identity:

N^{\text{cov}} = 9 - 0 + [1(\mathrm{Cl}) + 2(\mathrm{CO}) + 4(2\times \mathrm{PPh_3}) + 2(2\times\mathrm{H})] = 9 + [1+2+4+2] = 9+9=18

Ionically, H2 undergoes formal heterolysis in the bookkeeping into two H⁻ ligands (each donor number 2e, each ionic charge $c=-1$ ), so the ligand charge sum becomes $\sum c_i = (-1)_{\text{Cl}} + (-1) + (-1) = -3$ , giving $\mathrm{OS} = 0-(-3) = +3$ , i.e. Ir(III), d⁶ ( $g_{Ir}-\mathrm{OS} = 9-3=6$ ).

N^{\text{ion}} = (9-3) + [2(\mathrm{Cl}) + 2(\mathrm{CO}) + 4(2\times\mathrm{PPh_3}) + 4(2\times\mathrm{H})] = 6+[2+2+4+4] = 6+12=18

Both give 18, confirming the product is an 18-electron, octahedral, Ir(III) d⁶ complex — and simultaneously showing the mechanistic content of oxidative addition in the ionic language: the metal's oxidation state rises by exactly 2 (Ir(I)→Ir(III)) while the electron count rises by exactly 2 (16→18), because H2, formally neutral, is reinterpreted as 2 H⁻ plus a +2 charge increment on the metal. This is precisely why oxidative addition is named as it is: oxidation state goes up while coordination number and electron count also go up, a combination only visible cleanly in the ionic bookkeeping.

Limiting cases, scope & common traps

Situation	What happens	Correct handling
All-neutral-ligand complex ( $\sum c_i = 0$ ), e.g. Cr(CO)₆, Fe(CO)₅	Ionic and covalent formulas collapse to the same term-by-term expression	OS equals the overall complex charge $q$ (often 0); no ambiguity
Charged complex, e.g. [Mn(CO)₅]⁻, [Co(CO)₄]⁻	The overall charge $q$ must appear explicitly in the covalent formula as $-q$	Track sign of $q$ carefully; anionic complexes raise $N^{\text{cov}}$
η⁵-Cp and other polyhapto π-ligands	Donor number depends on hapticity, not just formula	η⁵-Cp: 5e (cov) / 6e (ion); η¹-Cp (σ-bonded, X-type): 1e (cov) / 2e (ion) — always specify hapticity
Bent vs. linear NO	NO is a "non-innocent," ambiguous-charge ligand	Linear NO ≈ NO⁺ (2e, L-type analogy) or 3e neutral covalent donor; bent NO ≈ NO⁻ (2e ionic, X-type-like) or 1e covalent donor — must be assigned per structure, not by formula alone
Metal–metal bonded dimers, e.g. Mn₂(CO)₁₀	Each metal gets 1e credit (covalent) or the M–M bond is "ionic-silent" (contributes 0 net charge, still 1e each in most treatments) for the bond to the other metal	Add 1 electron per metal–metal single bond in either convention; do not double-count
Bridging ligands (µ-CO, µ-X)	Donor number splits between the two metals it bridges	A µ-CO typically donates a net 2e shared between two metals (commonly counted as 1e to each, or the full 2e assigned by an agreed convention) — always state the convention used

Common trapCommon misconception 1: "Oxidation state is a real, physically measured property of the metal."

Oxidation state, like electron count itself, is a formalism, not an observable. X-ray crystallography measures bond lengths and angles; spectroscopy measures energy gaps and magnetic moments; nothing measures "the charge on iron" directly. Different formalisms (ionic vs. covalent, or different electronegativity conventions for near-covalent M–C bonds) can assign different formal oxidation states to the same real electron density, which is why, for example, some highly covalent early-transition-metal alkylidene or metallocene systems have "textbook" oxidation states that are debated. The 18-electron total, by contrast, is far more robust because it is forced to be convention-independent by the algebra shown above.

The chemist behind it

The systematic ionic/covalent dual bookkeeping for organometallic electron counting is usually credited to Malcolm L. H. Green, who in the 1970s–1990s formalized the "L, X, Z" ligand classification (neutral 2-electron donors L, one-electron/anionic-radical donors X, and empty-orbital acceptor ligands Z) that underlies the modern presentation of both counting schemes in a single unified notation — the Covalent Bond Classification (CBC) or "MLX" method. Green's scheme, together with the parallel long-standing "ionic" tradition traceable to classical Werner coordination chemistry (in which ligands are always closed-shell ions and the metal always carries a formal oxidation number), gave organometallic chemists a rigorous, mutually-checkable pair of languages rather than a single ad hoc rule.

Origins · 1 minMalcolm L. H. Green and how this was discovered

Malcolm L. H. Green

1936–2020 · University of Oxford, organometallic and inorganic chemistry

Green spent his career at Oxford's Inorganic Chemistry Laboratory, working extensively on early transition-metal and lanthanide organometallic complexes, including pioneering work on niobium, tantalum, and titanium metallocene chemistry, and on catalytically relevant metal-hydride and metal-alkyl species. His 1995 paper formalizing the MLX (Covalent Bond Classification) system gave the field a compact, unambiguous notation — writing, e.g., ferrocene as MX₂L₂ with the Cp ligands each decomposed as one X-donor plus two L-donors reflecting their η⁵-hapticity — that maps directly onto both the covalent and ionic electron-counting totals developed here, and is now standard in advanced organometallic textbooks and in computational/database cataloguing of ligand types.

History

- 1827 — Zeise's salt, K[PtCl₃(C₂H₄)], the first alkene complex, synthesized without any electron-counting theory to explain it.

- 1951 — Ferrocene, Fe(C₅H₅)₂, discovered independently by Kealy & Pauson and by Miller, Tebboth & Tremaine; its sandwich structure and 18-electron stability catalyze the birth of modern organometallic chemistry.

- 1956–1960s — Sidgwick's "effective atomic number" (EAN) rule (extending 1927 work) becomes formalized into the modern 18-electron rule as organometallic carbonyl and metallocene chemistry expands rapidly.

- 1970s–1995 — M. L. H. Green develops and publishes the covalent bond classification (L, X, Z / MLX) notation, unifying covalent and ionic electron counting under one ligand-typing scheme.

- 1980s–present — Ionic vs. covalent electron counting becomes a standard dual-method presentation in organometallic textbooks (Crabtree, Hartwig, Elschenbroich, Miessler & Tarr), used to teach oxidation-state assignment alongside the 18-electron rule.

Why do we need two different electron-counting methods at all — wouldn't one suffice?

Each method is optimized for a different job. The covalent method is fast and mechanical: you never have to decide a ligand's "charge," so it is ideal for a quick 18-electron check on an unfamiliar complex. The ionic method is slightly more work but simultaneously yields the formal oxidation state and dⁿ configuration, which are essential for reasoning about crystal-field splitting, magnetism, redox reactivity, and mechanistic steps like oxidative addition. Keeping both available, and cross-checking that they agree, catches arithmetic slips and clarifies which quantities (electron count vs. oxidation state) are convention-dependent.

Is oxidation state from the ionic method the same as the "real" charge on the metal measured experimentally?

No. Oxidation state is a formal bookkeeping charge, assigned by the convention that every M–L bond is heterolyzed toward the more electronegative atom. Real metal centers in organometallic complexes carry much smaller partial charges than their formal oxidation states suggest, because M–C and M–H bonds are substantially covalent. Techniques like X-ray photoelectron spectroscopy, Mössbauer spectroscopy, and computed atomic charges (Bader, NPA) give experimentally or computationally grounded charge estimates that often disagree numerically with the formal oxidation state, though the two are usually correlated in sign and trend.

How do you decide whether NO is a 1-, 2-, or 3-electron donor?

Nitrosyl is the classic "non-innocent" ligand. A linear M–N–O geometry (bond angle near 180°) corresponds to NO acting like NO⁺ (isoelectronic with CO), donating 2 electrons ionically or being treated as a 3-electron covalent donor by some conventions; a bent M–N–O geometry (angle near 120°) corresponds to NO acting like NO⁻, donating 2 electrons ionically as an X-type ligand or 1 electron covalently as a radical-like donor. You must look at the experimental M–N–O bond angle (from crystallography) or use the total electron count itself as a consistency check to decide which assignment applies — the formula "NO" alone does not determine the donor number.

Does the covalent method ever give a different total electron count from the ionic method for the same real molecule?

No — if applied correctly and completely, the two totals are algebraically forced to agree, as shown by the general derivation. If a student computes different totals, it means the methods have been mixed (e.g., ionic ligand donor numbers used with the neutral metal electron count) or a ligand's ionic charge/hapticity was misassigned (e.g., forgetting that η¹-Cp is a 1-electron X-type ligand covalently, not 5-electron). A mismatch is a diagnostic that some step in the bookkeeping was inconsistent, not evidence that the methods are fundamentally different.

How does hapticity (η-number) affect the ligand's donor count?

Hapticity tells you how many contiguous atoms of a π-ligand are engaged in bonding to the metal, and the donor number scales with it. An η¹-allyl (σ-bonded through one carbon) is a 1-electron covalent / 2-electron ionic X-type ligand, just like an alkyl; an η³-allyl (bonded through all three carbons in an allylic π-system) is a 3-electron covalent / 4-electron ionic donor, combining X-type and L-type character (formally X + L). Cp similarly ranges from η¹ (1e/2e, X-type) to η⁵ (5e/6e, X + L₂ in the covalent-bond-classification notation). Always identify the hapticity from the structure (bond lengths, NMR fluxionality data) before assigning a donor number.

Why does ferrocene's electron count come out the same whether iron is treated as Fe(0) or Fe(II)?

Because the two treatments differ only in where four electrons (two per Cp ligand) are formally assigned — to the metal (as part of its neutral d⁸ configuration reduced by donation) or to the ligand (as part of the closed-shell Cp⁻ anion) — and the total pool of electrons being redistributed is fixed by the real molecule. In the covalent view, Fe(0) has 8 valence electrons and two neutral Cp radicals each donate 5, totalling 8+10=18. In the ionic view, Fe(II) has 6 d-electrons and two Cp⁻ anions each donate 6, totalling 6+12=18. The four "extra" ionic electrons (2 per Cp) are exactly the electrons removed from the metal's count when its oxidation state rises from 0 to +2, so the grand total cannot change.

Where does this topic lead next in organometallic chemistry?

This dual-counting framework is the direct foundation for classifying ligands formally as L, X, or Z-type donors (the covalent bond classification), for systematically assigning metal oxidation state and dⁿ configuration in any new complex, and for predicting coordination number and geometry preferences via the 18-electron rule together with steric (cone angle) considerations. It also underlies mechanistic electron bookkeeping through elementary organometallic steps such as oxidative addition, reductive elimination, and migratory insertion, each of which has a signature change in electron count and oxidation state.