Bond Order and Bond Stabilization

Bond order is not a historical curiosity confined to introductory diagrams — it remains the fastest diagnostic chemists reach for whenever an electron is added or removed from a $\pi$ or $\sigma$ system. In atmospheric and combustion chemistry, the bond-order logic applied above to $\mathrm{O_2}$ and $\mathrm{NO}$ directly explains why superoxide ($\mathrm{O_2^-}$, bond order 1.5) is a reactive intermediate in oxidative stress biochemistry — its weaker, longer bond relative to $\mathrm{O_2}$ makes it markedly more reactive, a fact enzymes such as superoxide dismutase exploit and defend against. In battery and energy-storage materials research, the same reasoning governs lithium-oxygen and lithium-sulfur cell discharge chemistry, where cycling between $\mathrm{O_2}$, $\mathrm{O_2^-}$, and $\mathrm{O_2^{2-}}$ (peroxide, in $\mathrm{Li_2O_2}$ discharge products) tracks exactly the bond-order ladder worked out in Example 2. In organometallic and catalytic chemistry, bond order in carbon monoxide and related diatomic ligands explains back-bonding: when CO binds a transition metal, electron density flows from filled metal $d$ orbitals into CO's empty antibonding $\pi^*$ orbital, lowering the effective C–O bond order from its free value near 3 and directly showing up as a measurably lower C–O stretching frequency in infrared spectroscopy — the single most common structural diagnostic used in organometallic labs to gauge a metal center's electron density. In semiconductor and photovoltaic materials, isoelectronic bond-order arguments (extended to $\mathrm{Si_2}$, $\mathrm{Ge_2}$, and heteronuclear III–V analogues of $\mathrm{N_2}$-like species) inform predictions of bond strength and thermal stability in advanced electronic materials. Even nitric oxide's physiological role as a signaling molecule in vasodilation is partly rationalized through its intermediate bond order and single unpaired electron, which make it a reactive but selectively controllable radical messenger rather than an inert closed-shell species.

Ask the tutor: How does the bond-order ladder change once $s$–$p$ orbital mixing swaps the ordering of $\sigma_{2p}$ and $\pi_{2p}$ going from N₂ to O₂?

Start from the single fact established for $\mathrm{H_2^+}$: combining two atomic orbitals $\phi_A$ and $\phi_B$ produces one bonding molecular orbital $\psi_+ \propto \phi_A + \phi_B$, lower in energy than the separated atomic orbitals, and one antibonding molecular orbital $\psi_- \propto \phi_A - \phi_B$, higher in energy. The bonding combination builds up electron density in the region directly between the nuclei; the antibonding combination has a nodal plane there, pulling density away from the internuclear axis and out to the far sides of each atom. Physically, density between the nuclei screens their mutual repulsion and simultaneously is attracted to both nuclei at once, which is why occupying $\psi_+$ lowers the total electronic energy relative to the separated atoms, while occupying $\psi_-$ raises it. Every additional electron pair that goes into a molecule has to go somewhere in this ladder of orbital energies. If it goes into a bonding orbital, the molecule is stabilized further; if it goes into an antibonding orbital, some or all of the stabilization already gained is cancelled. Bond order is simply the bookkeeping device that keeps track of this running total. Because the antibonding orbital is always raised in energy by somewhat more than the bonding orbital is lowered (a consequence of the overlap integral in the secular equation, covered under MO Energy Levels and the Secular Equation), filling both the bonding and antibonding orbital derived from a given pair of atomic orbitals leaves a molecule net less stable than the separated atoms — this is exactly why $\mathrm{He_2}$ does not form a stable ground-state molecule, despite having a perfectly good bonding $\sigma_{1s}$ orbital available. The picture generalizes immediately to many-electron diatomics. Build the full molecular orbital diagram — $\sigma_{1s}, \sigma_{1s}^*, \sigma_{2s}, \sigma_{2s}^*, \pi_{2p}, \sigma_{2p}, \pi_{2p}^*, \sigma_{2p}^*$ in the usual ordering — fill it from the bottom up obeying the Pauli principle and Hund's rule, and then simply count. Every electron in an orbital labeled with a bonding character (no star) is a "+1" contribution; every electron in an antibonding orbital (starred) is a "−1" contribution. Divide the net total by two because a full single bond, in the Lewis sense, corresponds to two electrons of net bonding character. Orbitals that are essentially atomic in character and do not change energy on bond formation — core-like lone-pair orbitals in more complex molecules — are nonbonding and are excluded from the count altogether.

The correlation with bond length now falls out of pure electrostatics: more net bonding electron density between the nuclei means a stronger net attractive force pulling them together, so the equilibrium separation at which attraction balances nuclear-nuclear repulsion shrinks. More net bonding density also means a deeper potential well, so more energy is required to pull the atoms apart to infinity — the bond dissociation energy rises. Both trends are monotonic in bond order across a well-matched series of related molecules or ions (same elements, same principal quantum number orbitals), though the numerical slope of length or energy versus bond order is not universal — it depends on which orbitals are involved and how strongly they overlap.

The central definition, applicable to any diatomic (or, with care, polyatomic fragment) once the MO occupation is known:

where:

• $n_b$ = number of electrons occupying molecular orbitals with net bonding character,
• $n_a$ = number of electrons occupying molecular orbitals with net antibonding character,
• electrons in strictly nonbonding orbitals are omitted from both counts,
• the factor of 2 converts a net electron imbalance into a bond count, matching the Lewis convention that one bond = one shared electron pair.

Reading: this formula says the molecule's net "bondedness" is exactly what is left after antibonding electrons cancel out the stabilization contributed by bonding electrons — it is a signed sum, not a raw electron count, and it can come out to zero, a positive integer, or a half-integer. For a homonuclear diatomic built only from $ns$ and $np$ atomic orbitals, filling orbitals from lowest to highest according to the aufbau principle and Hund's rule, the same relation can be written directly in terms of total electron count $N$ once you know how many of those electrons land in $\sigma_g/\pi_u$ (bonding) versus $\sigma_u^*/\pi_g^*$ (antibonding) levels:

where the specific orbital ordering (whether $\sigma_{2p}$ falls below or above $\pi_{2p}$) depends on $s$–$p$ mixing, a detail developed fully when the second-period diatomics are treated as a series. A second relationship, empirical but extremely reliable within a matched series, links bond order to the two experimentally measurable quantities chemists actually care about:

where:

• $D_e$ is the equilibrium bond dissociation energy,
• $R_e$ is the equilibrium bond length,
• $p$ is an empirical exponent near 1–2 depending on the series (not a universal constant — the point of this relation is the trend direction, not a single fitted power law).

Reading: you should not expect a single numerical formula for $D_e(\text{bond order})$ or $R_e(\text{bond order})$ that works across all diatomics; what is robust is the ordering — within an isoelectronic-like series sharing the same core orbitals, higher bond order always means shorter, stronger bonds.

Example 1 — Foundational: $\mathrm{H_2}$, $\mathrm{H_2^+}$, and $\mathrm{H_2^-}$ compared. Take the three simplest possible diatomic hydrogen species and rank them by bond order, then check the ranking against real bond energies. For $\mathrm{H_2^+}$ (1 electron total): the single electron occupies $\sigma_{1s}$, so $n_b = 1,\ n_a = 0$.

For $\mathrm{H_2}$ (2 electrons): both occupy $\sigma_{1s}$ (opposite spins, Pauli-allowed), so $n_b = 2,\ n_a = 0$.

For $\mathrm{H_2^-}$ (3 electrons): two fill $\sigma_{1s}$, the third must go to $\sigma_{1s}^*$, so $n_b = 2,\ n_a = 1$.

The predicted ordering of bond strength is $\mathrm{H_2} (1.0) > \mathrm{H_2^+} (0.5) \approx \mathrm{H_2^-} (0.5)$, and experiment agrees closely: $D_e(\mathrm{H_2}) \approx 4.75\ \text{eV}$ versus $D_e(\mathrm{H_2^+}) \approx 2.65\ \text{eV}$, while $\mathrm{H_2^-}$ is only weakly bound (a fraction of an eV, with the extra electron mostly diffuse). Note the asymmetry deliberately built into this example: $\mathrm{H_2^+}$ and $\mathrm{H_2^-}$ share the same bond order of 0.5 but are not identical molecules — $\mathrm{H_2^-}$'s antibonding electron is more diffuse and less tightly bound than $\mathrm{H_2^+}$'s bonding electron is compact, so bond order gives the right ordinal ranking but is not claiming numerical equality of $D_e$.

Limiting check: setting the antibonding population to zero identically ($n_a=0$) reduces the formula to $\text{BO}=n_b/2$, so a fully-filled single bonding orbital with no antibonding population whatsoever gives exactly bond order 1 — the Lewis single bond is recovered exactly, as it must be.

Example 2 — Applied: ranking $\mathrm{O_2}$, $\mathrm{O_2^+}$, $\mathrm{O_2^-}$, and $\mathrm{O_2^{2-}}$ by bond length. This is the textbook demonstration that bond order tracks real, measured bond lengths, and it is worth doing the electron count explicitly rather than quoting the answer. Neutral $\mathrm{O_2}$ has 16 electrons total (8 + 8). Filling the standard second-period MO diagram (with $\sigma_{2p}$ below $\pi_{2p}$, the correct ordering once $2p$ dominates over $s$–$p$ mixing, appropriate for $\mathrm{O_2}$ and $\mathrm{F_2}$) up through the $2p$ manifold:

The $1s$ core orbitals are nonbonding for chemical purposes (essentially unperturbed core orbitals that both stabilize and destabilize identically and are always excluded by convention), and $\sigma_{2s}/\sigma_{2s}^*$ cancel exactly (2 bonding, 2 antibonding). That leaves the valence $2p$-derived orbitals to set the bond order:

Removing one electron to form $\mathrm{O_2^+}$ removes it from the highest-occupied, antibonding $\pi_{2p}^*$ level ($n_a: 2\to 1$):

Adding one electron to form $\mathrm{O_2^-}$ (superoxide) adds it to $\pi_{2p}^*$ ($n_a: 2\to 3$):

Adding a second electron to form $\mathrm{O_2^{2-}}$ (peroxide) fully occupies $\pi_{2p}^*$ ($n_a: 2\to 4$):

The predicted bond-length ordering (shorter bond ⟺ higher bond order) is

and this matches the experimental sequence closely: $R_e(\mathrm{O_2^+}) \approx 112\ \text{pm}$, $R_e(\mathrm{O_2}) \approx 121\ \text{pm}$, $R_e(\mathrm{O_2^-}) \approx 128\ \text{pm}$, $R_e(\mathrm{O_2^{2-}}) \approx 149\ \text{pm}$ (peroxide, e.g. in $\mathrm{Na_2O_2}$). Bond order 2.0 for neutral $\mathrm{O_2}$ also reproduces the familiar Lewis-structure "double bond," while simultaneously explaining, via the two unpaired electrons forced into separate degenerate $\pi_{2p}^*$ orbitals by Hund's rule, why $\mathrm{O_2}$ is paramagnetic — a fact no simple Lewis structure can capture.

Limiting check: as the extra electrons are added one at a time from $\mathrm{O_2^+}$ to $\mathrm{O_2^{2-}}$, each addition goes into the same antibonding orbital $\pi_{2p}^*$, so bond order decreases by exactly $0.5$ per added electron and bond length increases monotonically — consistent with the formula's linear dependence on $n_a$ at fixed $n_b$.

Example 3 — Exam/competition level: bond order and bond length of $\mathrm{NO}$ and its ions, with an explicit non-round comparison to a diatomic force-constant estimate. Nitric oxide, $\mathrm{NO}$, is a heteronuclear radical with 15 electrons (7 + 8), making it an excellent test because it is not symmetric and forces genuine electron-counting rather than pattern memorization. Filling the MO diagram in the same order as before (appropriate once $2p$-dominated ordering applies, as it does past $\mathrm{N_2}$):

Fifteen electrons: $2+2+2+2+2+4+1 = 15$, correctly accounted for. As before the $1s$ core and the $\sigma_{2s}/\sigma_{2s}^*$ pair cancel, leaving

This half-integer bond order is a genuinely useful prediction: it says $\mathrm{NO}$'s bond should sit strictly between a double and a triple bond in both length and strength, and indeed $R_e(\mathrm{NO}) \approx 115\ \text{pm}$, intermediate between $R_e(\mathrm{O_2}) \approx 121\ \text{pm}$ (BO 2) and $R_e(\mathrm{N_2}) \approx 110\ \text{pm}$ (BO 3), and $D_0(\mathrm{NO}) \approx 6.5\ \text{eV}$, likewise intermediate between $D_0(\mathrm{O_2}) \approx 5.2\ \text{eV}$ and $D_0(\mathrm{N_2}) \approx 9.8\ \text{eV}$. Now push further: removing the single antibonding $\pi_{2p}^*$ electron gives the nitrosyl cation $\mathrm{NO^+}$ (14 electrons, isoelectronic with $\mathrm{N_2}$ and $\mathrm{CO}$):

Experimentally $R_e(\mathrm{NO^+}) \approx 106\ \text{pm}$, shorter than neutral $\mathrm{NO}$'s $115\ \text{pm}$, exactly as the bond-order jump from 2.5 to 3.0 predicts, and the stretching force constant rises correspondingly from $k(\mathrm{NO}) \approx 1595\ \text{N/m}$ to $k(\mathrm{NO^+}) \approx 2445\ \text{N/m}$ — a roughly 53% stiffer bond accompanying a bond-order increase of only 20% (2.5 → 3.0), a reminder (see the caveats below) that force constants and dissociation energies do not scale linearly with bond order even though they scale monotonically with it.

Limiting check: for the isoelectronic pair $\mathrm{NO^+}$ and $\mathrm{N_2}$ (both 14 electrons, both bond order 3.0 by the identical electron count), the bond order formula predicts comparable bond lengths — and indeed $R_e(\mathrm{N_2}) = 110\ \text{pm}$ is close to $R_e(\mathrm{NO^+}) = 106\ \text{pm}$, with the small residual difference attributable to the unequal nuclear charges (heteronuclear polarization) rather than to any difference in bond order.

Bond order as used here is the natural vocabulary of molecular orbital theory, and its clearest early champion was Robert S. Mulliken, who spent much of the 1920s–1930s systematically working out how atomic orbitals combine into molecular orbitals for real diatomic molecules and translating spectroscopic term symbols into orbital occupation diagrams. Where Linus Pauling was building valence-bond theory and resonance structures from localized bonds, Mulliken pushed the rival — and ultimately complementary — molecular orbital picture, insisting that electrons in a molecule occupy orbitals delocalized over the whole nuclear framework, with bonding and antibonding character determined by how the wavefunction behaves in the internuclear region. His painstaking correlation of spectroscopic data (vibrational frequencies, bond lengths, dissociation energies measured from electronic band spectra) with predicted orbital occupations across the first- and second-row diatomics is exactly the empirical foundation underlying the bond-order/bond-length/bond-energy correlation used throughout this article. He introduced much of the notation still used today — $\sigma$, $\pi$, $g$, $u$ labels carried over from atomic term symbols — and his 1966 Nobel Prize in Chemistry was awarded explicitly "for his fundamental work concerning chemical bonds and the electronic structure of molecules by the molecular orbital method."