Oxides, Peroxides and Superoxides of Group 1

Potassium superoxide's ability to simultaneously absorb CO₂ and release O₂ is a live-support workhorse reaction:

This single reaction converts exhaled CO₂ (the problem) directly into breathable O₂ (the solution) without needing any external oxygen supply, which is why KO₂ canisters and related "oxygen candle" and self-contained self-rescuer chemistry underlie submarine atmosphere regeneration, mine-rescue breathing apparatus, and early closed-circuit spacecraft life-support concepts. Sodium peroxide's clean hydrolysis to H₂O₂ (Example 2) made Na₂O₂ historically important as an industrial bleaching and oxidizing agent, and it remains a convenient solid-state source of "active oxygen" in some cleaning and water-treatment formulations, since it avoids transporting and stabilizing liquid hydrogen peroxide directly. The lattice-energy logic developed here — that cation size selects anion charge state by controlling how much electrostatic stabilization is available to "pay for" a more highly charged anion — generalizes far beyond Group 1. It explains why alkaline-earth peroxides (BaO₂ is far more stable/isolable than the analogous MgO₂, mirroring exactly the Li/Cs contrast but shifted up one group), why certain transition-metal and lanthanide oxide/peroxide chemistries show comparable radius sensitivity, and it is the same reasoning that will be generalized quantitatively under Lattice Energy and Thermodynamic Stability of s-Block Salts, where the same Born–Haber/Kapustinskii machinery is used to rationalize solubility trends, thermal-decomposition temperatures of carbonates and nitrates, and the broader "radius-ratio" logic that runs through essentially all of ionic solid-state chemistry. The superoxide radical anion O₂⁻ also has a substantial afterlife in biochemistry: it is the primary reactive oxygen species generated during mitochondrial electron transport leakage and by NADPH oxidase in immune cells, and its controlled detoxification by the metalloenzyme superoxide dismutase (which cycles a metal center between oxidation states to disproportionate 2 O₂⁻ + 2H⁺ → H₂O₂ + O₂, structurally the same disproportionation chemistry seen in Example 2) is essential to aerobic life — a striking case of the same O–O bond-order/MO logic appearing in both a submarine's air scrubber and a living cell's antioxidant defense.

Ask the tutor: How would this lattice-energy argument change if we used a Born–Haber cycle with real experimental lattice enthalpies instead of Kapustinskii estimates — would the Na/K crossover point shift?

Start from O₂ itself. The molecular orbital diagram of the oxygen molecule is the textbook example of why simple Lewis structures fail: the ground state is a triplet, with two unpaired electrons in the degenerate π orbitals (π₂ₚₓ and π₂ₚᵧ). The valence MOs, in increasing energy, are σ₂ₛ, σ₂ₛ, σ₂ₚᵤ, {π₂ₚₓ, π₂ₚᵧ}, {π₂ₚₓ, π₂ₚᵧ}, σ₂ₚᵤ. For neutral O₂ (12 valence electrons) the filling gives (σ₂ₛ)²(σ₂ₛ)²(σ₂ₚ)²(π₂ₚ)⁴(π₂ₚ)², with the last two electrons singly occupying the two degenerate π orbitals (Hund's rule) — this is the paramagnetism you can demonstrate by pouring liquid O₂ between the poles of a magnet. Bond order is computed the usual way,

Adding electrons to make the anions simply continues filling π: one electron gives superoxide, O₂⁻ (bond order 1.5), two electrons gives peroxide, O₂²⁻ (bond order 1, both π orbitals fully occupied, diamagnetic, no unpaired spins). Each additional electron in an antibonding π* orbital weakens and lengthens the O–O bond and lowers the vibrational stretching frequency — a trend you can watch directly in Raman/IR spectra of the isolated solids.

Now bring in the metal. Oxide, O²⁻, is not simply "O₂²⁻ with another electron and no O–O bond" — it is atomic oxygen fully reduced, with no O–O linkage at all; it is the anion of the binary oxide M₂O. Whether a solid alkali-metal oxide contains isolated O²⁻ ions, O₂²⁻ dimers, or O₂⁻ radicals is not decided by which anion is intrinsically "most stable" in the gas phase (in fact O²⁻ is unstable as a free gas-phase ion — it is only the surrounding lattice of cations that stabilizes it enough to exist at all). It is decided by which anion, paired with the cation actually present, gives the lowest-energy ionic solid.

The lattice-energy competition can be pictured with a simple electrostatic scaling argument before ever writing a full Born–Haber cycle. Coulombic lattice energy scales, to a first approximation via the Born–Landé/Kapustinskii-type relation, as

For a fixed cation radius $r_+$, replacing O²⁻ ($z_-=2$, $r_-\approx 140$ pm) with O₂⁻ ($z_-=1$, $r_-\approx 158$ pm, treating the whole ion as a single quasi-spherical unit) roughly halves the charge product and increases the denominator — both effects lower $U_L$. This loss of lattice energy is a bigger relative penalty when $r_+$ is small (because $r_+ + r_-$ is dominated by $r_-$'s change) and a smaller relative penalty when $r_+$ is already large (because $r_+$ dominates the sum and the anion swap barely moves it). That asymmetry — small cations "feel" the anion-charge change strongly, large cations barely feel it — is the entire content of the Li → Cs trend, and it is worked through quantitatively below.

The governing relationship is the lattice-energy dependence on ionic charge and radius, captured compactly by the Kapustinskii equation (a generalized, structure-independent version of the Born–Landé/Born–Mayer treatments):

where:

• $\nu$ = number of ions per formula unit,
• $z_+, z_-$ = magnitudes of cation and anion charges,
• $r_+ + r_-$ = sum of ionic radii in picometres,
• the $34.5$ pm term is an empirical compressibility correction (Kapustinskii's constant).

Reading: lattice energy grows linearly with the product of the charges but only inversely with the sum of the radii. Because the charge product term is multiplicative while the radius term is additive, doubling an anion's charge (2 vs 1) is a much bigger lever on $U_L$ when the radii are small (so the fractional change in $r_++r_-$ from switching anions is large) than when the cation is already big. The second governing relationship is simply an energy-cycle statement of competitive stability: for two candidate solids $\text{M}_2\text{O}$ (containing O²⁻) and $\text{M}_2\text{O}_2$ (containing O₂²⁻) formed from the same metal and O₂, the more stable product is the one with the more negative overall formation enthalpy,

where:

• $\Delta H_{\text{sub}}$ = sublimation enthalpy of the metal (atomization),
• $IE_1$ = first ionization energy of M,
• the O-based term bundles the gas-phase energy cost of making O²⁻, O₂⁻ or O₂²⁻ from O₂ (dissociation plus one or two electron affinities, with the second electron affinity of oxygen being strongly endothermic),
• $U_L$ = lattice energy of the specific M/anion pair, released on solid formation (a stabilizing, negative contribution to $\Delta_f H$).

Reading: everything on the right-hand side except $U_L$ is essentially fixed by the choice of anion (it doesn't care which alkali metal you use) — the electron affinity ladder for O₂ → O₂⁻ → O₂²⁻ is a fixed gas-phase cost. The only term that depends on which metal you paired the anion with is $U_L$. So the entire Li → Cs trend in product identity is a trend in how much lattice energy each cation can supply to offset the increasingly unfavorable gas-phase cost of building up more negative charge on oxygen.

Limiting-case check. As $r_+ \to \infty$ (an arbitrarily huge, hypothetical monocation), the Kapustinskii denominators $r_++r_-(\text{O}^{2-})$ and $r_++r_-(\text{O}_2^-)$ become nearly equal (both dominated by $r_+$), so $U_L(\text{oxide})$ and $U_L(\text{superoxide})$ differ almost only by the $z_+z_-$ charge-product factor of 2 vs 1 — the lattice energy of the oxide never quite catches down to that of the superoxide relative to what the fixed +852 kJ mol⁻¹ gas-phase penalty demands, so extremely large cations should always favor the superoxide (or even the neutral O₂ dioxygenyl adduct route) over oxide. This matches the empirical fact that only the very largest alkali metal, Cs, gives essentially pure superoxide on combustion, while Rb sits close to the same regime and K is borderline (often isolated as a K₂O₂/KO₂ mixture depending on O₂ pressure and conditions).

Example 1 (foundational): identify the product and its magnetism for Na + O₂. Sodium metal burned in excess dry O₂ gives predominantly sodium peroxide, Na₂O₂, not Na₂O. Using the MO count: Na₂O₂ contains O₂²⁻, bond order 1 (both π* orbitals doubly occupied), so it is diamagnetic — confirmed experimentally by the absence of any measurable magnetic susceptibility beyond normal diamagnetic core contributions. The O–O bond length in Na₂O₂ is about 149 pm, matching the peroxide entry in the bond-order table. If instead Na₂O (containing isolated O²⁻, no O–O bond at all) were the product, it would show no O–O stretching band in the Raman spectrum — but Na₂O₂ shows a clear O–O stretch near 730–790 cm⁻¹, confirming the peroxide assignment structurally as well as thermodynamically. Sodium sits at the crossover point predicted by the lattice-energy argument: too large a cation to make Li-type oxide favorable, but still small/charge-dense enough (compared to K, Rb, Cs) to stabilize the doubly-charged peroxide rather than defaulting fully to superoxide.

Example 2 (applied): why does Na₂O₂ react with water to liberate H₂O₂, and what does KO₂ do instead? Peroxide hydrolysis is simply protonation of the O₂²⁻ ion without breaking the O–O bond:

Each oxygen atom in O₂²⁻ simply picks up an H⁺; the O–O single bond, being a genuine σ bond between two anionic oxygens, survives protonation intact, and the product is hydrogen peroxide. Contrast this with superoxide hydrolysis. Superoxide is a radical anion (bond order 1.5, one unpaired electron); protonation of O₂⁻ is more complex because the resulting neutral HO₂• radical is unstable and disproportionates:

Here only half the oxygen ends up as peroxide; the other half is released as O₂ gas, because superoxide disproportionates (2 O₂⁻ → O₂²⁻ + O₂) far more readily than peroxide does — a direct consequence of the odd-electron, radical character of O₂⁻ making a second O₂⁻ + O₂⁻ combination (forming one O₂²⁻ and releasing one O₂) both spin-allowed and thermodynamically easy. This O₂-releasing hydrolysis (and analogously, CO₂ absorption, discussed below) is the reaction exploited industrially.

Example 3 (exam/competition level): estimate the crossover cation radius at which oxide and superoxide become equally favorable, using the derivation's own numbers. From the Kapustinskii-based derivation, for a generic monovalent cation of radius $r_+$ (in pm), the lattice-energy gap between the "per 2 M" oxide and superoxide routes can be approximated by interpolating the two computed data points. At $r_+=76$ pm (Li), $\Delta U_L = 1054$ kJ mol⁻¹; at $r_+=167$ pm (Cs), $\Delta U_L = 763$ kJ mol⁻¹. Treating this dependence as approximately linear over this range (a reasonable local approximation since the Kapustinskii function varies smoothly),

The net thermodynamic preference for oxide over superoxide is $\Delta U_L(r_+) - 894$ kJ mol⁻¹ (the fixed gas-phase cost difference derived earlier). Setting this to zero to find the crossover radius:

This crude linear estimate places the crossover cation radius at roughly 126 pm — squarely between Na⁺ (102 pm, favoring peroxide, an intermediate case not captured by this two-anion-only model) and K⁺ (138 pm, favoring superoxide). Given how coarse the model is (it ignores peroxide entirely and uses only two calibration points), landing within about 12 pm of the K⁺ radius, and correctly bracketing the experimentally observed Na/K changeover, is a reasonable success for a back-of-envelope electrostatic argument — precisely the kind of order-of-magnitude consistency check expected of a Kapustinskii-level estimate rather than a claim of quantitative precision.

The systematic recognition that alkali-metal combustion products follow a size-dependent trend across oxide/peroxide/superoxide emerged from mid-20th-century solid-state and structural inorganic chemistry, particularly X-ray and magnetic-susceptibility studies that definitively distinguished O₂²⁻ from O₂⁻ species (settling long arguments about whether "potassium peroxide" was really a distinct paramagnetic superoxide). The lattice-energy/radius-ratio framework itself descends directly from the ionic-model machinery built by Born, Landé, Haber, and later Kapustinskii for computing lattice energies from purely electrostatic and structural data — the same machinery used throughout main-group solid-state chemistry.