Noble Gas Compounds

Xenon fluorides remain the workhorse fluorinating and oxidizing agents of choice whenever a mild, selective, byproduct-clean oxidant is needed in synthetic inorganic and organofluorine chemistry: XeF₂ is used industrially and in research labs to fluorinate aromatic rings, deoxygenate certain substrates, and etch silicon in microelectronics fabrication — XeF₂ vapor-phase etching is a standard MEMS (microelectromechanical systems) technique precisely because it etches silicon isotropically without needing a plasma, leaving only volatile SiF₄ and Xe gas as byproducts. XeF₂'s clean, gaseous-byproduct chemistry is prized in cleanroom environments where liquid or ionic etchants would contaminate delicate structures. XeF₆ and XeOF₄ are used as potent, controllable oxidizing and fluorinating reagents in the synthesis of other high-oxidation-state or unusual-bonding compounds, including some of the only known Xe–C and Xe–N bonded species (organoxenon compounds such as $\text{C}_6\text{F}_5\text{–Xe}^+$ salts), which are of active interest as extremely powerful, selective fluorinating and oxidizing agents in modern synthetic chemistry, sometimes used to install fluorine at otherwise inaccessible positions in drug-candidate molecules. Beyond synthesis, xenon compounds and the underlying bonding principles are pedagogically central to modern inorganic chemistry because they are the sharpest available test case for hypervalent bonding models — every computational chemistry course that teaches natural bond orbital (NBO) analysis or 3-center-4-electron bonding uses XeF₂ or $\text{I}_3^-$ as the canonical example, directly linking back to the same electron-counting and orbital-symmetry tools used throughout main-group and Interhalogen Compounds chemistry, and to the periodicity arguments underlying The Inert-Pair Effect. Radon chemistry, though radiochemically limited to trace-scale study, is directly relevant to nuclear waste and radon-mitigation science, where understanding Rn's predicted (and partly observed) fluoride chemistry informs strategies for immobilizing radioactive radon in solid matrices.

Ask the tutor: Why does XeF₆ hydrolyze in stages to give XeOF₄ before XeO₃, rather than jumping straight to the fully oxidized oxide?

Xenon's electron configuration, $[Kr]4d^{10}5s^2 5p^6$, is a closed shell in the same sense that neon's or argon's is — full $s$ and $p$ subshells give a spherically symmetric, low-polarizability charge cloud in principle. But "closed shell" does not mean "high ionization energy" once you go far enough down a group. The outer $5p$ electrons of xenon are shielded by a full $n=4$ shell including ten $d$ electrons, and they sit in a large, diffuse orbital far from the nucleus. The first ionization energy of xenon, 1170 kJ mol⁻¹, is lower than that of hydrogen (1312), oxygen (1314), or even nitrogen (1402) — it is not a coincidence that Bartlett's key insight was recognizing that $PtF_6$, an oxidizer strong enough to ionize $O_2$ (forming $O_2^+[PtF_6]^-$), should also be strong enough to ionize xenon. Once xenon is partially oxidized — even formally to Xe(II) — the resulting Xe–F or Xe–O bonds are not weak, exotic curiosities. They are well-described, isolable, thermodynamically stable covalent bonds with bond enthalpies around 130 kJ mol⁻¹ (Xe–F in XeF₂), comparable to many "normal" halogen bonds. The shapes of the resulting molecules follow standard VSEPR rules once you correctly count xenon's lone pairs: Xe(II) in XeF₂ retains three lone pairs, Xe(IV) in XeF�4 retains two, and Xe(VI) in XeF₆ retains one (stereochemically active, causing the distorted structure). The electronic structure underlying these hypervalent species is best understood not through $sp^3d$ or $sp^3d^2$ hybridization (which would require significant d-orbital mixing energetically implausible for main-group atoms) but through three-center-four-electron (3c–4e) bonding, the same model used to describe the linear F–Xe–F unit and the axial F–I–F fragments in interhalogens. A linear F–Xe–F fragment uses one xenon $5p$ orbital and the two fluorine $2p_z$ orbitals to build three MOs: bonding, nonbonding, and antibonding. Four electrons (two from the Xe lone pair being oxidized/engaged, two from the two F atoms) fill the bonding and nonbonding MOs, leaving the antibonding orbital empty. The result is two Xe–F bonds each with bond order roughly 0.5, exactly consistent with observed Xe–F bond lengths intermediate between single and non-bonded distances, and with the absence of any need to invoke populated d orbitals on xenon.

The VSEPR structures themselves are worth visualizing directly, since they are the most exam-relevant, immediately checkable output of the whole bonding analysis.

The controlling thermodynamic quantity for noble-gas reactivity is the overall enthalpy of forming a bond, built from ionization energy, electron affinity/bond formation terms, and lattice or intermolecular stabilization. For the general oxidative fluorination

the reaction enthalpy can be decomposed (a Born–Haber-style cycle) as

where:

• $IE_1(\mathrm{Xe}) = 1170\ \text{kJ mol}^{-1}$ is the energy cost of removing the first electron from xenon (a formal bookkeeping term; the bonding is covalent, not ionic, but the ionization energy still tracks how "available" the xenon lone pair is),
• $D(\mathrm{F-F}) = 159\ \text{kJ mol}^{-1}$ is the F–F bond dissociation enthalpy (unusually weak due to lone-pair repulsion, a key enabling fact),
• $D(\mathrm{Xe-F}) \approx 130\ \text{kJ mol}^{-1}$ is the average Xe–F bond enthalpy,
• $n$ is the number of F₂ molecules consumed (1, 2, or 3 for XeF₂, XeF₄, XeF₆).

Reading this: fluorine's own bond is anomalously weak (comparable to a single Xe–F bond), so breaking one F–F bond and forming two Xe–F bonds is close to thermoneutral or exothermic even though xenon's ionization energy is substantial. This is exactly the same weak-F–F-bond argument used to rationalize why fluorine is such a potent oxidizer generally, a point that recurs in Interhalogen Compounds. The second governing relationship is the VSEPR electron-counting rule applied to a hypervalent central atom:

with the oxidation-state bookkeeping

more concretely, since xenon starts with 8 valence electrons and each Xe–F bond (or Xe=O, counted as using 2 electrons from Xe in a simple electron-counting sense) removes one electron from the "remaining lone pair pool" per single bond formed:

where a Xe=O double bond is counted as using 2 of Xe's electrons (one "slot"), matching how it is drawn as a single ligand position in VSEPR even though it's a formal double bond.

The third governing idea addresses why these hypervalent Xe–F bonds don't require exotic orbitals: the 3-center-4-electron (3c–4e) bonding scheme, already sketched in the MO figure above, replaces the older $sp^3d$/$sp^3d^2$ hybridization picture.

Example 1 (foundational) — Predict the shape of XeOF₄ from first principles. Xenon oxytetrafluoride, XeOF₄, has one Xe=O and four Xe–F bonds. Using the electron-counting rule, each Xe–F bond uses one "slot," and the Xe=O double bond is treated as occupying one ligand position (its extra π bond doesn't add a lone-pair-equivalent domain, matching how VSEPR treats multiple bonds as a single electron domain). Total bonding domains $=4\ (\text{F}) + 1\ (\text{O}) = 5$. Xenon's 8 valence electrons: 4 are used for the four Xe–F single bonds, and the Xe=O double bond consumes the remaining electron pairs needed for π donation from Xe (in the simplest Lewis picture, treat O as taking 2 of Xe's electrons for the σ framework, same as a single bond, since VSEPR domain-counting doesn't distinguish bond order for steric number purposes). So:

Since lone pairs must be an integer, the correct way to handle this (standard in Xe(VI) oxyfluoride counting) is to recognize Xe(VI) has one lone pair regardless of how many of its six "bonds" are to O vs F — the oxidation state fixes lone pair count at 1 (as in XeF₆), independent of which of the six positions are O vs F. Steric number $=5\ (\sigma\text{ domains}) + 1\ (\text{lone pair}) = 6$, an AX₅E₁ system. Six domains are octahedral; one lone pair distorts it, giving a square-pyramidal geometry with the lone pair opposite the apex — matching the experimentally observed structure: XeOF₄ is a square pyramid, O at the apex, four F atoms in the basal plane, F–Xe–F(basal) angles slightly less than 90° (bent toward the lone pair, ~91.8° observed, and O=Xe–F ~ 91.8°, C4v symmetry).

Example 2 (applied) — Estimate the enthalpy of forming XeF₄ from Xe and F₂, and compare to the experimental value. Using the decomposition above with $n=2$:

Plugging in $IE_1 = 1170$ kJ mol⁻¹, $D(\mathrm{F-F}) = 159$ kJ mol⁻¹, and an average $D(\mathrm{Xe-F}) \approx 130$ kJ mol⁻¹ (a reasonable estimate from bond-enthalpy tables, though this is a crude single-bond-enthalpy model that does not capture 3c–4e delocalization energy):

This naive Lewis-bond-enthalpy estimate is wildly endothermic and clearly wrong — it overweights the "ionization cost" as if the reaction proceeded through a genuinely ionized Xe²⁺(or Xe⁴⁺) species rather than delocalized covalent bonding. The experimental enthalpy of formation of XeF₄(g) is actually exothermic, $\Delta H_f^\circ \approx -218$ kJ mol⁻¹ from the elements, i.e.

The lesson (a genuine limiting-case check): the crude "full ionization energy paid up front" model fails because real Xe–F bonds are covalent 3c–4e bonds, not built by first fully ionizing xenon; the ionization energy is a useful qualitative predictor of which noble gases can react at all (a threshold test), but it must not be used as a literal additive term in a quantitative thermochemical cycle without a compensating, large "covalent bonding/orbital mixing" stabilization term that is not simply $4\times D(\text{Xe–F})$ from an average bond-enthalpy table. This is exactly why the reaction is run practically (see below) rather than estimated from naive electronegativity/ionization arguments alone.

Example 3 (exam/competition-level) — Rationalize the bent, not linear, Xe–O–Xe bridging angle in $[\text{XeOXe}]^{2+}$-type or $\text{Xe}_2\text{F}_3^+$-related oxo-bridged cations, and predict the F–Xe–F angle deviation in real XeF₆ using the pseudo-VSEPR/AIM picture. Real gas-phase XeF₆, unlike a hypothetical ideal $O_h$ AX₆ species, shows F–Xe–F angles compressed from 90° toward roughly 80–85° in several of its fluxional $C_{3v}$-like structures, because the seventh electron domain (the lone pair) is stereochemically active and pushes into the space between three fluorines on one triangular face, exactly as the AX₆E₁ VSEPR count in the derivation above predicts. Electron diffraction and matrix-isolation IR studies show XeF₆ interconverting rapidly (barrier a few kJ mol⁻¹) among several distorted structures — capped-octahedral ($C_{3v}$), pentagonal-pyramidal-like, and others — because the potential energy surface is extremely flat near the "ideal" $O_h$ geometry: the lone pair is not rigidly localized but sloshes across faces. Quantitatively, if we model the lone pair as occupying roughly the same steric volume as a fluorine substituent (a common approximation), the observed contraction of adjacent F–Xe–F angles by 5–10° below the ideal 90° matches the general VSEPR rule that lone pairs compress adjacent bond angles more than bonding pairs do, by an amount on the order of

a magnitude entirely consistent with lone-pair/bond-pair repulsion being larger than bond-pair/bond-pair repulsion (the same hierarchy used to explain why $\text{NH}_3$'s H–N–H angle, 107°, is less than tetrahedral 109.5°, and why $\text{ClF}_3$'s equatorial F–Cl–F is compressed below 120°). The limiting case check: if XeF₆ genuinely had no stereochemically active lone pair (as some early, later-disproven models suggested, treating Xe(VI) as using $d$ orbitals to make it a "true" $O_h$ AX₆ species), the F–Xe–F angles would all be exactly 90°/180° and the molecule would show a single sharp vibrational/NMR signature rather than the fluxional, temperature-dependent, multiply-split spectra actually observed — the data rule out the naive non-lone-pair-active picture decisively.