Walsh Diagrams and Molecular Shape

Walsh diagrams remain the fastest qualitative route to predicting whether a reactive intermediate is bent or linear, planar or pyramidal — a question that matters enormously in mechanistic organic and organometallic chemistry. Carbene chemistry leans directly on the CH₂ Walsh analysis: whether a carbene reacts as a singlet (concerted, stereospecific cyclopropanation via the filled $2a_1$-like lone pair acting as both nucleophile and, through the empty $1b_1$-like p orbital, electrophile) or as a triplet (stepwise, radical-like, non-stereospecific) is controlled by the same singlet–triplet gap that the Walsh diagram rationalizes, and synthetic chemists routinely tune substituents to stabilize one spin state over the other for exactly this reason (electronegative substituents like halogens stabilize the filled $2a_1$-derived lone pair and favor singlet carbenes, as seen in :CCl₂ and :CF₂). The same logic underlies why silylenes (:SiH₂ and its substituted analogs) are almost universally ground-state singlets, in contrast to :CH₂'s triplet ground state — silicon's larger 3s–3p energy gap shifts the relative stabilization of the Walsh orbitals. In main-group and organometallic chemistry, the AH₃ pyramidalization diagram directly explains the geometry and reactivity of amine versus borane fragments used as ligands: NH₃'s pyramidal, lone-pair-bearing nitrogen makes it an excellent σ-donor ligand for Lewis acids and transition metals, while BH₃'s planarity (and consequent empty p orbital) makes it the archetypal Lewis acid, forming adducts like H₃N–BH₃ whose stability is a textbook example of donor–acceptor bonding built from exactly the frontier orbitals the Walsh diagram identifies. Phosphine (PH₃) and its heavier congeners follow the same pyramidal pattern with even smaller bond angles, directly relevant to ligand design in homogeneous catalysis, where the "cone angle" and pyramidalization of phosphine ligands around a metal center governs steric and electronic properties central to catalyst tuning. In photochemistry and atmospheric chemistry, Walsh-diagram thinking predicts geometry changes upon electronic excitation or ionization — essential for interpreting UV/vis absorption band shapes (vibrational progressions reveal how much the equilibrium angle shifts between ground and excited state) in species from water photolysis products (OH, forming from H₂O photodissociation) to interstellar and cometary radicals (CH₂, NH₂, and related species detected by rotational and electronic spectroscopy in astrochemical environments). The same framework extends, with appropriate generalization to more atoms and more symmetry elements, to understanding pseudorotation and geometry preferences in larger main-group clusters, and its logical descendant — full ab initio potential energy surfaces correlated with orbital symmetry — underlies modern computational predictions of reactive intermediate geometries used throughout mechanistic organic chemistry and computational catalyst design. Understanding Walsh diagrams also deepens appreciation of the symmetry-adapted linear combination machinery central to Molecular Orbitals of Polyatomic Molecules, and connects naturally to the vibronic coupling ideas used in treating the Jahn-Teller Effect and to frontier-orbital arguments used throughout Woodward-Hoffmann Rules for pericyclic reactions.

Ask the tutor: How would the Walsh diagram change for an AH₂ molecule where A is a heavier atom with a much larger ns–np energy gap, like SiH₂ compared to CH₂?

Consider a generic AH₂ molecule, H–A–H, and parametrize its geometry by the H–A–H bond angle θ, running from 180° (linear, point group D∞h) down to something like 90–100° (bent, point group C2v). At θ = 180°, the central atom A contributes a 2s orbital (σg symmetry), a 2pz orbital along the molecular axis (σu symmetry, taking z as the H–A–H axis), and a degenerate pair of 2px, 2py orbitals perpendicular to that axis (πu symmetry). The two hydrogen 1s orbitals combine into a symmetric SALC (σg symmetry) and an antisymmetric SALC (σu symmetry); they have no component that can match πu, so the πu pair on A remains strictly nonbonding at linear geometry — it simply does not see the hydrogens at all when the molecule is perfectly linear. Four MOs matter for a first-row AH₂ hydride with a light central atom: 1σg (bonding combination of A 2s and the symmetric H SALC), 1σu (bonding combination of A 2pz and the antisymmetric H SALC), the nonbonding 1πu pair (pure A 2px, 2py, doubly degenerate), and 2σg (the antibonding partner of 1σg, mixing A 2s and H SALC out of phase). As θ decreases from 180°, the molecular symmetry drops to C2v, and every degeneracy tied to the linear axis is broken. In C2v (with the molecule kept in the yz-plane and z bisecting the H–A–H angle), the labels become a1, b2, b1, a1 for these same four orbitals in order, and the crucial event is what happens to the formerly degenerate πu pair: the in-plane component (the one that can point toward the hydrogens as the angle closes) becomes a1 and is allowed by symmetry to mix with the A 2s orbital and with the H2 SALC, while the out-of-plane component remains b1 and stays essentially a pure, nonbonding 2p orbital on A throughout the whole bending motion. That single symmetry-driven split is the entire content of the Walsh diagram for AH₂. The b1 orbital, having no in-plane hydrogen partner to mix with, is nearly flat as a function of θ. The 2a1 orbital (the bent-geometry descendant of 1πu's in-plane lobe) drops sharply in energy as θ decreases from 180° toward 90°, because bending allows it to rehybridize with A's 2s character and to develop bonding overlap with both hydrogens simultaneously — geometrically, an sp-like hybrid pointing into the H–A–H bite angle overlaps far better with both H 1s orbitals at 90–104° than a pure unhybridized p orbital does at 180°. Meanwhile 1a1 (built from 1σg) rises gently on bending because the symmetric H2 combination loses some bonding overlap with A's 2s orbital as the H–H distance shrinks and the H 1s lobes begin to repel each other, and 1b2 (from 1σu) also rises somewhat because bending reduces the effective overlap of the antisymmetric hydrogen combination with A's pz-derived orbital. The higher-lying 3a1 (the antibonding partner) falls slightly as θ decreases. All four curves together constitute the classic Walsh AH₂ diagram; its single decisive feature is the steep downward slope of 2a1 with decreasing angle. The analogous story for planar AH₃ (D3h, point group with a threefold axis) versus pyramidal AH₃ (C3v) runs in parallel. At D3h the central atom's valence set is 2s (a1′), the in-plane pair 2px, 2py (e′), and the out-of-plane 2pz (a2″); the three hydrogen 1s orbitals form SALCs of symmetry a1′ + e′. The a1′ hydrogen combination bonds with A's 2s to give 1a1′ (strongly bonding) and 2a1′ (antibonding); the e′ hydrogen combination bonds with A's in-plane 2p pair to give a bonding e′ pair and an antibonding e′* pair; and A's 2pz (a2″) is left with no planar partner, so it is nonbonding in the D3h reference and becomes the LUMO of a closed-shell planar AH₃ system (or the frontier orbital that a radical or anion must occupy). As the molecule pyramidalizes toward C3v, the a2″ orbital drops in energy — it mixes in A 2s character and gains bonding overlap with the hydrogen a1 SALC as the pyramid folds up — while the formerly nonbonding a2″ orbital and one component of the bonding manifold reorganize into the C3v labels 1a1, 1e, 2a1, 2e. The key downward-sloping curve on pyramidalization is again the orbital derived from a2″, now labeled 2a1 in C3v; occupying it drives the molecule toward a pyramidal shape exactly as occupying the bent-favoring 2a1 orbital drives AH₂ toward a bent shape.

The organizing quantity is the total electronic energy at the one-electron level, written as a sum over occupied MOs:

where: $\varepsilon_i(\theta)$ is the energy of MO $i$ as a function of bond angle $\theta$ (the Walsh-diagram curve for that orbital), $n_i$ is its occupation number (0, 1, or 2), and the sum runs over all occupied orbitals.

Reading: the equilibrium angle is the value of $\theta$ that minimizes $E_{\text{tot}}(\theta)$; since the core-like $\sigma$ orbitals change slowly and roughly cancel between different molecules with the same skeleton, the frontier (HOMO) orbital's slope usually decides the answer — this is Walsh's rule. The angle-dependence of each orbital energy is captured, to leading (second) order in the bending distortion, by vibronic perturbation theory. Let $Q$ be a bending coordinate (proportional to $180° - \theta$, i.e. $Q=0$ at the linear/planar reference geometry) and let $H^{(0)}$ be the electronic Hamiltonian at $Q=0$ with eigenstates $\varphi_i^{(0)}$, energies $\varepsilon_i^{(0)}$. Expanding the one-electron Hamiltonian in $Q$,

where: the first-order term is the linear vibronic coupling operator and the second-order term contains the direct curvature (geometric) contribution to the orbital energy. Standard non-degenerate perturbation theory applied orbital-by-orbital gives, to second order,

where: the linear term vanishes by symmetry whenever $\varphi_i^{(0)}$ belongs to a nondegenerate irreducible representation at the reference (high-symmetry) geometry — true for $\sigma_g,\sigma_u$ in D∞h and for $a_1', a_2''$ in D3h — the sum over $j$ is the vibronic (pseudo-Jahn–Teller-type) mixing term, and the last term is the direct force-constant-like contribution.

Reading: every Walsh curve is, to lowest order, a parabola in the bending coordinate; whether it curves up or down is set by the sign and size of the vibronic coupling matrix element to nearby orbitals of the correct symmetry — this is precisely why the πu-derived 1b1/2a1 split shows one flat and one steep member: only the in-plane member has a nonzero, large coupling matrix element to A's 2s orbital through the bending motion.

Example 1 — Foundational: locating BeH₂ and H₂O on the same Walsh diagram and rationalizing 180° vs. 104.5°. Valence electron count for BeH₂: Be (1s²2s²) contributes 2 valence electrons from 2s; each H contributes 1; total $= 2 + 1 + 1 = 4$. These 4 electrons fill $1a_1$ (2 electrons, the strongly bonding combination of Be 2s + H₂ symmetric SALC) and $1b_2$ (2 electrons, Be 2pz + H₂ antisymmetric SALC). Both are bonding, both are already used at 4 electrons, and the shape-determining orbital $2a_1$ is empty. Since $\varepsilon(1a_1)$ and $\varepsilon(1b_2)$ both rise slightly upon bending (their bonding overlaps degrade as the H⋯H distance shrinks and as the σg/σu bonding character is disrupted), the total energy $E_{\text{tot}}(\theta) = 2\varepsilon_{1a_1}(\theta)+2\varepsilon_{1b_2}(\theta)$ is minimized at $\theta = 180°$: BeH₂ is linear. This is confirmed spectroscopically — gas-phase BeH₂ (observed by matrix isolation and high-temperature spectroscopy) shows no permanent dipole moment and IR/Raman patterns consistent with D∞h symmetry, and modern coupled-cluster calculations place its equilibrium angle at 180° to within computational precision. Valence electron count for H₂O: O (2s²2p⁴) contributes 6 valence electrons; each H contributes 1; total $=6+1+1=8$. Filling from the bottom: $1a_1$ (2), $1b_2$ (2), $2a_1$ (2), $1b_1$ (2) — all four Walsh orbitals doubly occupied. The decisive orbital $2a_1$, which drops steeply with decreasing angle, is fully occupied, contributing strongly to bending. The competing rise of $1a_1$ and $1b_2$ on bending is smaller in magnitude than the fall of $2a_1$ (this asymmetry is what ab initio Walsh-diagram calculations on H₂O consistently show — the $2a_1$ curve has by far the largest curvature of the four), so the energy minimum sits well off 180°.

Minimizing this sum places the equilibrium angle at $\theta_{eq} \approx 104.5°$, the experimentally measured H–O–H angle in gaseous water (microwave spectroscopy value: 104.48°). The limiting case check: if the 2 electrons that would occupy $2a_1$ were instead removed (as in the hypothetical 6-valence-electron species, isoelectronic with BeH₂-like filling), Walsh's rule predicts a return toward linearity — and this is exactly the trend observed experimentally: BeH₂ (4 e⁻, linear) → CH₂ triplet (6 e⁻, bent ~134°, only partial $2a_1$ occupation) → H₂O (8 e⁻, bent 104.5°, full $2a_1$ occupation) shows the bond angle closing monotonically as $2a_1$ occupation increases from 0 to 2, exactly as the orbital-filling argument demands.

Example 2 — Applied: predicting and explaining the singlet–triplet gap's effect on CH₂'s geometry, and comparing to isoelectronic NH₂⁺. Methylene, CH₂, has 6 valence electrons (C: 4, two H: 1 each). Its ground state is a triplet, ³B₁, with electron configuration (in the bent, C2v labeling) $(1a_1)^2(1b_2)^2(2a_1)^1(1b_1)^1$, the two "extra" electrons beyond the BeH₂-like 4-electron core occupying $2a_1$ and $1b_1$ singly with parallel spins (Hund's rule maximizes exchange stabilization for two electrons in near-degenerate orbitals). Its lowest singlet state, ¹A₁, instead pairs both extra electrons into $2a_1$: configuration $(1a_1)^2(1b_2)^2(2a_1)^2$. Applying Walsh's rule to each configuration separately: in the triplet, only one electron occupies the steeply bending-favoring $2a_1$ orbital, while the other sits in the angle-insensitive $1b_1$; the net bending force is roughly half what it would be with $2a_1$ doubly occupied, so the triplet's equilibrium angle is pulled only modestly away from 180°. In the singlet, both extra electrons occupy $2a_1$, doubling the bending stabilization relative to a single-electron occupation, and the singlet bends much further. Experimental and high-level ab initio values bear this out precisely: triplet CH₂ (³B₁) has an equilibrium bond angle of about 133.8°, while singlet CH₂ (¹A₁) has an equilibrium bond angle of about 102.4° — a difference of roughly 31°, entirely rationalized by whether $2a_1$ holds one or two electrons. The singlet–triplet energy gap itself is small (about 9 kcal/mol, or ≈ 3140 cm⁻¹, with the triplet lower), which is why both states are experimentally accessible and both angles have been measured by electronic spectroscopy. Now compare to the isoelectronic cation NH₂⁺ (N contributes 5 valence electrons, loses 1 for the +1 charge, plus 2 H at 1 each: $5-1+1+1=6$, same valence count as CH₂). Because NH₂⁺ is isoelectronic with CH₂, Walsh's rule predicts the same qualitative behavior: a triplet ground state bent near 180° minus a modest deflection, and a low-lying singlet bent more sharply. Experimentally, ground-state NH₂⁺ is indeed a ³B₁ triplet with a bond angle near 145–150° (somewhat more open than CH₂'s triplet, consistent with nitrogen's higher electronegativity slightly altering the s/p mixing energetics but not the qualitative Walsh-diagram topology), and its excited singlet state is substantially more bent, around 115–120°. The Walsh framework correctly predicts the ordering and direction of the angle change between spin states across an entire isoelectronic series, even though it does not, by itself, pin down the precise numerical angle without a real calculation of the orbital-energy curves.

Example 3 — Exam/competition level: ranking the pyramidalization angle of BH₃, CH₃, NH₃, and predicting the effect of removing an electron from NH₃⁺. Valence electron counts: BH₃ has $3 + 1+1+1 = 6$; CH₃ has $4+1+1+1=7$; NH₃ has $5+1+1+1=8$; the radical cation NH₃⁺ has $8 - 1 = 7$. Using the AH₃ Walsh diagram (planar $1a_1' < 1e' < a_2'' < 2e'$, correlating to pyramidal $1a_1 < 1e < 2a_1 < 2e$): the first 6 electrons always fill $1a_1$ and $1e$ (the two A–H bonding sets, present in all four species). The decisive orbital is the one descending from $a_2''$ — call it $2a_1$ at pyramidal geometry. BH₃ has zero electrons in this orbital: no pyramidalization driving force, planar, D3h, confirmed experimentally (BH₃ itself is unstable as a monomer and dimerizes to B₂H₆, but isoelectronic BF₃ and the transient BH₃ fragment observed spectroscopically are both planar). CH₃ has one electron in this orbital ($2a_1)^1$: partial pyramidalization drive, and indeed the methyl radical is known to be very shallowly pyramidal with an extremely low inversion barrier (essentially planar on the vibrationally averaged structure, with an sp² framework only slightly puckered — early Walsh-diagram treatments correctly flagged CH₃ as intermediate). NH₃ has two electrons in this orbital, $(2a_1)^2$: full pyramidalization drive, and NH�3 is robustly pyramidal with ∠HNH = 106.7° and an inversion barrier of about 2076 cm⁻¹ (≈ 5.8 kcal/mol), large enough to give a well-defined pyramidal equilibrium structure while still allowing quantum-mechanical tunneling inversion (the basis of the ammonia maser). Now predict NH₃⁺: removing one electron from neutral NH�3's $(1a_1)^2(1e)^4(2a_1)^2$ configuration removes it from the highest-occupied, pyramidalization-favoring $2a_1$ orbital (ionization from a nonbonding-ish, largely N-lone-pair-derived orbital is energetically the easiest route, consistent with NH3's low first ionization energy of 10.07 eV corresponding almost entirely to lone-pair character). This leaves NH₃⁺ with configuration $(1a_1)^2(1e)^4(2a_1)^1$ — the same orbital occupation pattern as neutral CH₃. Walsh's rule therefore predicts NH₃⁺ should be substantially less pyramidal than neutral NH₃, closer to CH₃'s shallow pyramidalization or even nearly planar. This prediction is borne out: photoelectron spectroscopy of NH₃ shows a long vibrational progression in the ν₂ (umbrella) mode upon ionization to NH₃⁺, the classic signature of a large geometry change along that coordinate — direct spectroscopic proof that removing the $2a_1$ electron flattens the pyramid substantially, exactly as the correlation diagram demands.

Arthur Donald Walsh (1916–1977) was a British physical chemist whose formative work, published as a landmark series of papers in the Journal of the Chemical Society in 1953, laid out correlation diagrams for AH₂, AH₃, AB₂, and other small polyatomics as a function of bond angle. Walsh worked at the time within the tradition of British spectroscopic chemistry — closely tied to the electronic-spectroscopy programs that were mapping out excited states of small radicals and hydrides by UV absorption — and his diagrams were explicitly built to explain why the excited electronic states of molecules like BH₂, CH₂, and NH₂ showed dramatically different rotational constants (hence different equilibrium geometries) from their ground states. He derived the orbital correlations using qualitative LCAO and perturbation arguments long before large-scale ab initio calculations were feasible, and the predictive power of his diagrams — later fully vindicated by configuration-interaction and coupled-cluster calculations across the periodic table — made "Walsh diagram" and "Walsh's rules" permanent fixtures of inorganic and physical-organic pedagogy.