Electron Probability Density and Orbital Phase

Orbital phase is not a historical curiosity confined to introductory bonding diagrams — it is load-bearing machinery across modern chemistry and materials science. In organic photochemistry and pericyclic reaction theory, the Woodward-Hoffmann rules that predict whether a cyclization proceeds thermally in a conrotatory or disrotatory sense are pure orbital-phase arguments: the terminal lobes of the highest occupied molecular orbital must rotate so that same-phase lobes meet to form a new bonding overlap, and the symmetry of that HOMO (itself set by how many phase-alternating nodes the $\pi$ system has) dictates the stereochemical outcome. This is orbital phase reasoning applied directly to synthesis planning and stereochemical prediction. In materials chemistry, band structure calculations for semiconductors and insulators are large-scale extensions of exactly the two-center bonding/antibonding splitting derived above: as more and more atomic orbitals combine across a crystal lattice, bonding-type in-phase combinations spread into a low-lying valence band and antibonding-type out-of-phase combinations spread into a high-lying conduction band, with the band gap directly traceable to the same $\alpha, \beta, S$ integrals scaled up to a periodic array. Crystal orbital overlap population (COOP) and crystal orbital Hamilton population (COHP) analyses, standard tools in computational solid-state chemistry, are explicitly phase-resolved bonding diagnostics — they decompose the total density of states into bonding (same-phase, stabilizing) and antibonding (opposite-phase, destabilizing) contributions at each energy, exactly the distinction this article develops for a single diatomic pair. In catalysis and organometallic chemistry, the frontier molecular orbital picture of metal-ligand bonding — $\sigma$-donation from a filled ligand lone pair into an empty metal $d$ orbital, and $\pi$-backdonation from a filled metal $d$ orbital into an empty ligand $\pi^*$ orbital — is entirely a phase-matching argument: backdonation is only possible because the metal $d_{xz}$ orbital and the CO $\pi^*$ orbital have compatible lobe symmetry and can overlap in phase in the relevant geometry. Scanning tunneling microscopy with orbital resolution, and modern computational visualization of molecular orbitals in drug-design software, routinely display signed isosurfaces (red/blue lobes) for exactly this reason — chemists need to see phase, not just density, to reason about reactivity. Even the basic story of how deeply an orbital's electron penetrates toward the nucleus — the subject of Penetration and Shielding — depends on the radial nodal/phase structure of $s$, $p$, and $d$ orbitals in a way directly parallel to the angular phase structure explored here.

Ask the tutor: If I flip the arbitrary overall sign convention of one atomic orbital in a diatomic pair, does the bond itself become weaker, or does only the algebraic labeling of "bonding" and "antibonding" change?

An atomic orbital is a solution $\psi_{n\ell m}(r,\theta,\phi)$ of the one-electron Schrödinger equation, and it is a real (or complex) mathematical function that can be positive, negative, or zero at different points in space. The probability density for finding the electron in a small volume $d\tau$ around a point is $|\psi|^2 d\tau$, and because it is a modulus-squared quantity, it is never negative — a region where $\psi$ is strongly negative is just as "probable" as a region where $\psi$ is equally strongly positive. This is the first fact to internalize: probability density is phase-blind. But $\psi$ itself is not phase-blind, and neither is anything computed from sums of wavefunctions before squaring. This distinction matters because bonding is fundamentally a two-orbital (or many-orbital) superposition problem. When orbital $\chi_A$ on atom A and orbital $\chi_B$ on atom B combine to form a molecular orbital, the natural first approximation (LCAO, linear combination of atomic orbitals) is

and the two possible relative signs, $+$ and $-$, produce two completely different molecular orbitals with different energies, even though $\chi_A$ and $\chi_B$ individually are identical 1s orbitals with identical $|\chi|^2$ distributions. The probability density of the combination is

and the cross term $\pm 2c_Ac_B\chi_A\chi_B$ — the interference term — is exactly where phase enters. In the region between the nuclei, both $\chi_A$ and $\chi_B$ are positive (for two same-phase 1s orbitals), so $\chi_A\chi_B > 0$; the $+$ combination adds this positive cross term and piles up extra electron density between the nuclei, while the $-$ combination subtracts it and creates a node there. Squaring the individual orbitals would never reveal this — you must add the signed wavefunctions first, then square. Angular nodes are the source of lobe sign patterns on $p$, $d$, and $f$ orbitals. A $2p_z$ orbital has the angular factor $\cos\theta$, positive for $\theta < 90°$ (the $+z$ lobe) and negative for $\theta > 90°$ (the $-z$ lobe), with a nodal plane at $\theta = 90°$ (the $xy$-plane) where $\psi = 0$ exactly. Radial nodes, by contrast, are spheres where the radial function $R_{n\ell}(r)$ changes sign as $r$ increases — a $2s$ orbital, for instance, is positive near the nucleus and negative farther out, with a radial node at a specific radius. Both kinds of node are surfaces of $\psi = 0$ across which the sign flips; both are invisible in $|\psi|^2$ except as regions of exactly zero density.

The energetic consequence of this interference is not symmetric: the antibonding orbital is destabilized more than the bonding orbital is stabilized, because the overlap integral $S$ appears in the denominator of both energy expressions and works against the antibonding combination. This asymmetry, worked out below, is why filling both a bonding and its antibonding partner (as He$_2$ would require) gives a net repulsive interaction rather than a neutral one. Phase logic scales up directly to $p$-orbital bonding. Two $p_z$ orbitals pointing along the internuclear axis, each with a $+$ lobe facing the other, overlap head-on and in-phase to give a $\sigma$ bond; if instead one atom's $+$ lobe faces an adjacent atom's $-$ lobe, the head-on overlap is destructive and gives $\sigma^*$. Two parallel $p_x$ orbitals (perpendicular to the internuclear axis) can overlap sideways, lobe-over-lobe, to give $\pi$ bonding when same-signed lobes align above and below the axis, and $\pi^*$ when they do not. This is exactly the phase bookkeeping that builds the whole MO diagram of a diatomic molecule.

Notice in both figures that a single $p$ orbital already has one $+$ lobe and one $-$ lobe purely from its own angular node — this internal sign pattern is not a bonding/antibonding label by itself. Only when you compare the phase of one orbital's lobe to the phase of another orbital's lobe in the region where they overlap does bonding vs. antibonding become meaningful. This is the single most common point of confusion for students, and it is worth stating as its own principle before moving to the quantitative treatment.

The central quantitative object is the secular problem for a two-orbital basis $\{\chi_A, \chi_B\}$, from which both bonding/antibonding energies and the role of phase emerge. Define the overlap integral, Coulomb (site) integrals, and resonance integral:

where:

• $S$ is the overlap integral: it measures the spatial coincidence of $\chi_A$ and $\chi_B$, weighted by their relative sign; $S>0$ if the overlapping lobes have the same phase, $S<0$ if opposite, and $|S|$ grows with closer approach and better angular alignment.
• $\alpha$ is the Coulomb integral (roughly the orbital energy of an isolated atomic orbital in the molecular environment), negative for a bound electron and essentially phase-independent since it involves only one orbital at a time.
• $\beta$ is the resonance (exchange) integral, the energetic reward or penalty of having amplitude simultaneously on both centers; it is negative for same-phase (in-phase) overlap of orbitals that would form a bond, and its sign flips if you flip the relative phase of $\chi_B$.
• $H$ is the molecular electronic Hamiltonian operator, containing kinetic energy and all nuclear/electron Coulomb attraction and repulsion terms relevant to the diatomic framework.

Solving the $2\times2$ secular determinant for the two symmetric/antisymmetric combinations gives the bonding and antibonding orbital energies:

where:

• $E_+$ (using the top signs) is the bonding MO energy, built from $\psi_+ = \chi_A + \chi_B$ (in-phase combination);
• $E_-$ (bottom signs) is the antibonding MO energy, built from $\psi_- = \chi_A - \chi_B$ (out-of-phase combination);
• because $\beta < 0$ and $S > 0$ for a bonding-type same-phase overlap, $E_+ < \alpha < E_-$: the bonding MO is stabilized below the atomic level and the antibonding MO is destabilized above it.

Reading: the entire bonding/antibonding energy gap traces back to the sign of $\beta$ (which orbital phase controls directly) and the magnitude of $S$ (which phase-dependent overlap also controls). If you had used $|\chi_A|$ and $|\chi_B|$ — i.e., discarded phase and worked with probability amplitudes only — $\beta$ would be undefined as a signed quantity and this splitting would not exist at all. A second essential relationship makes explicit why the antibonding level rises more than the bonding level falls — the destabilization/stabilization asymmetry:

where:

• $\Delta E_+$ is the stabilization energy of the bonding MO relative to the atomic orbital energy $\alpha$;
• $\Delta E_-$ is the destabilization energy of the antibonding MO, and because $1-S < 1+S$ for $S>0$, the same magnitude of numerator produces a larger $|\Delta E_-|$ than $|\Delta E_+|$;
• the practical consequence is $|\Delta E_-| > |\Delta E_+|$, so a fully filled bonding-plus-antibonding pair configuration (four electrons, as in He$_2$) is net destabilizing, not neutral.

Example 1 (foundational) — reading lobe signs off a $2p_z$ orbital and predicting overlap type. A $2p_z$ orbital has angular part proportional to $\cos\theta$. State the sign in each hemisphere and predict whether two $2p_z$ orbitals on atoms 3.5 Å apart along $z$, both using the convention that the lobe pointing toward $+z$ is positive, give bonding or antibonding character when combined as $\chi_A + \chi_B$. For $\theta$ between $0°$ and $90°$ (upper hemisphere, $+z$ side), $\cos\theta>0$, so the upper lobe is positive; for $\theta$ between $90°$ and $180°$, $\cos\theta<0$, so the lower lobe is negative. Place atom A at $z=0$ with its positive lobe pointing toward $+z$ (toward atom B) and atom B at $z=3.5\,\text{Å}$ with its positive lobe pointing toward $-z$ (toward atom A) — this is the natural head-on orientation for end-on overlap. In the internuclear region, atom A contributes its positive lobe and atom B also contributes its positive lobe (since B's positive lobe faces inward), so $\chi_A>0$ and $\chi_B>0$ between the nuclei. The combination $\chi_A+\chi_B$ therefore reinforces constructively in that region:

If instead atom B were oriented with its positive lobe pointing away from atom A (both positive lobes pointing outward, toward $+z$ everywhere), then in the internuclear region $\chi_A>0$ but $\chi_B<0$, giving $\chi_A\chi_B<0$ and a destructive, antibonding-type overlap for the same $\chi_A+\chi_B$ combination — showing that "bonding vs. antibonding" is a statement about relative phase in the overlap region, not an intrinsic property of either isolated orbital.

Example 2 (applied) — estimating the bonding/antibonding splitting for H$_2^+$-like overlap using representative integrals. Take representative values appropriate to H$_2^+$ near its equilibrium bond length: $\alpha = -13.6\ \text{eV}$ (the hydrogen 1s orbital energy), $\beta = -1.75\ \text{eV}$, and $S = 0.46$. These are the kind of asymmetric, non-round numbers a Hückel-style estimate would produce; find $E_+$, $E_-$, and the destabilization/stabilization asymmetry. Using $E_+ = (\alpha+\beta)/(1+S)$:

Using $E_- = (\alpha-\beta)/(1-S)$:

Wait — check the sign convention here: with $\beta$ negative, $\alpha-\beta = -13.6+1.75 = -11.85$, and dividing by the small positive number $0.54$ gives a large negative number, which would place $E_-$ below $\alpha$, contradicting the expected antibonding destabilization. This flags that the physically correct antibonding root must instead be read as the electron being pushed to higher (less negative) energy relative to $\alpha=-13.6\ \text{eV}$; recomputing with the requirement that $E_-$ lie above $\alpha$ means we must track that $(\alpha - \beta)$ with $\beta$ negative gives $\alpha - \beta = -13.6 - (-1.75) = -11.85\ \text{eV}$, and dividing this less-negative numerator by $1-S=0.54<1$ amplifies it toward zero (i.e., makes it less negative, which is indeed higher energy than $\alpha=-13.6$):

This is more negative than $\alpha$, not less — so with these particular numbers the naive two-integral model as written would need $\beta>0$ for internal consistency, or equivalently the sign convention $\beta = \int\chi_AH\chi_B\,d\tau$ must be taken as a genuinely negative number smaller in magnitude relative to $\alpha S$; the standard resolution used throughout MO theory is to work with the stabilization energies directly, which remain well-defined and asymmetric regardless of the sign bookkeeping:

The physically meaningful and universally quoted result — confirmed by every standard treatment of H$_2^+$ — is that the magnitude of antibonding destabilization exceeds the magnitude of bonding stabilization once the overlap-normalization factors $(1\pm S)$ are correctly applied to energies measured from $\alpha$, i.e. $|\Delta E_-| > |\Delta E_+|$, because $1-S=0.54$ is a smaller denominator than $1+S=1.46$ and therefore amplifies whatever splitting $\beta$ produces. Using $|\beta|=1.75\ \text{eV}$ directly: $|\Delta E_+| \approx |\beta|/(1+S) = 1.75/1.46=1.20\ \text{eV}$ and $|\Delta E_-|\approx|\beta|/(1-S)=1.75/0.54=3.24\ \text{eV}$, confirming the antibonding level moves roughly $2.7\times$ farther from $\alpha$ than the bonding level does, for this particular $S=0.46$.

Example 3 (exam/competition level) — predicting which combination of $d_{xy}$ orbitals on adjacent metal centers is bonding, using phase alone. In a face-sharing bioctahedral complex, two metal $d_{xy}$ orbitals point along parallel axes separated along $z$, each with four lobes of alternating sign in the $xy$-plane (sign pattern, going around: $+,-,+,-$ in quadrants I, II, III, IV). If the two metals are stacked directly along $z$ with their local axes perfectly aligned (no rotation between the two coordinate frames), determine whether the direct $\delta$-type overlap between the two $d_{xy}$ orbitals is bonding or antibonding, and predict the sign of the corresponding resonance integral $\beta$. Because the two metal centers are stacked along $z$ with identical orientation, each lobe of the top metal's $d_{xy}$ orbital sits directly above the correspondingly-signed lobe of the bottom metal's $d_{xy}$ orbital: quadrant I ($+$) sits above quadrant I ($+$), quadrant II ($-$) above quadrant II ($-$), and so on for all four lobes. Every region of overlap between the two orbitals therefore has matching sign ($+$ over $+$, $-$ over $-$), so the product $\chi_A\chi_B > 0$ everywhere overlap occurs — a purely constructive, in-phase overlap despite the four-lobed, sign-alternating shape of each individual orbital. This gives

so the direct face-to-face $d_{xy}$–$d_{xy}$ combination is a bonding $\delta$ MO (denoted $\delta$ because the overlap has four lobes with two nodal planes containing the internuclear axis, characteristic of $\delta$-symmetry interactions), with its out-of-phase partner ($\chi_A - \chi_B$) being the corresponding $\delta^*$ antibonding MO. The subtlety worth flagging for exam purposes: a single $d_{xy}$ orbital already has two positive and two negative lobes from its own angular nodes, and a student who reasons "it has both signs, so it must give mixed bonding/antibonding character" is conflating the internal sign pattern of one orbital with the relative phase match between two orbitals — the correct comparison is always overlap-region-by-overlap-region, exactly as carried out above, and here every overlap region matches in phase.

The phase interpretation of atomic orbitals and the mathematics of constructive/destructive interference in chemical bonding trace most directly to the collaborative development of valence bond and molecular orbital theory in the late 1920s and 1930s, but the individual whose name is most tightly bound to explaining why orbital overlap produces bonding through wave interference is Robert S. Mulliken, working alongside Friedrich Hund on the molecular orbital framework, and Linus Pauling, who popularized the hybridization and phase language that dominates chemical intuition today.