Ionic Radii and Isoelectronic Series

Ionic radii and the isoelectronic-series logic behind them are load-bearing tools well beyond the descriptive-chemistry classroom. In battery materials science, the choice between Na⁺- and Li⁺-based intercalation chemistries hinges directly on radius: Na⁺ (102 pm, CN=6) is roughly 45% larger than Li⁺ (76 pm, CN=6), which changes diffusion barriers, preferred coordination polyhedra, and achievable energy density in sodium-ion versus lithium-ion battery cathodes — a direct, economically consequential echo of the isoelectronic/periodic size logic developed here. In geochemistry and mineral physics, radius-ratio and CN-dependent radius arguments predict which trace cations substitute for which major cations in silicate and oxide minerals (e.g., why Fe²⁺ substitutes readily for Mg²⁺ in olivine, and why larger ions like Sr²⁺ preferentially replace Ca²⁺ rather than smaller Mg²⁺), underpinning models of planetary differentiation and ore-deposit formation. In coordination chemistry and supramolecular design, crown ethers and cryptands are engineered with cavity sizes tuned to match specific alkali or alkaline-earth ionic radii (the classic size-selectivity of 18-crown-6 for K⁺ over Na⁺ or Li⁺ is a direct application of the radius differences tabulated here), a principle now exploited in ion-selective electrodes and in nuclear waste remediation (selective sequestration of Cs⁺ or Sr²⁺ from radioactive waste streams). In solid-state materials design, the tolerance factor for perovskite structures (ABX₃, widely used for solar-cell and ferroelectric materials) is built directly from Shannon ionic radii of the A-site, B-site, and X-site ions, and correctly predicting whether a candidate composition will form a stable perovskite, and in which distorted polymorph, remains a first-pass screening tool in high-throughput materials discovery. Finally, the lanthanide contraction discussed here — the compression of 4f-block radii that leaves 4d and 5d transition-metal ions of the same group nearly indistinguishable in size — is the direct forerunner of the far larger, spin-orbit-driven size and reactivity anomalies encountered among the heaviest elements, where Relativistic Effects in Heavy Elements take over as the dominant organizing principle.

Ask the tutor: How would the radius-ratio prediction and the CFSE argument change if magnetite's Fe²⁺ were replaced by a d⁰ or d¹⁰ ion instead?

An ion's electron cloud has no sharp boundary — the electron density falls off exponentially rather than stopping at some radius $r_{\rm ion}$. What crystallography actually measures is a set of internuclear distances $d(M\text{–}X)$ from X-ray or neutron diffraction. To convert a distance into two separate radii, one needs an assumption that partitions $d$ between cation and anion. Historically this partition was done in different ways (Landé's assumption that anion–anion contact along a face diagonal fixes the anion radius; Pauling's assumption that radius scales inversely with effective nuclear charge for isoelectronic ions), producing different absolute scales. The scale in universal modern use is Shannon's (1976, refined from Shannon and Prewitt 1969), built by fixing $r(\mathrm{O}^{2-}) = 140$ pm in six-coordination and then adjusting all other radii, cation and anion, self-consistently so that summed radii reproduce thousands of measured bond lengths across many structure types. This is why Shannon radii are internally consistent (radius sums correctly predict $d$) even though the absolute cation/anion split is a convention. Two physical effects are folded into any such radius. The first is electronic: for a fixed nuclear charge $Z$, adding electrons increases electron-electron repulsion and expands the cloud (anion formation, or descending an isoelectronic series toward lower Z), while removing electrons shrinks it — dramatically so if an entire principal shell is emptied, as happens for main-group cations that lose their valence $ns/np$ electrons entirely and expose the smaller underlying noble-gas core. The second is geometric: the same ion's tabulated radius grows with coordination number, because in a more crowded coordination polyhedron the surrounding anions (or cations) push each other apart, allowing (and requiring) the central ion's effective radius, deduced from $M$–$X$ distances, to expand along with it. Both effects are large enough to move an ion between different preferred coordination geometries, which is why radius-ratio arguments are used, cautiously, to rationalize crystal structure choice.

The starting point is the definition of effective nuclear charge acting on the outermost electrons, from Slater's rules or a self-consistent-field calculation:

where:

• $Z$ is the atomic number (nuclear charge),
• $\sigma$ is the screening constant from all other electrons,
• $Z_{\rm eff}$ is the net attractive charge felt by a reference electron in the outermost shell.

Reading: for a fixed electron configuration, increasing $Z$ while holding the electron count (and hence $\sigma$, to good approximation) fixed increases $Z_{\rm eff}$ linearly with $Z$. This is exactly the condition met by an isoelectronic series. A simple hydrogen-like scaling gives the qualitative size law: the most probable radius (or expectation value $\langle r \rangle$) of an orbital with principal quantum number $n$ scales as

where:

• $n$ is the principal quantum number of the outermost occupied shell,
• $Z_{\rm eff}$ is as above,
• the proportionality constant absorbs the Bohr radius and angular-momentum-dependent corrections.

Reading: this is the single equation underlying essentially every radius trend in the periodic table. Within a period (constant $n$, roughly constant screening from underlying shells), radius falls as $Z$ rises. Down a group ($n$ increases by one shell, and $Z_{\rm eff}$ changes only modestly because of $s$/$p$ penetration and $d$/$f$-block screening anomalies), radius rises — except where the lanthanide/actinide contraction intervenes. The Shannon radius sum rule, which makes all of this operational for crystal chemistry, is

where:

• $d(M\text{–}X)$ is the experimentally measured internuclear (bond) distance,
• $r_M(\mathrm{CN})$, $r_X(\mathrm{CN})$ are the tabulated effective ionic radii of cation and anion at the coordination number found in that particular structure,
• the split between $r_M$ and $r_X$ is fixed once and for all by the convention $r(\mathrm{O}^{2-}, \mathrm{CN}=6) = 140\ \mathrm{pm}$.

Reading: because both radii carry an explicit CN label, the same ion has several correct radii in the table — one for each coordination environment it is found in — and using the wrong one for the structure at hand is the single most common source of error in applying these tables. A useful empirical rule of thumb for the CN dependence, consistent with Shannon's compilations, is that radius increases roughly linearly with CN over the accessible range (CN 4 to 8 for most main-group and transition-metal ions):

where:

• $\beta$ is an empirical slope, typically 3–5 pm per unit of coordination number for main-group and $d$-block cations,
• $r(\mathrm{CN}=6)$ is the commonly tabulated "default" octahedral radius.

Reading: this is not a fundamental law — it is a compact way to interpolate/extrapolate Shannon's tabulated CN=4, 6, 8 values when a CN=5 or CN=7 radius is needed and not separately tabulated. Finally, the radius-ratio rule connects these sizes to preferred coordination geometry by a purely geometric argument (derived below):

where:

• $r_+/r_-$ is the ratio of cation to anion radius,
• the critical ratios mark the geometric threshold below which anion–anion contact forces a lower coordination number.

Reading: the rule is a lower bound, not a guarantee — real structures are also controlled by covalency, Madelung energetics, and crystal-field stabilization, so radius-ratio predictions succeed best for hard, largely ionic salts (alkali/alkaline-earth halides and oxides) and fail more often for softer or more covalent pairs.

Example 1 (foundational): Cation vs. atom vs. anion for the same element, sulfur. Compare atomic sulfur, S²⁻, and (hypothetically) S⁴⁺/S⁶⁺, to see the size collapse on ionization in both directions. Atomic radius of S (covalent radius, roughly) is about 105 pm; sulfide, S²⁻, six-coordinate Shannon radius is 184 pm — nearly 75% larger than the neutral atom, because two extra electrons enter the same $3p$ shell, increasing electron-electron repulsion sharply while $Z_{\rm eff}$ on any one $3p$ electron decreases (screening from the added electrons outweighs the fixed $Z=16$). Going the other way, S⁴⁺ (six-coordinate, as found e.g. in some sulfite-related solids) has $r \approx 37$ pm and S⁶⁺ (as in sulfate-type environments, four-coordinate) is close to 12 pm — removing all six $3s^23p^4$ valence electrons exposes the compact $[\mathrm{Ne}]$ core, held by a large effective nuclear charge with no more $n=3$ shell to buffer it. The full sequence

illustrates, within a single element, both halves of the general isoelectronic/ionization argument: adding electrons expands the cloud; removing them (especially emptying a shell) collapses it, with the largest single drop occurring exactly when the outermost shell is entirely vacated.

Example 2 (applied): Which spinel or which halide structure — comparing Mg²⁺ and Fe²�high spin) via radius ratio. MgO and FeO both crystallize with the rock-salt structure; use radius-ratio reasoning to check this is expected, and then use CN-dependence to show why care is needed comparing across techniques. Shannon six-coordinate radii: $r(\mathrm{Mg}^{2+}) = 72.0$ pm, $r(\mathrm{Fe}^{2+}, \text{high spin}) = 78.0$ pm, $r(\mathrm{O}^{2-}) = 140$ pm. Computing both ratios,

Both values lie comfortably in the 0.414–0.732 window for octahedral coordination, correctly predicting the rock-salt structure for both — and correctly predicting that FeO, with its slightly larger cation, sits a bit further from the CN=4 boundary than MgO, i.e., is even more securely six-coordinate. Now suppose someone mistakenly used the four-coordinate radius of high-spin Fe²⁺ ($r=63$ pm, appropriate to a tetrahedral site as found in some ferrites) in place of the correct six-coordinate value: they would compute $\rho = 63/140 = 0.450$, still inside the octahedral window but close enough to the 0.414 boundary to wrongly suggest FeO might be borderline tetrahedral, when in reality using the correct CN-matched radius shows it is not close to that boundary at all. This is precisely the trap the CN label on Shannon radii is designed to prevent: radii must be pulled from the table entry matching the coordination environment actually being modeled, never mixed across CN values.

Example 3 (exam/competition level): Predicting the inverse-vs-normal spinel preference of Fe₃O₄ using ionic radii and site-preference energetics. Magnetite, Fe₃O₄, is formally Fe²⁺Fe³⁺₂O₄ and adopts the inverse spinel structure — Fe³⁺ ions split evenly between tetrahedral and octahedral sites, while Fe²⁺ occupies only octahedral sites — rather than the normal spinel arrangement (all Fe²⁺ in tetrahedral sites, all Fe³⁺ octahedral). Rationalize this using ionic radii together with crystal-field stabilization energy (CFSE), showing that radius arguments alone are insufficient. Shannon radii: high-spin Fe²⁺ is $r=63$ pm (CN=4) or $78$ pm (CN=6); high-spin Fe³⁺ is $r=49$ pm (CN=4) or $64.5$ pm (CN=6). Purely on size grounds, the larger ion, Fe²⁺, "should" prefer the larger octahedral hole and the smaller ion, Fe³⁺, "should" prefer the smaller tetrahedral hole — that is exactly the normal spinel prediction, and it is wrong for magnetite. The resolution requires bringing in Periodic Trends in Atomic Radii's companion concept, crystal-field theory: Fe²⁺ is a $d^6$ ion and Fe³⁺ is $d^5$. In an octahedral field, high-spin $d^6$ has CFSE $= -0.4\Delta_o$ (four electrons in $t_{2g}$, two in $e_g$, net stabilization beyond the spherical average), while high-spin $d^5$ has CFSE $=0$ exactly (half-filled shell, no net stabilization in either geometry). Since only Fe²⁺ gains a nonzero, geometry-dependent stabilization energy and it gains more of it in octahedral than tetrahedral coordination, the electronic driving force pulls Fe²⁺ preferentially into the octahedral sites — directly opposing the naive radius-ratio packing argument. Because Fe³⁺ is CFSE-indifferent (0 either way), it distributes across both remaining sites, which is exactly the outcome observed: Fe³⁺ tetrahedral + Fe³⁺ octahedral + Fe²⁺ octahedral, the inverse spinel. This example shows explicitly that radius-ratio/geometric packing arguments give a first, size-based guess, but electronic stabilization energies (CFSE) can override that guess whenever a partially-filled $d$-shell with nonzero, site-dependent stabilization is present — precisely the caveat flagged in the "Limiting cases" discussion below.