Zinc Blende, Wurtzite and Rutile Structures

Drop a crystal of the semiconductor GaAs on a table. Place a crystal of ordinary window-glass filler, rutile TiO₂, beside it. Together they bracket an entire branch of solid-state chemistry. That branch asks: what happens to ionic-solid structures when bonding stops being purely electrostatic? What happens when it starts caring about direction? The Cesium Chloride and Fluorite Structures describe binaries built from hard, round, charge-only ions. These are spheres packed to satisfy Madelung sums. Zinc blende, wurtzite and rutile are different. They appear once the anion and cation begin to share electron density directionally. They appear once bonds start to look like bonds, rather than mutual electrostatic leases. This matters enormously in practice. Zinc blende and wurtzite are the two competing tetrahedral arrangements. Nearly every technologically important MX solid with substantial covalent character adopts one of them. Examples include ZnS itself, GaAs, GaN, CdS, SiC, and AlN. The entire semiconductor and optoelectronics industry is built from atoms in one of these two lattices. The two structures differ only in the stacking sequence of identical tetrahedral layers. Rutile, meanwhile, is the archetypal structure for MX₂ compounds. In these, the cation is too small and too covalently demanding for fluorite's cubic eightfold hole. Instead, it squeezes into a distorted octahedral hole. This drags chains of edge-sharing octahedra into existence. TiO₂, SnO₂, MnO₂, MgF₂ and many transition-metal difluorides crystallize this way. The big picture is a coherent story. It concerns radius ratio and orbital hybridization competing with pure Madelung energetics. Given a stoichiometry (MX or MX₂) and a rough size ratio, the Tetrahedral and Octahedral Holes framework tells you which holes are geometrically available. The degree of covalency — driven by electronegativity difference, polarizability, and orbital overlap — decides which hole the cation actually chooses. It also decides whether close-packing is even the right starting point. Zinc blende is cubic close-packed anions with cations in half the tetrahedral holes. Wurtzite is the hexagonal close-packed twin of the same local environment. Rutile is a distorted hexagonal-ish anion array, with cations in one-half of the octahedral holes. These octahedra are arranged so they share edges into infinite chains. This is a direct structural consequence of a d⁰ or d^n cation. That cation is too small and too charge-dense for the cubic fluorite hole. This article builds all three structures from first principles. These include packing, coordination number, Pauling's rules, and Madelung constants. It shows precisely how covalency, radius ratio and site-preference energetics select one polymorph over another. It sets up directly for Perovskite Structure, a ternary elaboration of the octahedral-network idea. It also sets up for Band Theory of Solids. The same tetrahedral sp³ networks that give zinc blende and wurtzite their geometry generate something else too. They generate exactly the band gaps that make GaAs a semiconductor.

Key takeaways

Zinc blende (sphalerite, ZnS) = ccp S²⁻ with Zn²⁺ in half the tetrahedral holes; wurtzite = hcp S²⁻ with Zn²⁺ in half the tetrahedral holes — same local ZnS₄ tetrahedron, different long-range stacking.
Both structures have coordination number 4:4 (each ion tetrahedrally surrounded by 4 of the other); they are the default choice once bonding gains real covalent/directional character (Pauling radius-ratio rule under-predicts tetrahedral coordination for these).
Zinc blende vs wurtzite is essentially decided by stacking-fault energy and next-nearest-neighbor (second coordination shell) electrostatics — cubic ABC stacking vs hexagonal AB stacking of identical tetrahedral sheets.
Rutile (TiO₂) has 6:3 coordination — octahedral Ti⁴⁺, each O²⁻ shared by 3 octahedra — built from chains of edge-sharing TiO₆ octahedra linked further by corner-sharing, giving the compound its acicular (needle-like) habit.
Radius ratio governs the overall trend M⁺X₈ (CsCl) → M⁺⁺X₆ (rutile/NaCl family) → M⁺⁺⁺⁺X₄ (zinc blende/wurtzite) as rᴹ/rX shrinks and/or covalency rises, but real structures are frequently determined by orbital hybridization (sp³ vs ionic sp³d² preference) rather than hard-sphere geometry alone.
Key idea: tetrahedral MX and distorted-octahedral MX₂ structures are what close-packed ionic frameworks turn into once directional covalent bonding — not just charge and size — starts to dictate which holes the cations will occupy and how those occupied polyhedra link together.

The core insight

Zinc blende, wurtzite, and rutile are all derivable from a close-packed (or near-close-packed) anion lattice with cations in a fraction of the tetrahedral or octahedral holes; which structure appears is decided by radius ratio together with the degree of covalent, directional bonding character, which favors lower coordination numbers (4 rather than 6 or 8) and can even distort the anion sublattice away from ideal close packing (as in rutile).

Where this lives today

Zinc blende and wurtzite are not museum pieces; they are the literal substrate of modern optoelectronics. Gallium arsenide (zinc blende), gallium nitride and indium nitride (wurtzite), cadmium sulfide and cadmium selenide quantum dots (both polytypes, tunable by synthesis conditions), and silicon carbide (both, with dozens of intermediate polytypes) underlie LEDs, laser diodes, solar cells, and high-power/high-frequency transistors. GaN's choice of wurtzite over zinc blende is directly responsible for the built-in spontaneous polarization that gives GaN/AlGaN heterostructures their high-electron-mobility-transistor (HEMT) behavior used in every modern RF power amplifier and 5G base station — a direct, technologically monetized consequence of the hcp-vs-ccp stacking asymmetry analyzed above. The same tetrahedral sp³ network that determines zinc blende/wurtzite geometry is exactly the periodic potential that gives rise to band structure and band gaps, foreshadowing Band Theory of Solids. Rutile TiO₂ is the world's dominant white pigment (paints, sunscreens, toothpaste, plastics) precisely because its very high refractive index (driven by strong, polarizable, covalent Ti–O bonding within the edge-sharing octahedral chains) gives outstanding light-scattering power per unit mass. TiO₂'s photocatalytic activity — self-cleaning coatings, water-splitting photoelectrodes, dye-sensitized solar cells — stems from the band gap set by the same Ti 3d/O 2p covalent network discussed structurally here. Rutile-type SnO₂ and doped variants (fluorine-doped, antimony-doped) are the transparent conducting oxides in touchscreens and low-emissivity glass. MnO₂ in the rutile-related pyrolusite structure is central to battery cathode chemistry (alkaline and some Li-ion primary cells). The rutile edge-sharing-octahedron motif reappears, generalized to a three-dimensional network, in the corner-sharing octahedral frameworks of Perovskite Structure materials (titanates, manganites) that dominate current research in ferroelectrics, colossal magnetoresistance, and photovoltaics.

Ask the tutor: How does the choice between zinc blende and wurtzite structure affect the electronic band structure and optical properties of a semiconductor like GaN?

The physical picture

Begin with the two limiting descriptions you already have. In CsCl, radius ratio near unity and largely ionic bonding gives 8:8 coordination — a simple cubic anion array with cations in the cubic holes. In fluorite (CaF₂), a smaller cation/anion charge ratio (M²⁺X⁻₂) with a moderate radius ratio gives 8:4 coordination — ccp cations with anions filling all tetrahedral holes. Both are essentially electrostatic constructions: maximize Madelung energy subject to the ions not overlapping. Now shrink the cation further, or — more importantly — make the M–X bond more covalent (small electronegativity difference, high cation charge density, or participation of cation d-orbitals/anion polarizability). Two things happen simultaneously. First, geometrically, a genuinely small cation can no longer comfortably fill a cubic (8-coordinate) or even simple octahedral (6-coordinate) hole without the anions clashing; a tetrahedral hole becomes the natural fit. Second and more importantly, covalent bonding actively prefers tetrahedral coordination when the cation can form four equivalent sp³ hybrid bonds — this is not just "the hole that fits" but "the hybridization that lowers total energy," the same sp³ preference you know from carbon and silicon chemistry. Zinc blende and wurtzite are the two ways to stack these ZnS₄ (or, generally, MX₄) tetrahedra into an infinite three-dimensional lattice. Picture a single close-packed layer of anions. Every anion layer, whether stacked ABAB… (hcp) or ABCABC… (ccp), sits atop tetrahedral and octahedral holes in an identical local pattern; the difference between hcp and ccp only appears in the third layer. Cations fill one type of tetrahedral hole (say, the ones pointing "up") in every anion layer, so cation and anion each individually form a diamond-like or wurtzite-like network. In zinc blende the anion sublattice is ccp (Fm-3m, same as diamond cubic when M=X), giving overall cubic $\overline{4}3m$ symmetry. In wurtzite the anion sublattice is hcp, giving hexagonal $6mm$ symmetry. Locally, every atom in both structures has exactly the same coordination environment: four nearest neighbors of the other kind arranged tetrahedrally, at essentially the same M–X distance. The structures are polytypes — different stacking sequences of chemically identical building blocks — and many II–VI and III–V compounds (ZnS, SiC, GaN, CdSe) show both forms, sometimes coexisting as stacking-fault-riddled crystals.

Figure

Zinc blende vs wurtzite: identical tetrahedra, different stacking of anion layers

Rutile takes a different path entirely, because it is an MX₂ compound (not MX): the cation is small, highly charged (typically 4+), and too undersized even for the octahedral hole to sit comfortably surrounded by six equidistant anions in an undistorted close-packed lattice. TiO₂ crystallizes with a distorted hexagonal-close-packed-like O²⁻ array (strictly, a distorted variant, not ideal hcp) with Ti⁴⁺ occupying half the octahedral holes, in a pattern that is not the isolated-octahedra arrangement of NaCl-type structures but instead forms infinite chains of edge-sharing TiO₆ octahedra running parallel to the crystallographic c axis, and these chains are cross-linked by corner-sharing into the tetragonal rutile lattice ( $P4_2/mnm$ ). Each Ti⁴⁺ is 6-coordinate; each O²⁻ is 3-coordinate, bonded to three different Ti⁴⁺ (trigonal-planar-ish local geometry), giving the required 6:3 = 2:1 stoichiometry ratio automatically. Edge-sharing (rather than exclusively corner-sharing, as fluorite's cation array effectively enforces) brings the two Ti⁴⁺ centers of adjacent octahedra unusually close, which is only tolerable because the high formal charge is compensated by substantial covalent Ti–O bonding (partial M–O π character from O 2p → Ti 3d donation) that screens the Ti···Ti electrostatic repulsion.

The core ideas

The organizing relationships across all three structures are radius ratio, Madelung energetics with the Born exponent, and coordination-number bookkeeping.

\rho = \frac{r_{+}}{r_{-}}

where: $r_+$ , $r_-$ are the cation and anion (or smaller/larger ion) radii; $\rho$ predicts, via Pauling's radius-ratio rules, the geometrically preferred coordination number — but for zinc blende/wurtzite/rutile this hard-sphere prediction is frequently overridden by covalency, so $\rho$ sets an outer bound rather than a strict verdict.

U = -\frac{N_A M z_+ z_- e^2}{4\pi\varepsilon_0 r_0}\left(1-\frac{1}{n}\right)

where: $U$ is lattice energy, $M$ the Madelung constant (structure-specific — 1.638 for zinc blende, 1.641 for wurtzite, 2.408 for rutile, all per formula unit referenced consistently), $z_\pm$ ionic charges, $r_0$ the shortest cation–anion distance, and $n$ the Born exponent describing short-range repulsion.

Reading: the near-equality of the zinc-blende and wurtzite Madelung constants (1.638 vs 1.641) is the single most important quantitative fact in this article — it proves, independent of any covalency argument, that pure electrostatics is essentially indifferent between the two stackings. Whatever does select one polytype over the other (bonding-orbital energetics, second-neighbor van der Waals terms, kinetic/growth effects) must come from beyond the point-charge Madelung sum. Rutile's much larger Madelung constant (2.408) simply reflects the higher charges (2+/1− vs 4+/2−) and different geometry, not a "better" structure in any absolute sense — Madelung constants are not comparable across stoichiometries without correcting for charge and formula-unit convention.

\text{CN}_{\text{cation}} : \text{CN}_{\text{anion}} = \frac{1}{x_{\text{cation}}} : \frac{1}{x_{\text{anion}}} \quad \text{for } M_{x}X_{y}

more concretely, electroneutrality/bond-valence bookkeeping requires

n_M \cdot \text{CN}_M = n_X \cdot \text{CN}_X

where: $n_M, n_X$ are the mole numbers of cation and anion per formula unit, and $\text{CN}_M,\text{CN}_X$ are their coordination numbers; this is why rutile's Ti(6) must pair with O(3) — $1\times 6 = 2\times 3$ — and why zinc blende/wurtzite's Zn(4) pairs with S(4) — $1\times4 = 1\times 4$ .

Derivation AWhy zinc blende and wurtzite have essentially identical Madelung constants

Goal. Show, from the geometry of tetrahedral stacking, why the electrostatic (Madelung) energy of zinc blende and wurtzite must be nearly equal, so that the polytype selection is not an electrostatic effect.

Step 1 — identical first coordination shell. In both structures every cation has exactly 4 anion nearest neighbors at the same distance $r_0$ (to an excellent approximation, since ideal wurtzite has $c/a = \sqrt{8/3} \approx 1.633$ , matching the tetrahedral angle exactly; real wurtzite structures deviate only slightly). The nearest-neighbor term in the Madelung sum,

E_1 = -\frac{4 z_+ z_- e^2}{4\pi\varepsilon_0 r_0},

is therefore identical for both structures by construction — this is the dominant term and already fixes ~90%+ of the total Madelung sum's magnitude.

Step 2 — second-shell (like-ion) environment differs only in stacking phase. The next contribution comes from the 12 second-nearest neighbors — same-species ions (cation–cation or anion–anion) that surround a given ion in the close-packed layers. In both ccp and hcp stacking, every ion has 12 nearest neighbors of its own kind at essentially the same distance ( $a$ , the close-packing repeat distance), because the local packing of any single close-packed layer, and the layer directly above and below it, is locally identical in ccp and hcp — the two stackings differ only in the registry of the third layer, which is a next-nearest-neighbor-of-neighbor effect.

Step 3 — third-shell terms are where ccp and hcp diverge, and they are small. The stacking sequence (ABCABC vs ABAB) only changes the relative phase of ions three or more coordination shells away. Because Coulomb interactions fall off as $1/r$ , and because the number of ions in each successive shell is common to both structures out to at least the second shell, the energy difference between the two Madelung constants arises from third-and-further shells, whose individual contributions are down by a factor of order $(r_0/r_{\text{shell}})$ relative to the first shell. Summing the full lattice gives $M_{\text{ZB}} = 1.6381$ and $M_{\text{W}} = 1.6410$ (both referenced per ion pair with $z_+z_- = 1$ ), a difference of about 0.18%.

Step 4 — energetic consequence. Converting the Madelung constant difference into an energy difference via the Born–Landé expression, for a typical $r_0 \approx 235$ pm (ZnS) and $z_+z_-=4$ , the electrostatic preference for wurtzite over zinc blende (or vice versa) is only a few kJ mol⁻¹ per formula unit — utterly swamped by lattice-defect energies, growth kinetics, temperature, and covalent second-order effects (crystal-field-like splitting of bonding/antibonding tetrahedral orbitals, weak differences in π-type overlap between next-nearest neighbors). This is precisely why so many compounds (ZnS, SiC, CdS, GaN) show polytypism, temperature-dependent phase transitions between zinc blende and wurtzite forms, and stacking-fault-rich real crystals: the thermodynamic driving force distinguishing the two forms is genuinely tiny.

Go deeper · Derivation B · 2 minRadius-ratio and covalency criteria for the CsCl → rutile/rock-salt → zinc-blende/wurtzite sequence

Derivation BRadius-ratio and covalency criteria for the CsCl → rutile/rock-salt → zinc-blende/wurtzite sequence

Goal. Derive, from simple sphere-packing geometry, the radius ratios at which 8-, 6-, and 4-coordination become geometrically stable, then show how covalency shifts real compounds toward lower coordination than the ratio alone would predict.

Step 1 — cubic (8-coordinate) limiting ratio. Place a cation at the body center of a cube of anions at the corners (CsCl-type). The critical ratio at which the cation just touches all eight corner anions while corner anions also touch along the cube edge is found from the cube geometry: cation–anion distance is $\frac{\sqrt3}{2}a$ and anion–anion (cube edge, at the point of instability) is $a$ . Setting $r_++r_- = \frac{\sqrt3}{2}a$ and $2r_- = a$ (anions just touching) gives

\rho_{\text{crit,8}} = \frac{r_+}{r_-} = \sqrt{3} - 1 \approx 0.732.

Step 2 — octahedral (6-coordinate) limiting ratio. For a cation at the center of an anion octahedron, the analogous construction (cation touching all six anions, anions touching each other along the octahedron edge) gives

\rho_{\text{crit,6}} = \sqrt{2} - 1 \approx 0.414.

Step 3 — tetrahedral (4-coordinate) limiting ratio. The same construction for a regular tetrahedron of anions around a central cation gives

\rho_{\text{crit,4}} = \sqrt{\tfrac{3}{2}} - 1 \approx 0.225.

Step 4 — where the real compounds sit, and why the rule already starts to fail. For ZnS, $r$ (Zn²⁺, CN4) ≈ 74 pm and $r$ (S²⁻, CN4) ≈ 184 pm gives $\rho \approx 0.40$ — solidly inside the octahedral range (0.414–0.732) by Pauling's rule, yet ZnS adopts 4-coordinate structures (zinc blende/wurtzite), not 6-coordinate rock salt. This is the textbook demonstration that radius-ratio rules, built on hard-sphere ionic-model geometry, systematically under-predict the tendency toward low coordination in compounds with substantial covalent character: directional sp³ hybridization on Zn²⁺ (using 4s and 4p orbitals with significant orbital overlap into S 3p) stabilizes the tetrahedral geometry well beyond what pure electrostatics/packing would select. The same reasoning explains why TiO₂ does not adopt fluorite (which the naive M⁴⁺X₂⁻ formula, high charge, and Ti⁴⁺/O²⁻ radius ratio $\rho = 60.5/140 \approx 0.43$ might suggest is plausible geometrically for an even higher coordination structure): Ti⁴⁺'s small size and high charge density instead drive it into the smaller octahedral, not cubic, hole, consistent with $\rho\approx0.43$ sitting just inside the octahedral range — and Ti–O covalency (π-donation from O 2p into empty Ti 3d) further stabilizes the edge-sharing chain motif over an idealized ionic rock-salt-like alternative, because edge sharing maximizes metal–metal $t_{2g}$ -orbital proximity and network rigidity along the chain axis, a purely covalent/crystal-packing stabilization with no hard-sphere analogue.

Step 5 — the general lesson. Radius ratio correctly predicts the coarse trend (bigger cation/anion ratio → higher coordination) but its quantitative breakpoints are reliable only for strongly ionic solids (alkali halides, alkaline-earth oxides). For MX and MX₂ compounds with electronegativity differences below roughly 2 (Pauling scale) — ZnS (Δχ ≈ 0.9), CdS, GaAs, SiC, TiO₂ (Δχ ≈ 1.9, borderline) — orbital hybridization energetics compete with, and often dominate, simple geometric packing, and this is exactly why the tier of structures discussed here is billed as belonging to "more covalent binaries."

Worked examples

Example 1 (foundational) — coordination-number bookkeeping and formula-unit content in zinc blende. Zinc blende has the ccp anion sublattice, conventional cubic cell edge $a = 541$ pm (real value for ZnS). The ccp anion sublattice contributes the standard ccp content of 4 anions per conventional cell (corners: $8\times\frac18=1$ ; faces: $6\times\frac12=3$ ; total $4$ ). Zinc cations occupy half of the 8 tetrahedral holes in the ccp array — the 4 holes that all have the same "handedness" of the two interpenetrating fcc lattices — so the cell also contains $4$ Zn²⁺.

Z = 4 \ \text{(ZnS formula units per conventional cell)}

Each Zn²⁺ sits at a tetrahedral hole with 4 equidistant S²⁻ neighbors at

r_0 = \frac{\sqrt3}{4}a = \frac{\sqrt3}{4}(541\ \text{pm}) = 234\ \text{pm},

matching the accepted Zn–S bond length in sphalerite (≈235 pm) to within rounding. As a limiting-case check: setting $a\to 0$ is unphysical (ions would overlap), but setting the anion radius equal to the cation radius (a fictitious "isoelectronic" limit) makes $\rho\to1$ and pushes the structure toward diamond-cubic geometry — indeed, when M = X (as in elemental Si, Ge, or diamond itself), the zinc-blende lattice degenerates exactly into the diamond structure, confirming the internal consistency of the tetrahedral-hole construction.

Example 2 (applied) — Madelung-energy comparison of zinc blende vs wurtzite for ZnS, and estimating the polymorph energy gap. Using the Born–Landé expression with Madelung constants $M_{\text{ZB}} = 1.6381$ , $M_{\text{W}} = 1.6410$ , $z_+z_- = 4$ , Born exponent $n \approx 9$ (typical for ions of the Ne/Ar-like configuration relevant to Zn²⁺/S²⁻), and $r_0 = 234$ pm from Example 1:

U = -\frac{N_A M z_+ z_- e^2}{4\pi\varepsilon_0 r_0}\left(1-\frac1n\right).

Compute the prefactor common to both structures:

\frac{N_A e^2}{4\pi\varepsilon_0 r_0} = \frac{(6.022\times10^{23})(1.602\times10^{-19})^2}{4\pi(8.854\times10^{-12})(234\times10^{-12})}\ \text{J mol}^{-1} = 5.93\times10^{5}\ \text{J mol}^{-1}.

Then

U_{\text{ZB}} = -(1.6381)(4)(5.93\times10^{5})\left(1-\tfrac19\right)\ \text{J mol}^{-1} = -3.454\times10^{6}\ \text{J mol}^{-1} = -3454\ \text{kJ mol}^{-1},

U_{\text{W}} = -(1.6410)(4)(5.93\times10^{5})\left(1-\tfrac19\right)\ \text{J mol}^{-1} = -3460\ \text{kJ mol}^{-1}.

The difference,

\Delta U = U_{\text{W}} - U_{\text{ZB}} \approx -6\ \text{kJ mol}^{-1},

says the wurtzite Madelung sum favors wurtzite by only about 6 kJ mol⁻¹ out of a total lattice energy of roughly 3450 kJ mol⁻¹ — a relative difference of 0.2%, confirming quantitatively the qualitative claim from the derivation above. For comparison, the actual measured enthalpy difference between real zinc-blende and wurtzite ZnS phases is only a few kJ mol⁻¹ (they are near-degenerate and interconvert around 1020 °C), consistent with the Madelung estimate but reminding us that the full energy difference also contains second-order covalent terms not captured by the point-charge model. As a limiting-case check, if the two Madelung constants had been artificially set exactly equal ( $M_{ZB}=M_W$ ), the formula correctly predicts $\Delta U = 0$ , i.e., no thermodynamic preference at all from electrostatics alone — precisely the near-realized situation.

Example 3 (exam / competition level) — deriving the O²⁻ coordination number in rutile from stoichiometry and octahedral occupancy, then checking the Ti···Ti edge-sharing distance. Rutile's tetragonal cell (space group $P4_2/mnm$ ) has $a = 459$ pm, $c = 296$ pm, with $Z = 2$ TiO₂ formula units per cell. Given that each Ti⁴⁺ is octahedrally 6-coordinate, use pure electroneutrality/coordination bookkeeping (no need for the full crystal structure) to find the O²⁻ coordination number, then rationalize the short c-axis repeat in terms of the edge-sharing octahedral chain. Coordination number of oxygen. Electroneutrality of bond count requires

n_{\text{Ti}}\cdot \text{CN}_{\text{Ti}} = n_{\text{O}} \cdot \text{CN}_{\text{O}}.

With $n_{\text{Ti}} = 1$ , $n_{\text{O}} = 2$ (per TiO₂ formula unit) and $\text{CN}_{\text{Ti}} = 6$ :

1 \times 6 = 2 \times \text{CN}_{\text{O}} \quad\Rightarrow\quad \text{CN}_{\text{O}} = 3.

So each O²⁻ bridges exactly three Ti⁴⁺ centers, consistent with the trigonal-planar-like local oxygen environment described earlier — this is a purely stoichiometric deduction, requiring no diagram. Edge-sharing chain repeat. Chains of edge-sharing TiO₆ octahedra run parallel to $c$ ; because two octahedra sharing an edge have their metal centers separated by roughly one octahedral edge length rather than a full corner-to-corner span, the repeat distance along the chain (essentially the $c$ -axis length, 296 pm) is markedly shorter than it would be for exclusively corner-sharing octahedra (which, for comparable Ti–O bond lengths ≈ 198 pm, would give a Ti···Ti corner-shared repeat close to $2\times198 = 396$ pm along a linear M–O–M corner bridge). Concretely, for a near-regular TiO₆ octahedron with Ti–O ≈ 198 pm, the O–O edge length is

\ell_{\text{edge}} = \sqrt{2}\,d_{\text{Ti–O}} = \sqrt2 (198\ \text{pm}) = 280\ \text{pm (unshared edge)},

while the shared edge in rutile is compressed (a well-known distortion of the rutile octahedron) to about 255 pm, pulling the two edge-sharing Ti⁴⁺ centers to a Ti···Ti separation of roughly 296 pm — matching the crystallographic $c$ parameter almost exactly, confirming that the c-axis repeat in rutile is structurally set by the edge-sharing octahedral chain, not by a simple close-packed anion repeat distance. As a limiting-case sanity check: if rutile instead used exclusively corner-sharing octahedra (the fluorite/rock-salt-like alternative), the expected metal–metal repeat would nearly double relative to the observed 296 pm — which is exactly why edge-sharing, not corner-sharing, is diagnostic of the rutile-type (and related MX₂ structures like MgF₂ and MnO₂) versus the isolated-octahedra motifs of NaCl-type MX compounds.

Limiting cases, scope & common traps

Regime / limit	What happens	Structure realized
$\rho\to 1$ , purely ionic, MX	Both ions same size, cubic (8:8) hole preferred	CsCl-type
$\rho \approx 0.4$ –0.7, ionic, MX	Octahedral (6:6) hole preferred	NaCl-type
$\rho \approx 0.2$ –0.4 but strong covalency, MX	sp³ hybridization overrides geometry; tetrahedral (4:4) preferred despite ratio suggesting octahedral	Zinc blende / wurtzite
Zinc blende vs wurtzite choice	Madelung energies nearly identical (ΔM ≈ 0.2%); decided by weak second-order covalent/packing terms, temperature, kinetics	Polytypism, phase transitions (e.g., ZnS at ~1020 °C)
Small, highly charged cation, MX₂	Too small for cubic fluorite hole; occupies half the octahedral holes; edge-sharing needed to satisfy 2:1 stoichiometry with 6:3 coordination	Rutile-type (TiO₂, SnO₂, MgF₂, MnO₂)
$M = X$ (isoelectronic limit)	Zinc-blende lattice degenerates to elemental diamond-cubic lattice	Diamond, Si, Ge, SiC (mixed)
High-pressure limit for ZnS/GaAs-type	Increasing pressure favors higher coordination (6:6) as compressibility overcomes covalent tetrahedral preference	NaCl-type high-pressure polymorphs

The chemist behind it

The person most associated with turning "which structure forms" into a predictive, quantitative rule is Linus Pauling, whose 1929 paper "The Principles Determining the Structure of Complex Ionic Crystals" formalized the radius-ratio criteria and the electrostatic-valence and coordination rules used throughout this article — even though, ironically, his own rules are the ones that most dramatically fail for zinc blende, which is precisely why zinc blende/wurtzite became a classic teaching example of covalency overriding naive ionic-radius geometry. The structural determinations themselves — zinc blende, wurtzite, and rutile — came earlier, from the pioneering X-ray crystallography of W. H. Bragg and W. L. Bragg in the 1910s, who solved ZnS and TiO₂ among the very first inorganic crystal structures ever determined by X-ray diffraction, establishing the method itself as much as the structures.

Origins · 1 minLinus Carl Pauling and how this was discovered

Linus Carl Pauling

1901–1994 · American chemist; Caltech; formalized ionic radius-ratio rules, electronegativity scale, and the nature of the chemical bond

Pauling's 1929 rules for ionic crystal structures — radius ratio, electrostatic valence, and (for complex structures) sharing of polyhedral edges/corners — remain the standard first-pass toolkit for predicting coordination numbers in ionic solids, even though, as this article emphasizes, they require correction once covalent, directional bonding becomes significant. His parallel work on electronegativity and hybridization (sp³, sp²) supplies the other half of the argument needed to understand why zinc blende and wurtzite defy the naive radius-ratio prediction.

History

- 1912–1914: W. H. and W. L. Bragg develop X-ray crystal structure determination; early mineral structures including zinc blende (ZnS) and rutile (TiO₂) solved among the first inorganic crystal structures.

- 1913: Sphalerite (zinc blende) structure of ZnS established, revealing the ccp-anion, half-tetrahedral-hole motif.

- 1920s: Wurtzite (hexagonal ZnS) structure solved, establishing it as the hcp polytype of the same tetrahedral coordination.

- 1929: Linus Pauling publishes radius-ratio and electrostatic-valence rules for predicting ionic crystal coordination numbers.

- 1950s–60s: Systematic polytypism studies (notably in SiC) reveal dozens to hundreds of stacking polytypes, cementing the modern understanding that zinc blende/wurtzite selection is a stacking-energetics problem, not a bonding-type problem.

Why don't zinc blende and wurtzite just interconvert freely at room temperature if their energies are so close?

Although the thermodynamic driving force between the two polytypes is small (a few kJ mol⁻¹, dominated by second-order covalent and lattice-vibrational terms beyond the near-identical Madelung sums), interconversion requires a cooperative, coordinated re-stacking of entire close-packed layers — breaking and reforming a large number of bonds simultaneously along extended stacking faults. This has a substantial kinetic barrier, so at ordinary temperatures the polytype set during crystal growth is effectively "frozen in," and only at high temperature (for ZnS, around 1020 °C) does thermal energy allow the bulk phase transition to proceed on laboratory timescales.

Does every MX compound with tetrahedral coordination have to be either zinc blende or wurtzite?

These are the two most common tetrahedral MX polytypes, but not the only imaginable ones — because the underlying anion sublattice is close-packed, in principle any ABAB.../ABCABC.../more exotic periodic stacking sequence could occur, and indeed silicon carbide alone exhibits over 200 distinct polytypes (labeled by Ramsdell notation, e.g., 3C, 2H, 4H, 6H) differing only in stacking period. Zinc blende (the purely cubic, "3C" sequence) and wurtzite (the purely hexagonal, "2H" sequence) are simply the two simplest, most common limiting cases.

Why does rutile have a distorted rather than ideal close-packed oxygen sublattice?

An undistorted hexagonal close-packed oxygen array, with cations in exactly half the octahedral holes, would place those cations at relatively long, purely ionic-like Ti–O distances and would not naturally accommodate the short, covalently reinforced Ti···Ti contacts along edge-shared octahedral chains. The oxygen sublattice relaxes (compressing the shared O–O edges from roughly 280 pm to about 255 pm) to let the octahedral chains form efficiently, trading a small ideal-packing penalty for a larger covalent bonding stabilization — a compromise geometry rather than either pure ionic close-packing or pure covalent network geometry.

How does radius ratio explain why TiO2 doesn't adopt the fluorite structure, given that fluorite is also an MX2 structure?

Fluorite requires cations large enough to sit comfortably in the cubic eightfold hole of a ccp anion lattice (roughly ρ > 0.73 for ideal hard-sphere stability, though real fluorites like CaF₂ sit somewhat below this). Ti⁴⁺ (ionic radius ≈ 60.5 pm, CN6) is far too small relative to O²⁻ (≈140 pm) for cubic 8-coordination; its radius ratio (~0.43) falls squarely in the octahedral range instead, so it occupies octahedral holes as in rutile rather than cubic holes as in fluorite. High cation charge (4+) also makes the shorter, more covalent Ti–O octahedral bonding more electrostatically and orbital-overlap favorable than a longer, more ionic 8-coordinate arrangement would be.

Is zinc blende more "covalent" than wurtzite, or vice versa?

No — because they share essentially identical local M–X bonding (same bond length, same coordination, same first-shell electron density), zinc blende and wurtzite have essentially the same degree of covalency in the primary Zn–S bond. The covalent character of an MX compound is set chiefly by electronegativity difference and orbital overlap, properties of the chemical identity of M and X, not of which stacking sequence the ccp/hcp anion sublattice happens to adopt.

Why is rutile's cation coordination number 6 while zinc blende/wurtzite's is 4, even though all three are "covalent" binaries?

The key variable is stoichiometry combined with cation size/charge, not "how covalent" the bond is in some vague absolute sense. MX compounds (1:1) with small, highly charge-dense, strongly covalent-bonding cations (Zn²⁺, Cd²⁺, Ga³⁺) favor tetrahedral sp³ hybridization at 4-coordination. MX₂ compounds (1:2) with small, highly charged, d⁰-ish or d-poor cations (Ti⁴⁺, Sn⁴⁺, Mn⁴⁺) are geometrically and electrostatically pushed toward octahedral 6-coordination, with the 2:1 stoichiometry then forcing 3-coordinate oxygen and edge-sharing chain connectivity to satisfy both bond-count and covalent Ti–O π-donation stabilization.

What is the practical significance of edge-sharing versus corner-sharing octahedra in oxide structures generally?

Edge-sharing octahedra place metal centers closer together than corner-sharing ones, which raises electrostatic cation–cation repulsion but, when compensated by covalent metal–oxygen bonding (as in rutile), produces rigid one-dimensional or chain-like connectivity with anisotropic physical properties (rutile's needle-like crystal habit, anisotropic refractive index and conductivity). Corner-sharing octahedral networks, by contrast (as in the perovskite and NaCl-related structures), tend to give more isotropic three-dimensional connectivity, generally lower internal strain, and are the default when the cation charge is lower or the network needs more flexibility (as in many perovskite ferroelectrics), making the edge-vs-corner-sharing distinction a direct structural predictor of anisotropy in physical properties.