The perovskite aristotype is the corner-sharing octahedral framework that organizes a huge slice of functional-materials physics: ferroelectricity, the manganite/cuprate oxides, and the halide photovoltaics. The ideal cell is the cubic prototype (, No. 221), a network of octahedra sharing every corner, with the larger cation filling the 12-coordinate cuboctahedral cavity between them. Almost all the interesting physics lives in the distortions of this prototype — octahedral tilting, polar off-centering, Jahn–Teller elongation — and the recurring lattice-dynamics theme is that the high-symmetry cubic phase is often dynamically unstable at , stabilized into existence only by anharmonicity and temperature.
Ideal cubic structure and stoichiometry
In the prototype the cation sits at the cell origin (Wyckoff ), the cation at the body center (), and the three anions at the face centers and permutations (), midway along each – edge. Each is octahedrally coordinated by six ; each octahedron shares all six corners with neighbors, so the framework is , and the cation occupies the resulting cuboctahedral cavity. Equivalently the structure is a cubic-close-packed array of layers with in one quarter of the octahedral holes (those coordinated entirely by ).
Charge neutrality admits several families. For oxides () the two common splittings are (CaTiO₃, SrTiO₃, BaTiO₃) and (LaMnO₃, LaAlO₃, the rare-earth orthoferrites); (KNbO₃, NaTaO₃) also occurs. For halides () neutrality forces — the optoelectronic lead/tin iodides CsPbI₃, CsSnI₃, MAPbI₃ all sit here. The cubic prototype has formula unit per cell; the distorted variants below enlarge the cell ( for tetragonal, for the orthorhombic structure).
Geometric stability: tolerance and octahedral factors
Whether a given triple adopts the perovskite topology, and with how much distortion, is captured to first order by two ionic-radius ratios. Goldschmidt’s tolerance factor measures how well the – and – bond lengths match the geometric constraint of the cubic cell — in the ideal structure the cube edge is along the –– axis and the face diagonal is along the – contact, so equality of the two requires
The octahedral factor asks separately whether the cation is large enough to sit stably in the octahedral cage; the rigid-sphere lower bound for octahedral coordination is , and very large pushes toward higher coordination. The empirical regimes are:
- (–): the cubic prototype is geometrically comfortable (SrTiO₃, the cubic halide perovskites at high ).
- (–): the cation is too small for the cavity, the framework relieves the underbonded by cooperative octahedral tilting, lowering the symmetry to tetragonal / orthorhombic / rhombohedral (CaTiO₃, LaMnO₃, GdFeO₃).
- : the cation is too large (or too small); corner-sharing can no longer accommodate it and the structure goes to hexagonal face-sharing polytypes or abandons the perovskite topology altogether (e.g. BaNiO₃-type, or the non-perovskite -phases of the cesium lead/tin halides).
This is a guide, not a law. predicts formability only probabilistically — a substantial fraction of compounds are not perovskites, and some outside the window are — because it ignores covalency, the -cation /lone-pair electronic driving forces, and temperature. It is especially treacherous for hybrid halide perovskites, where the site is a molecular cation (methylammonium , formamidinium ) with no well-defined hard-sphere radius; one assigns an effective radius from the rotational envelope, and the classification degrades. The standard refinement is the data-driven Bartel tolerance factor , a single dimensionless combination of , , and the -cation oxidation state fitted to a large structure database, which separates perovskite from non-perovskite far more reliably than across both oxides and halides (Bartel et al., Sci. Adv. 5, eaav0693 (2019)).
Octahedral tilting, Glazer notation, and phase transitions
When , the rigid octahedra rotate about pseudocubic axes to shorten – contacts while preserving corner connectivity (the octahedra are far stiffer than the – cage). Glazer notation classifies the resulting tilt systems by the rotation amplitude about each of the three cubic axes and the phase of successive layers: a superscript means adjacent octahedra along that axis rotate in phase, means out of phase, and means no tilt. Thus is the untilted cube; is a single in-phase tilt (tetragonal); (two equal out-of-phase tilts plus one in-phase) is the ubiquitous / orthorhombic system of most distorted oxide and low- halide perovskites; gives rhombohedral . The 23 distinct Glazer systems and their space groups follow from group theory on the - and -point tilt modes (Glazer, Acta Cryst. B 28, 3384 (1972); Howard & Stokes, Acta Cryst. B 54, 782 (1998)).
A representative cooling sequence is cubic tetragonal orthorhombic, each step condensing an additional tilt (or polar) instability:
The microscopic origin is a set of soft modes at the cubic zone boundary. The two tilt instabilities live at the point (out-of-phase tilts, irrep) and the point (in-phase tilts, ); the ferroelectric/off-centering instability lives at . Computing the dynamical matrix of the cubic phase at therefore generically returns imaginary frequencies , : the harmonic energy surface is a multi-well, the cubic structure a saddle point. Reading this correctly is the whole subject of imaginary modes: artifact vs. instability — a tilt mode at / that deepens with -mesh and supercell convergence is a genuine instability, distinct from the spurious imaginary branches near that signal an under-converged calculation or a violated acoustic sum rule. Because the cubic phase is entropically stabilized — the free-energy minimum at finite sits at zero average tilt even though the enthalpy does not — its phonons must be computed anharmonically: the self-consistent harmonic approximation (anharmonic phonons & SSCHA) renormalizes the soft / modes to real, positive frequencies above and locates the transitions as the temperature where the renormalized passes through zero.
Key examples
Inorganic halide perovskites — CsPbI₃, CsSnI₃. The flagship single-junction and tandem photovoltaic absorbers. Their defining complication is polymorphism / phase stability: the photoactive black perovskite phase ( cubic at high , distorting to tetragonal and orthorhombic on cooling) is only marginally stable against a yellow, photo-inactive non-perovskite -phase (edge-sharing, -type for CsPbI₃) into which it readily converts at room temperature and humidity. CsPbI₃ has — well inside the tilting regime and close to the perovskite/non-perovskite boundary — which is exactly why the black phase is so fragile and why A-site/B-site alloying (Br, FA, Sn) is used to push back toward 1. The electronic structure is dominated by the lone pair of the cation (Pb , Sn ): the valence-band maximum is antibonding –, giving the small effective masses and defect tolerance that make these materials good absorbers, while the conduction band is . Heavy-element spin–orbit coupling is not optional — for Pb it splits the conduction band by and reorders the band edges, so any quantitative gap requires SOC. CsSnI₃ adds the Sn/Sn oxidation problem (self-doping to a degenerate -type metal).
Hybrid organic–inorganic — MAPbI₃, FAPbI₃. Replacing Cs⁺ with the molecular methylammonium on the site gives MAPbI₃, the compound that launched the perovskite-PV field (Kojima et al., J. Am. Chem. Soc. 131, 6050 (2009)). The dipolar MA⁺ cation rotates nearly freely in the cavity at room temperature; its reorientational dynamics couple to the tilts and broaden the picture beyond a static lattice, but the band edges remain a lead-iodide framework property. MAPbI₃ runs the same cubic tetragonal (, ) orthorhombic (, ) tilt sequence. Formamidinium FAPbI₃ has a larger cation ( closer to 1) and a more favorable gap, making it the current absorber of choice, but it shares the black-/yellow- stability problem and is usually stabilized by Cs/MA/Br alloying.
Oxide perovskites. The classical functional oxides:
- CaTiO₃ — the namesake mineral (perovskite, after L. A. Perovski). Strongly tilted (, but tilted); the prototype distorted oxide perovskite.
- SrTiO₃ — the canonical quantum paraelectric: an incipient ferroelectric whose polar mode softens on cooling but is arrested by quantum fluctuations before it condenses, so it never becomes ferroelectric at ambient pressure. It does undergo an antiferrodistortive -point tilt transition at (cubic tetragonal ), a textbook soft-mode zone-boundary instability.
- BaTiO₃ — the prototypical ferroelectric: the polar mode condenses, taking the structure through cubic tetragonal orthorhombic rhombohedral on cooling, each phase with the Ti off-centered along a progressively lower-symmetry direction (an order–disorder/displacive hybrid). The Ti second-order Jahn–Teller off-centering is the driver.
- LaMnO₃ — with high-spin Mn (, ): a cooperative Jahn–Teller elongation of the octahedra, orbital ordering, and A-type antiferromagnetism — the parent of the colossal-magnetoresistive manganites.
Quantitative band gaps and the magnetic/orbital ordering of the late- oxides (LaMnO₃ and relatives) are a stress test for the exchange–correlation functional: semilocal DFT badly underestimates the gap and can miss the insulating, orbitally ordered ground state because of self-interaction error, and the standard fixes — Hubbard or hybrid functionals — are needed to recover the correct ground state. The same SIE caution, plus mandatory SOC, governs the halide-perovskite gaps.
Prerequisites
crystal symmetry & space groups · ionic radii, close packing & coordination polyhedra (foundational crystal chemistry)
Builds toward: perovskite variants — double & antiperovskites
Key references
- Tolerance factor (original) — V. M. Goldschmidt, Naturwissenschaften 14, 477 (1926).
- Octahedral tilt classification — A. M. Glazer, Acta Cryst. B 28, 3384 (1972); group-theoretical analysis: C. J. Howard & H. T. Stokes, Acta Cryst. B 54, 782 (1998).
- Revised (data-driven) tolerance factor — C. J. Bartel, C. Sutton, B. R. Goldsmith, R. Ouyang, C. B. Musgrave, L. M. Ghiringhelli & M. Scheffler, Sci. Adv. 5, eaav0693 (2019).
- Hybrid perovskite photovoltaics (first report) — A. Kojima, K. Teshima, Y. Shirai & T. Miyasaka, J. Am. Chem. Soc. 131, 6050 (2009).
- Halide-perovskite electronic structure & SOC — J. Even, L. Pedesseau, J.-M. Jancu & C. Katan, J. Phys. Chem. Lett. 4, 2999 (2013).