# Grain boundary A **grain boundary** is the interface between two [[Crystallite|crystallites]] of the same phase whose lattices meet at an angle. Almost every metal, [[Ceramic|ceramic]] and rock is a polycrystal: it solidified from many separate nuclei, each of which grew with its own orientation until it ran into its neighbours, and the surfaces where those growth fronts met are the grain boundaries. They occupy a vanishing fraction of the volume — a two-atom-thick film in a structure whose grains are micrometres across — and yet they govern strength, [[Diffusion|diffusion]], [[Corrosion|corrosion]] and [[Electrical_resistivity_and_conductivity|electrical resistivity]], because every one of those processes is easier or harder at a boundary than in the perfect lattice. In the microsim below the reader has one control, the mean grain size `d`, logarithmic from 10 nm to 1 mm. A seeded Voronoi map is rebuilt at each setting, so the grains visibly coarsen or refine, and the Hall–Petch line `σ_y = σ_0 + k_y/√d` moves a marker on the strength axis. The copper preset takes `σ_0` = 25 MPa and `k_y` = 0.11 MPa·m^0.5, with steel presets alongside, and the regime below about 10 nm where the trend reverses is drawn but marked ILLUSTRATIVE. On the [[Materials_science]] flagship this page serves Part II, Fundamentals › Structure, in the section *Microstructure*: it is the point on the spine where the subject stops being one crystal and becomes an aggregate of them, and where a [[Dislocation|dislocation]] first meets something it cannot glide through. ## High and low angle boundaries The natural first division is by misorientation angle. Below roughly 15° a boundary can be built entirely from dislocations: a symmetric tilt boundary of misorientation `θ` is a wall of edge dislocations of Burgers vector `b` stacked a distance `D = b/sin θ ≈ b/θ` apart, and a twist boundary is a crossed grid of screw dislocations.[^read-shockley1950][^hull-bacon-ch9] The dislocations in such a low-angle boundary are far enough apart that their cores do not overlap, so the lattice between them is nearly perfect and the boundary is, structurally, a defect of the crystal rather than a new kind of interface. For copper, with `b` = 0.256 nm, a 1° tilt boundary has its dislocations 14.7 nm apart and a 5° boundary 2.9 nm apart. Above about 15° the spacing `b/θ` falls below the core diameter, the cores merge, and the description in terms of separate dislocations stops meaning anything. High-angle boundaries are treated instead as their own two-dimensional structures, disordered over one or two atomic layers, whose atoms belong fully to neither lattice. The distinction is not cosmetic: the two classes differ by roughly an order of magnitude in mobility and by a factor of several in energy, and only high-angle boundaries act as the strong obstacles that the microsim's Hall–Petch line assumes. The exception is the coherent twin, a boundary of large misorientation whose atoms nevertheless sit on sites shared by both lattices. Twins have energies an order of magnitude below an ordinary high-angle boundary and are correspondingly hard to move, which is why annealed [[Copper|copper]] is full of straight twin bands. ## Describing a boundary A grain boundary needs more numbers than a dislocation does. Specifying the orientation of grain B relative to grain A takes three parameters — an axis and a rotation angle — and specifying which plane the boundary cuts through that relationship takes two more, the direction of the boundary normal. These five macroscopic degrees of freedom mean that "a 38° boundary in copper" is not one object but a whole family, and this is the main reason grain-boundary properties are still tabulated sparsely. Three further microscopic parameters, the rigid translation of one lattice relative to the other, are fixed by relaxation rather than chosen. For special misorientations the two lattices, extended through one another, share a fraction of their sites. The coincidence-site lattice describes such a boundary by `Σ`, the reciprocal of that shared fraction: `Σ3` means one site in three is common to both lattices, and in a face-centred cubic metal `Σ3` is the 60°⟨111⟩ relationship of the coherent twin.[^kronberg1949] Low-`Σ` boundaries tend to have low energy and low mobility, and the notation is the standard shorthand, but the correspondence is loose — a `Σ3` misorientation on the wrong plane is an incoherent boundary with ordinary energy. Modern work therefore reports the full five-parameter distribution, measured grain by grain from orientation maps. ## Boundary energy A boundary costs energy because its atoms are in the wrong positions: bonds are stretched, compressed or missing, and the excess free energy per unit area, `γ_gb`, is what the material would recover by removing the boundary. High-angle boundaries in metals run near a third of the free-surface energy, a few tenths of a joule per square metre, which is also why a grain boundary that meets a free surface makes a groove at the angle where the three tensions balance. Low-angle boundaries have the energy of their dislocation content, and because the dislocations are individually cheap but get more numerous as `θ` grows, the sum is not linear. Read and Shockley derived `γ(θ) = γ_0·θ·(A − ln θ)`, which rises steeply from zero, flattens as the cores begin to interact, and matches measurements up to about 15° before the model loses its basis.[^read-shockley1950] Above that the energy is nearly independent of angle except at the special low-`Σ` misorientations, where it dips into cusps; the coherent twin sits in the deepest of these. Energy per unit area matters because it is multiplied by an area that grows fast as grains shrink. Modelling grains as cubes of edge `d`, each face shared between two grains, gives a boundary area per unit volume of `3/d`. At `d` = 100 µm that is 3 × 10⁴ m² of boundary in every cubic metre; at the sim's finest setting, `d` = 10 nm, it is 3 × 10⁸ m²/m³, or about 300 m² of interface in a cubic centimetre of metal. Stored boundary energy is what drives everything in the next section. ## Excess volume Because the atoms at a boundary cannot pack as efficiently as those in the lattice, a boundary carries free volume: the material is slightly less dense there. The quantity is small — a fraction of an atomic volume per unit boundary area, giving a boundary expansion of a few per cent of an atomic spacing — but it is the structural reason for several properties at once. The loose packing is what lets atoms move along a boundary far faster than through the lattice, what makes boundaries preferred sites for segregated solutes and for the nucleation of second phases, and what makes them absorb the [[Vacancy_defect|vacancies]] and interstitials produced by deformation or irradiation. Excess volume also depends on misorientation in the same way energy does, falling into cusps at the coincidence misorientations. A coherent twin, whose atoms are on lattice sites, has essentially no excess volume, which is why twins neither speed diffusion nor collect impurities. ## Boundary migration A boundary moves when there is a free-energy difference across it, and to a good approximation its velocity is proportional to the driving pressure: `v = M·P`, where `M` is the mobility. In an annealed metal with no stored strain the only driving force is the boundary's own curvature, `P = 2·γ_gb/R` for a boundary of radius `R`, and it always points toward the centre of curvature — so small grains, which are convex, shrink and disappear while large ones grow. Integrating that over a structure gives the parabolic [[Grain_growth|grain growth]] law `d² − d_0² = k·t`, the variant sim on the *Grain growth* page, with `k` carrying an [[Arrhenius_equation|Arrhenius]] temperature dependence through the mobility.[^burke-turnbull1952] Mobility is extraordinarily sensitive to chemistry. A few parts per million of a solute that segregates to the boundary drags on it, and the drop in mobility between a nominally pure and a slightly impure metal can be orders of magnitude; second-phase particles pin boundaries outright, which is how a fine grain size is kept through a hot process. When pinning fails locally, a few grains grow far beyond the rest, the phenomenon of [[Abnormal_grain_growth|abnormal grain growth]]. The geological version of the same physics is the clearest illustration of what the sim's control means. Magma that cools slowly at depth nucleates few crystals and grows them large, producing the coarse phaneritic texture of granite in which the grains are visible to the eye; the same melt erupted and chilled at the surface nucleates many crystals and grows none of them far, giving the fine aphanitic texture of basalt, and quenched faster still it forms glass with no grains at all.[^earle-ch3] Grain size is a record of thermal history, in a lava flow and in a weld alike. ## Complexion A grain boundary can exist in more than one structural state at the same misorientation, and switch between them. Because the boundary has its own local composition and its own degree of disorder, it behaves as a quasi-two-dimensional phase, and a change of temperature or of bulk solute content can drive a discontinuous transition from one state to another — a clean boundary to one with an adsorbed monolayer, an ordered structure to a nanometre-thick disordered film. Such states are called complexions rather than phases, because they cannot exist without the abutting crystals.[^dillon2007] The practical importance is that mobility and energy jump at the transition. A boundary that suddenly acquires a disordered film can become orders of magnitude more mobile, which is one explanation for abnormal grain growth appearing in a narrow window of temperature or dopant level, and for the sharp effect small additives have on [[Sintering|sintering]].[^dillon2007] ## Effect to the electronic structure To a conduction electron a grain boundary is a plane of scattering centres. In a coarse-grained metal the grains are far larger than the electron mean free path — about 40 nm in copper at room temperature — so boundary scattering adds almost nothing and the [[Electrical_resistivity_and_conductivity|resistivity]] is set by phonons and point defects. As `d` approaches and then falls below the mean free path, boundary scattering takes over and resistivity rises steeply; Mayadas and Shatzkes wrote the standard model, treating boundaries as partially reflecting planes characterised by one reflection coefficient.[^mayadas1970] This is not an academic point. Copper interconnects in [[Semiconductor_device_fabrication|integrated circuits]] now have cross-sections of a few tens of nanometres, so their grains are necessarily that small and their resistivity is several times the bulk value, which is one of the limits on further scaling. In [[Semiconductor|semiconductors]] the effect is stronger still, because a boundary carries electronic states in the [[Band_gap|band gap]] that trap carriers and build a potential barrier; the efficiency gap between single-crystal and multicrystalline silicon solar cells is largely this. ## Defect concentration near grain boundaries The loosely packed boundary is a sink, a source and a highway for point defects. It absorbs excess [[Vacancy_defect|vacancies]] and interstitials, which is why a fine-grained metal is more tolerant of irradiation damage, and it emits vacancies where they are needed, which is what allows the diffusional creep of fine-grained solids under load at high temperature. Diffusion along a boundary is faster than through the lattice because the activation energy is lower — roughly half the lattice value in face-centred cubic metals — so the two paths trade places as the temperature falls. Above about half the melting temperature the lattice route dominates because its far greater cross-sectional area outweighs its higher barrier; below it, boundary diffusion carries most of the transport, and the crossover moves to higher temperature as the grains get finer. Solutes also concentrate at boundaries. Because an atom that does not fit the lattice fits the disordered boundary better, it lowers the system's energy by [[Segregation_(materials_science)|segregating]] there, and equilibrium boundary concentrations of hundreds of times the bulk level are ordinary for impurities like sulfur or phosphorus in steel. Segregation is the mechanism behind temper embrittlement and a good deal of intergranular [[Corrosion|corrosion]], and it is also the deliberate tool behind the solute-drag control of mobility described above. ## Relationship between theory and experiment The sim's readout is the oldest useful empirical statement about grain boundaries, and the one whose theory is least settled. In 1951 Eric Hall, measuring mild [[Steel|steel]], and in 1953 Norman Petch, measuring cleavage strength, found strength varying with the inverse square root of grain size: `σ_y = σ_0 + k_y/√d`, with `σ_0` the friction stress of a single crystal.[^hall1951][^petch1953] The classical explanation is a pile-up. Dislocations gliding on a [[Slip_(materials_science)|slip]] plane stop at the boundary, because the slip systems of the next grain are misoriented, and pile up behind the leader; the stress at the head of a pile-up of `n` dislocations is `n` times the applied shear, so a longer pile-up — a larger grain — triggers slip next door at a lower applied stress. Since the number a pile-up holds scales with grain diameter, the critical stress scales as `1/√d`.[^hull-bacon-ch9] Putting the copper preset through the equation shows what is at stake. With `σ_0` = 25 MPa and `k_y` = 0.11 MPa·m^0.5, a 1 mm grain size gives 28 MPa, 100 µm gives 36 MPa, 10 µm gives 60 MPa, 1 µm gives 135 MPa, 100 nm gives 373 MPa and 10 nm gives 1.1 GPa.[^presets-cn] Refining the grain is the only classical mechanism that raises strength without spending [[Ductility|ductility]] — unlike [[Work_hardening|work hardening]], [[Solid_solution_strengthening|solid-solution]] or [[Precipitation_hardening|precipitation]] strengthening — so metallurgists reach for it first. The readout is a yield stress in the [[Stress–strain_curve|stress–strain curve]] sense: a 0.2 % offset, not a sharp event.[^univphys-ch12] The last decade of the slider is where theory and experiment part company. Metals with [[Nanoparticle|nanoscale]] grains leave the line somewhere between 10 and 30 nm, peak and then soften as the grains are refined further — the inverse Hall–Petch regime, attributed to deformation moving out of the grains into the boundaries themselves, by sliding and rotation, once there is no room inside a grain for a pile-up.[^schiotz1998] The crossover is where this page's earlier geometry bites: with a boundary thickness of about 0.5 nm, the fraction of atoms lying in boundaries is `3·δ/d` — 0.0015 % at `d` = 100 µm but 15 % at 10 nm. A material one-seventh interface is no longer a crystal with boundaries in it. The sim draws the reversal because the physics is real, and marks it ILLUSTRATIVE because its position and depth are a sketch, not a measured law. ## See also - [[Grain_growth]] — the variant sim, `d² − d_0² = k·t` under an anneal-time control - [[Crystallite]] - [[Microstructure]] - [[Abnormal_grain_growth]] - [[Dislocation]] — the pile-up behind the Hall–Petch exponent - [[Recrystallization_(metallurgy)]] - [[Segregation_(materials_science)]] - [[Work_hardening]] ## References [^hall1951]: Hall, E. O. (1951). "The Deformation and Ageing of Mild Steel: III Discussion of Results." *Proceedings of the Physical Society. Section B* 64 (9): 747–753. https://doi.org/10.1088/0370-1301/64/9/303 [^petch1953]: Petch, N. J. (1953). "The Cleavage Strength of Polycrystals." *Journal of the Iron and Steel Institute* 174: 25–28 (volume and pages to pin; no DOI asserted). [^read-shockley1950]: Read, W. T.; Shockley, W. (1950). "Dislocation Models of Crystal Grain Boundaries." *Physical Review* 78 (3): 275–289 (DOI to pin). [^kronberg1949]: Kronberg, M. L.; Wilson, F. H. (1949). "Secondary Recrystallization in Copper." *Transactions of the American Institute of Mining and Metallurgical Engineers* 185: 501–514 (the coincidence-site lattice; pages to pin). [^burke-turnbull1952]: Burke, J. E.; Turnbull, D. (1952). "Recrystallization and Grain Growth." *Progress in Metal Physics* 3: 220–292 (the curvature driving force and the parabolic growth law; pages to pin). [^dillon2007]: Dillon, S. J.; Tang, M.; Carter, W. C.; Harmer, M. P. (2007). "Complexion: A new concept for kinetic engineering in materials science." *Acta Materialia* 55 (18): 6208–6218 (DOI to pin). [^mayadas1970]: Mayadas, A. F.; Shatzkes, M. (1970). "Electrical-Resistivity Model for Polycrystalline Films: the Case of Arbitrary Reflection at External Surfaces." *Physical Review B* 1 (4): 1382–1389 (DOI to pin). [^schiotz1998]: Schiøtz, J.; Di Tolla, F. D.; Jacobsen, K. W. (1998). "Softening of nanocrystalline metals at very small grain sizes." *Nature* 391 (6667): 561–563 (DOI to pin). [^hull-bacon-ch9]: Hull, D.; Bacon, D. J. *Introduction to Dislocations*, 5th ed. (2011), Butterworth-Heinemann, Ch. 9 Dislocation Arrays and Crystal Boundaries (tilt and twist walls, `D = b/sin θ`, pile-ups against a boundary and the `1/√d` argument) (page to pin). Not a Portal Book. [^earle-ch3]: Earle, Steven. *Physical Geology* (2015), BCcampus Open Education, Ch. 3 Intrusive Igneous Rocks, pp. 67–92 (cooling rate against crystal size: phaneritic versus aphanitic versus glassy textures) (page to pin). https://open.umn.edu/opentextbooks/textbooks/physical-geology [^univphys-ch12]: Sanny, J.; Ling, S. et al. *University Physics Volume 1* (2016), OpenStax, Ch. 12 Static Equilibrium and Elasticity, pp. 565–610 (stress, strain and the elastic moduli that fix what a yield stress means) (page to pin). https://open.umn.edu/opentextbooks/textbooks/university-physics-volume-1 [^presets-cn]: *Citation needed.* The copper preset `σ_0` = 25 MPa and `k_y` = 0.11 MPa·m^0.5, and the steel presets, are supplied by the M9 sim row; the six yield stresses quoted in *Relationship between theory and experiment* are computed here from them. A tabulated Hall–Petch fit for annealed high-purity copper, with its grain-size range and its source, would let the presets be checked, and would fix where the fitted line should be cut off at the fine end. ## Further reading - Earle, Steven. *Physical Geology* (2015), BCcampus — Ch. 3, for cooling rate against crystal size, the geological twin of the sim's control. - Sanny, J.; Ling, S. et al. *University Physics Volume 1* (2016), OpenStax — Ch. 12, for the stress and strain definitions the Hall–Petch readout uses. - Hull, D.; Bacon, D. J. *Introduction to Dislocations*, 5th ed. (2011) — Ch. 9, for tilt walls and pile-ups. <!-- MATTERSIM:BEGIN g33 — Matter & Energy Cluster microsim (framework build, specs/sims/Grain_boundary.json); do not hand-edit inside --> **Microsim — three.js (Wikitube framework):** *Grain boundary* <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/matter/Grain_boundary.html" data-title="Grain boundary"></div> *Built from `MICROSIM_GUIDE/specs/sims/Grain_boundary.json`; part of the [[PORTAL_Matter|Matter portal]] spine (section sims and See-also variants).* <!-- MATTERSIM:END --> ## Wikipedia : Wikitube **Strict pair:** [Wikipedia](https://en.wikipedia.org/wiki/Grain_boundary) : [Wikitube](https://en.wikitube.io/wiki/Grain_boundary) · pinned revision [1369639347](https://en.wikipedia.org/w/index.php?oldid=1369639347) · 2026-09-11 ## Previous hub tags Hubs: `Life_Physics`. Portals: [[PORTAL_Materials_science]]. --- *Matter & Energy Cluster child articles, wave 1 · 2026-09-11 · drafted · Materials_science row M9 · sim pending (matter/Grain_boundary).*