Spotlight #67: Materials Science — Dislocations, Phase Diagrams & Grain Growth

I. Edge Dislocation Glide — Why Metals Are Weak (and How We Fix That)

Here is a puzzle that baffled physicists for decades: the theoretical shear strength of a perfect metal crystal — the stress needed to slide one half of the crystal over the other simultaneously — is about G / 6 to G / 30, where G is the shear modulus. For copper, that works out to roughly 1–10 GPa. Yet real copper yields at just 50–100 MPa. That is two to three orders of magnitude weaker than theory predicts. The reason is dislocations.

An edge dislocation is a line defect in a crystal: an extra half-plane of atoms that terminates inside the lattice. Instead of sliding an entire plane of atoms simultaneously — the process requiring GPa of stress — a dislocation moves by breaking and re-forming one bond at a time along its line. The dislocation advances by one lattice spacing (one Burgers vector) per broken bond, sweeping across the slip plane like a ruck in a carpet. The stress needed to move a dislocation is the Peierls–Nabarro stress:

tau_PN = (2G / (1 - nu)) * exp(-2*pi*d / (b * (1 - nu)))

  G   : shear modulus (GPa)
  nu  : Poisson's ratio (~0.33 for most metals)
  d   : spacing between adjacent slip planes (m)
  b   : Burgers vector magnitude (lattice spacing, ~0.25 nm for Cu)

For Cu (FCC): tau_PN ~ 10^-4 * G  ~ 4 MPa  (slip on {111} planes)
For Si (DC):  tau_PN ~ 10^-2 * G  ~ 2 GPa  (explains silicon's brittleness)

The exponential dependence on d/b is the key insight. FCC metals like copper, aluminium, and gold have close-packed {111} planes with large spacing and small Burgers vectors, giving very low Peierls stress and excellent ductility. Diamond-cubic silicon has small plane spacing and large Burgers vectors, so dislocations are nearly immobile at room temperature — hence silicon shatters rather than bends.

Strengthening mechanisms

If dislocations are what make metals weak, then blocking dislocation motion is how we make them strong again. Every strengthening mechanism in structural metallurgy works by impeding dislocation glide:

• Work hardening — dislocations pile up and tangle, blocking each other. Cold-rolling a metal doubles or triples its yield stress.
• Solid-solution hardening — solute atoms distort the lattice locally, creating stress fields that pin dislocations. This is why stainless steel (Fe + Cr + Ni) is far stronger than pure iron.
• Precipitation hardening — fine particles of a second phase (e.g., Al&sub3;Cu in duralumin) act as obstacles that dislocations must either cut through or bow around (the Orowan mechanism).
• Grain boundary hardening — grain boundaries block dislocation motion entirely, described by the Hall–Petch relation: sigma_y = sigma_0 + k / sqrt(d), where d is grain diameter.

II. Binary Alloy Phase Diagrams — Reading the Recipe for a Microstructure

A phase diagram is a map. For a binary alloy (two components, say tin and lead), it tells you which phases — liquid, solid solution alpha, solid solution beta, intermetallic — are stable at every combination of composition and temperature. It is the single most important diagram in metallurgy: every casting process, heat treatment, and soldering specification relies on it.

The most iconic feature is the eutectic point: the single composition at which the alloy has the lowest possible melting point, and where the liquid freezes directly into two intimately mixed solids without passing through a two-phase mushy zone. In the Sn–Pb system, the eutectic is at 61.9 wt% Sn and 183 °C — the basis of traditional 60/40 electronics solder, chosen precisely because it solidifies sharply (no pasty range) and at the lowest possible temperature.

Eutectic reaction:
  L  -->  alpha + beta     (at T_eutectic, single composition)

For Sn-Pb:
  T_eutectic = 183 °C,  C_eutectic = 61.9 wt% Sn
  alpha phase: Pb-rich solid solution (max 19.2 wt% Sn at eutectic T)
  beta  phase: Sn-rich solid solution (max 2.5 wt% Pb at eutectic T)

Lever Rule (in a two-phase alpha + L region):
  At overall composition C_0, temperature T:
    x_alpha = (C_L - C_0) / (C_L - C_alpha)    [fraction solid]
    x_L     = (C_0 - C_alpha) / (C_L - C_alpha) [fraction liquid]

  C_alpha : composition of alpha phase (read from left boundary)
  C_L     : composition of liquid phase (read from right boundary)
  The "lever": C_0 is the fulcrum; lengths give phase fractions.

The lever rule is so named because the two-phase tie line acts like a lever: the fraction of each phase is inversely proportional to its distance from the overall composition, just as a heavier mass must sit closer to the fulcrum to balance a lighter one. It is a direct consequence of mass conservation and applies to any two-phase region in any equilibrium phase diagram.

Reading a cooling path

Consider a hypoeutectic Sn–Pb alloy at 40 wt% Sn cooled slowly from 300 °C. The cooling path crosses the liquidus at about 240 °C, where the first solid (Pb-rich alpha phase) nucleates. As temperature falls through the two-phase L + alpha region, the solid fraction grows and both phases change composition, following the liquidus and solidus boundaries. At 183 °C, the remaining liquid (now at exactly 61.9 wt% Sn) transforms eutectic: it freezes into a fine lamellar mixture of alpha and beta. The final microstructure is primary alpha dendrites surrounded by a eutectic matrix — exactly what you see in the cross-section of old plumbing solder.

The simulation lets you place a cursor anywhere on the phase diagram and see the lever rule computed live, with colour-coded phase fractions and compositions updated as you drag. The animated cooling mode shows nucleation events, lever-rule evolution, and the final frozen microstructure in a side-by-side panel.

III. Grain Growth — The Potts Model and Why Annealed Metals Go Soft

When a cold-worked metal is heated — annealed — three processes occur in sequence: recovery (point defects annihilate), recrystallisation (new strain-free grains nucleate and grow), and finally normal grain growth (large grains grow at the expense of small ones, driven purely by reduction of grain boundary area). It is this last process that our simulation models.

The driving force is thermodynamic. Grain boundaries are high-energy regions — atoms at a boundary are in a less favourable environment than atoms inside a grain, with excess energy of roughly 0.5–1 J m⁻² for high-angle boundaries in metals. The total grain boundary energy of a polycrystal is proportional to the total boundary area, which scales as 1/d (d = mean grain diameter). Coarsening reduces this energy by eliminating boundaries, growing the mean grain size over time.

The Potts model represents each lattice site as a spin variable Q ∈ {1, 2, …, Q_max}, where each value represents a distinct grain orientation. Sites with the same Q belong to the same grain; boundaries exist wherever neighbouring sites differ. At each Monte Carlo step, a lattice site is chosen at random, assigned a trial orientation from its neighbours, and the trial is accepted or rejected by a Metropolis criterion:

Hamiltonian:
  H = J * sum_{<i,j>} (1 - delta(Q_i, Q_j))
  J  : boundary energy per site-pair (J > 0 penalises unlike neighbours)
  delta : Kronecker delta (1 if same grain, 0 if different)

Metropolis acceptance rule:
  delta_H = H_new - H_old
  P_accept = 1            if delta_H <= 0  (favourable move)
  P_accept = exp(-delta_H / kT)  otherwise

Mean grain area growth law (parabolic law):
  <A>(t) = <A>(0) + 2*M*gamma*t
  or equivalently:  d(t)^2 - d(0)^2 = K*t
  K : rate constant (proportional to boundary mobility M and energy gamma)
  d(t) : mean grain diameter at time t
  exponent n = 0.5 in ideal (normal) grain growth

The parabolic growth law d ∝ t^0.5 is a mean-field result: it assumes all grains are identical spheres (or circles in 2D) and all boundaries move at the same speed. Real metals often show d ∝ t^n with n < 0.5, because solute drag, precipitate pinning (Zener pinning), or texture effects slow boundary motion selectively. The simulation measures n directly from the slope of a log–log plot of mean area versus Monte Carlo time.

Zener pinning

Adding a dispersion of fine, immobile particles to the simulation (inert sites that never change orientation) arrests grain growth: boundaries bow around particles and eventually pin. The Zener limit for the maximum grain diameter is d_Z = (4r) / (3f), where r is particle radius and f is volume fraction. This is why fine oxide particles are deliberately introduced into oxide-dispersion-strengthened (ODS) steels used in nuclear reactors: they freeze the grain structure at service temperature, preventing the strength loss that would otherwise follow from grain coarsening.

Try It Yourself

All three simulations run in your browser with no installation:

Closing Thought

Materials science is, at its heart, the discipline that connects atomic-scale behaviour to engineering performance. The three topics in this spotlight span seven orders of magnitude in length scale — from the 0.25 nm Burgers vector of a copper dislocation, through the micrometre-scale lamellae of eutectic solder, to the millimetre-scale grains of an annealed steel sheet — yet all three are governed by the same underlying principle: systems minimise free energy, and the rate at which they do so determines everything from yield strength to service life. The simulations here make that journey across scales tangible, letting you explore in minutes what took metallurgists decades to understand.