Two ways to rebuild the same crystal
In guide 1 you learned that a polymorphic transition takes one fixed set of atoms and re-houses them in a different structure — iron's BCC alpha becoming FCC gamma, graphite becoming diamond, white tin becoming grey. That guide told you WHAT changes. This guide asks the sharper question: by what physical route do the atoms actually get from the old arrangement to the new one? Nature has, broadly, only two answers, and almost every structural transformation you will ever meet is one, the other, or a blend of the two.
Picture a brick wall you want to change into a differently-shaped wall. One way is to dismantle it brick by brick, carry the bricks around, and re-lay them in the new pattern — slow, laborious, and every brick ends up wherever it happens to be carried. The other way is to lean on the whole wall and shear it sideways, so it tilts into a new shape while every brick keeps touching exactly the same neighbours it always had — fast, and the new wall remembers precisely how the old one was oriented. The first route is a reconstructive transformation; the second is a displacive transformation. That single difference — do the atoms let go of their neighbours, or hold on — cascades into almost everything else about how the change behaves.
Reconstructive: break every bond and start over
A reconstructive transformation is the dismantle-and-re-lay route. To turn the old structure into a genuinely different one — different coordination, different packing — atoms must break their strong bonds, wander off by diffusion, and re-settle in the new pattern. Because breaking bonds costs a great deal of energy, there is a large barrier to get over, so the new phase has to be born as tiny nuclei that then grow, and the whole thing is slow and strongly thermally activated: heat it up and it speeds up, cool it down and it can stall for geological ages. Crucially, since each atom finds its new home more or less independently, the daughter crystal usually forgets how the parent was oriented — there is no built-in relationship between the two.
Carbon gives the most famous example. At room temperature and pressure, graphite is the truly stable form of carbon, and diamond is only metastable — thermodynamics says a diamond ought to turn into graphite. It does not, because the conversion is reconstructive: every one of diamond's strong covalent bonds would have to break and re-form into graphite's sheets, an enormous kinetic barrier that at room temperature is essentially never crossed. So 'diamonds are forever' is a statement about kinetics, not stability. Running it the other way — graphite to diamond — takes brute force: industrial synthesis uses about 5 to 6 GPa of pressure and 1400 to 1600 degrees C, usually with a molten metal catalyst to help the bonds re-sort.
Tin shows the same mechanism at its most destructive. Above 13.2 degrees C the stable form is white beta-tin, a shiny metallic tetragonal crystal; below it, the stable form is grey alpha-tin, which has the open diamond-cubic structure and is a brittle semiconductor. The switch is reconstructive and, like all reconstructive changes, sluggish to start — but the diamond-cubic packing is far less dense, so the transformed tin swells by about 27 percent (density drops from 7.27 to 5.77 g/cm^3) and simply crumbles to grey powder. Worse, it is autocatalytic: a speck of grey tin seeds its neighbours, so the rot spreads. This is tin pest, blamed over the centuries for organ pipes disintegrating in cold churches (the tale of Napoleon's soldiers' buttons is a colourful legend, almost certainly untrue, but it captures the idea).
Displacive: nudge, do not break
A displacive transformation is the lean-and-shear route, and it is a completely different animal. No strong bonds break; instead every atom shifts by a tiny amount — typically less than a fifth of an interatomic spacing, always less than one full bond length — and it shifts cooperatively, in lockstep with all its neighbours, so no atom ever swaps who it is bonded to. Because nothing has to break and no atom has to diffuse anywhere, the change can sweep through a crystal at close to the speed of sound (martensite fronts have been clocked at up to about 1000 m/s). And because there is no diffusion, the amount that transforms depends on how cold you make it, not on how long you wait — it is athermal, not time-dependent.
Holding on to your neighbours has vivid consequences. Because the whole lattice shears as one, the daughter crystal is born locked to the parent in a fixed orientation relationship — you can predict exactly which parent plane and direction line up with which daughter plane and direction. The shear also tilts the transformed region bodily, so a flat polished surface visibly buckles into relief, and the transformed plate lies along a specific, near-undistorted plane called the habit plane. Finally, that shape change is so violent it must be accommodated: the plate often splits into fine alternating bands, transformation twins, which cancel out the average distortion the way pleats let a stiff skirt still hang straight. The archetype of all this is the martensitic transformation in quenched steel, and it earns the whole of the next guide.
One more gift of the displacive route: because the lattice merely leaned over, it can lean back. Cool a shape-memory alloy and it transforms displacively; warm it and it un-transforms, snapping back to the parent shape — the trick behind self-bending eyeglass frames and heart stents. There is always a temperature gap, a hysteresis, between the forward and reverse changes, because starting the shear still costs a little energy in each direction. And because the shift is small and cooperative, the daughter's symmetry is almost always a subgroup of the parent's — a gentle symmetry-lowering rather than a total rebuild, which is exactly the thread guide 5 will pick up.
Iron at the crossroads: one pair, two routes
The most beautiful thing about this dichotomy is that the SAME parent-daughter pair can go either way, depending only on how you drive it. Take iron. Above 912 degrees C it is FCC gamma (austenite); below, it is BCC alpha (ferrite). Recall the packing numbers from earlier rungs: BCC has packing factor 0.68 and coordination 8, while FCC packs at 0.74 with coordination 12. So the FCC form is actually the denser one — which is why iron does the counter-intuitive thing and shrinks by about 1 percent when you HEAT it through 912 degrees, a genuine anomaly among metals.
Now cool austenite back down and watch the fork in the road. Cool it slowly and carbon has time to diffuse: the transformation runs reconstructively, atoms re-sort, and you get soft ferrite plus layered pearlite, a classic diffusion product with no memory of the parent orientation. But quench it fast into cold water and diffusion is frozen out — the carbon is trapped with nowhere to go. Iron cannot wait, so it takes the displacive escape route instead: the FCC lattice shears bodily into a strained body-centred-tetragonal cell in a diffusionless instant, giving hard, brittle martensite. Same two structures, same two atoms; slow means reconstructive and soft, fast means displacive and hard. The entire art of heat-treating steel lives in that choice.
RECONSTRUCTIVE vs DISPLACIVE
(break & re-sort) (nudge & shear)
----------------------------------------------------------------
nearest bonds BROKEN KEPT (only bent / stretched)
atom movement long-range DIFFUSION < 1 bond length, cooperative
speed slow (needs jumps) ~ speed of sound
temperature thermally activated athermal (amount ~ how cold)
depends on time? YES (wait -> more) NO (cool more -> more)
orientation usually FORGOTTEN parent<->daughter LOCKED (OR)
shape change none (just regrows) macroscopic shear + surface relief
energy barrier large (bonds break) small
reverse? sluggish, may stall clean, with hysteresis (shape memory)
symmetry link need NOT be group-sub usually group -> subgroup
----------------------------------------------------------------
examples graphite <-> diamond austenite -> martensite (quenched)
white <-> grey tin BaTiO3 cubic -> tetragonal
slow gamma -> alpha fast gamma -> alpha' (same Fe pair!)A spectrum, and some honest caveats
Do not treat reconstructive and displacive as a hard binary — they are the two ends of a spectrum. Real transformations can sit in between: a small displacive shuffle may need a little short-range hopping to complete, or a mostly-reconstructive change may keep a partial orientation relationship. And this mechanism axis is not the only way to classify a transformation. Guide 4's order-disorder changes and guide 5's continuous, second-order transitions cut the cake along a different line entirely — the thermodynamic order of the transition (does a property jump, or slide smoothly to zero?), which is a separate question from whether bonds break. Two different filing systems for the same drawer of phenomena.
- Is it fast or slow, and does time matter? Sweeping through near-instantly and caring only about how cold you go points to displacive; needing to hold at temperature and getting further the longer you wait points to reconstructive.
- Is there a fixed orientation relationship between parent and daughter? A locked, predictable OR is the fingerprint of a displacive (cooperative) shear; a forgetful, near-random daughter says reconstructive.
- Does a polished surface tilt into visible relief? Surface upheaval means a macroscopic shape change, hence a shear, hence displacive. No relief, just new grains, means reconstructive.
- Does the coordination number change a lot? A big jump in nearest-neighbour count (like 8 to 12, or metallic tin to diamond-cubic tin) forces bonds to break — reconstructive. Little change in coordination allows a gentle displacive shuffle.