Rusting without water: a battery with no puddle
The first three guides in this rung all needed a wet electrolyte: metallic corrosion ran as an electrochemical cell with an anode dissolving, a cathode consuming oxygen, and salty water carrying the ions between them. Now take the water away and turn up the heat — a turbine blade at 1000 degrees C, an exhaust manifold glowing dull red, a heating element in a furnace. Bare metal meets hot air and grows a crust of oxide called scale. This is high-temperature oxidation, and it is corrosion's dry cousin.
The surprise is that it is still a battery, still the same anode-and-cathode chemistry — the roles have simply moved. At the metal-scale interface the metal gives up electrons (M becomes M^2+ plus 2 electrons): that is the anode, the oxidation half-reaction. At the scale-air interface oxygen grabs those electrons (O2 plus 4 electrons becomes 2 O^2-): that is the cathode, the reduction half-reaction. What is astonishing is what plays the part of the salty water. There is no puddle — the solid oxide scale itself is the electrolyte, ferrying metal ions and oxygen ions, and also the wire, carrying the electrons across. The scale is the entire electrochemical cell squeezed into a film a few micrometres thick.
How the scale grows: the parabolic clock
Picture the scale as a wall being built from both sides at once. To add a new layer of oxide, either metal ions must trek outward through the existing scale to reach the air, or oxygen ions must trek inward to reach the metal. Either way the journey is a slow shuffle of atoms through a solid — it is diffusion, the same random-walk hopping through the crystal that earlier rungs used to case-harden steel. And here is the beautiful consequence: the thicker the wall already is, the longer that trek, so the slower the next layer forms. The scale smothers its own growth.
Turn that into arithmetic. When diffusion through the scale sets the pace, the thickness x obeys the parabolic rate law: x^2 = k_p times t, so x = sqrt(k_p times t). Thickness grows as the square root of time — the classic signature of a self-limiting, protective scale. A quick worked example makes it vivid. Suppose a scale reaches 10 micrometres after 1 hour. After 4 hours it is not 40 but 10 times sqrt(4) = 20 micrometres; after 100 hours only 10 times sqrt(100) = 100 micrometres. To double the scale you must wait four times as long, and the metal loss per year keeps easing off. That square-root crawl is exactly the behaviour you want at red heat.
But not every metal is so lucky, and the shape of the growth curve tells you which kind of scale you have. If the scale is porous or cracked, oxygen reaches fresh metal without any real barrier, so the rate never eases: thickness grows straight-line with time, x = k_l times t, the linear law — runaway loss, the mark of a useless scale. At the other extreme, very thin films at lower temperatures grow by a logarithmic law, x proportional to log(t), slamming almost to a halt after a few nanometres; that is the regime of the ultra-thin protective films of the previous guide. Parabolic is the workhorse middle case of a thick protective scale.
THREE OXIDATION KINETICS (scale thickness x vs time t) x | . linear x = k_l * t | . ' (porous scale: | . ' NO protection, | . ' runaway loss) | . ' | .' _____...---- parabolic x = sqrt(k_p * t) | .' __..---'' (dense scale, diffusion- | .' _.-'' limited: self-slowing) | ._-' ________________________ logarithmic x ~ log(t) | .-' __.-' (ultra-thin film: nearly | _:.--' stops -> best protection) |_//___________________________________ t parabolic & logarithmic = protective | linear = non-protective
The Pilling-Bedworth ratio: does the oxide fit?
Whether a scale can even become dense enough to protect turns on a wonderfully simple question: when a chunk of metal turns into oxide, does the oxide take up more room than the metal it ate, or less? The answer is the Pilling-Bedworth ratio (PBR) — the volume of oxide produced divided by the volume of metal consumed. In symbols, PBR = (M_oxide times rho_metal) divided by (n times M_metal times rho_oxide), where M is molar mass, rho is density, and n is the number of metal atoms in one oxide formula unit (n = 2 for Al2O3). It is just a bookkeeping of volumes, yet it foretells the whole fate of the scale.
Think of it as fitting a lid onto the metal. If PBR is less than 1 the oxide is too small — it cannot cover the surface it grew from, so it forms a loose, porous, gappy layer that shields nothing; oxygen pours straight through and the metal oxidises linearly forever. This is the fate of magnesium and the alkali metals. If PBR is between about 1 and 2 the oxide fits snugly, forming a dense continuous film under mild compression that seals the surface: protective, parabolic, exactly what you want. And if PBR is much greater than 2 the oxide is far too big for the space — it is crammed in under enormous compressive stress, so it buckles, cracks, and flakes off, like too many coats of paint crazing on a windowsill. That exposes bare metal and the ruin starts over.
Two quick numbers show the rule at work. For aluminium forming Al2O3: PBR = (101.96 times 2.70) / (2 times 26.98 times 3.95) = 275 / 213 = 1.29 — comfortably in the protective band, which is exactly why aluminium is famous for its clingy oxide. For magnesium forming MgO: PBR = (40.3 times 1.74) / (1 times 24.31 times 3.58) = 70 / 87 = 0.81 — below 1, porous and non-protective, which is why magnesium burns so fiercely once lit. One ratio, and you already know which metal shields itself and which does not.
PILLING-BEDWORTH RATIO PBR = (M_ox * rho_metal) / (n * M_metal * rho_ox)
PBR < 1 oxide too SMALL -> porous, gappy NON-protective
Mg 0.81 Li 0.57 K 0.45 (linear kinetics)
1 < PBR < 2 oxide FITS -> dense, adherent PROTECTIVE
Al 1.29 Ni 1.70 Fe ~1.8 Cr 2.0
scale grows as x = sqrt(k_p * t) (parabolic)
PBR > 2 oxide too BIG -> compressive often cracks / spalls
Nb 2.7 Ta 2.5 Mo 3.3 W 3.4 (breakaway loss)
n = number of metal atoms per oxide formula unit (Al2O3 -> n = 2)The heroes of red heat: chromium, aluminium, and adhesion
A handful of elements form nearly ideal scales, and they are the reason high-temperature engineering is possible at all: chromium (Cr2O3, chromia), aluminium (Al2O3, alumina), and silicon (SiO2, silica). Each grows a dense, slow, tightly-bonded oxide that starves its own reaction. This is the very same trick as the passivation of the previous guide — a thin protective oxide standing guard — only now the guard holds in dry gas at red heat instead of in wet acid. It is no coincidence that stainless steel survives a furnace and a saltwater splash for the same reason: its chromium raises a protective oxide in both worlds. Add enough chromium and aluminium and you build a superalloy — the nickel-based superalloys that let a jet engine run hotter than the melting point of the metal beneath their scale.
But forming a good oxide is only half the battle; the scale must also stay put. Here a quieter villain appears: the oxide and the metal expand by different amounts when heated, a thermal expansion mismatch. Hold the part at temperature and all is well, but every time it cools the two materials shrink at different rates, the scale is stressed, and on enough thermal cycles it cracks and spalls off — flaking away to bare the metal so oxidation restarts. This 'breakaway' loss is why a part can pass a steady furnace test yet fail in real service that switches on and off. Engineers fight it by adding tiny amounts of 'reactive elements' like yttrium or hafnium, which pin the scale to the metal and dramatically improve its adhesion.
When even the best alloy oxide is not enough — inside the very hottest section of a modern jet engine — engineers stop asking the metal to survive the heat and instead insulate it. A thermal barrier coating, typically a porous layer of yttria-stabilised zirconia sprayed a fraction of a millimetre thick, is a ceramic blanket that can drop the metal's temperature by a hundred degrees or more, with a metallic bond-coat beneath it that grows the protective alumina scale. Blade, bond-coat, and ceramic together are why turbine gas can run far above the alloy's own limit.
- Name the service: the peak temperature, the gas (air, combustion products, steam, sulphur-bearing), and — crucially — whether it cycles hot-and-cold or holds steady.
- Pick an element that raises a slow, dense oxide: chromium, aluminium, or silicon — check that its Pilling-Bedworth ratio sits in the protective 1-to-2 band.
- Confirm parabolic (self-slowing) kinetics, not linear, at the working temperature — a straight-line weight-gain curve is a red flag for a porous scale.
- If the part thermal-cycles, guard adhesion: add reactive elements (Y, Hf), match expansion, and consider a bond-coat plus thermal barrier coating for the hottest zones.
Honest limits: a rule of thumb, not a law of nature
The Pilling-Bedworth ratio is a lovely first sieve, but do not mistake it for a verdict. It gets the extremes right — PBR well below 1 really is doomed, PBR above 3 really does tend to spall — yet the middle band is full of honest exceptions. A ratio inside 1 to 2 is necessary for a dense scale but not sufficient: iron sits near 1.8 and still oxidises briskly, because its scale is a stack of three different oxides (wustite, magnetite, haematite) laced with fast diffusion paths and pores. The ratio counts only volume; it says nothing about how fast ions diffuse, how well the oxide grips, or whether the scale is one clean phase or a leaky sandwich. Two of those — diffusion rate and adhesion — often matter more than PBR itself.
One last connection ties this rung to the last. High temperature attacks a metal on two fronts at once: the surface oxidises while the interior slowly stretches under load — creep, from the failure rung. The two conspire, because oxidation notches and pits the surface and those notches become the cracks that creep and fatigue then drive inward. This is why the same nickel superalloys engineered for creep are also engineered for oxidation, and why the very high-temperature oxidation you have just met is inseparable from how hot-section parts are chosen. The last guide of this rung pulls all of protection together — coatings, cathodic protection, inhibitors — and then turns to how the non-metals rot: polymers that swell, chain-scission, and weather in the sun, and ceramics that slowly dissolve.