From guessed lattices to measured ones
All the way up this ladder you have taken crystal structure on faith. We said iron is body-centred cubic, aluminium is face-centred cubic with a packing factor of 0.74, that a unit cell of copper is about 0.36 nm on a side. But how does anyone actually know the atoms sit there, in that pattern, at that spacing? You cannot see an atom in a light microscope — guide 1 showed light tops out near half a micron, ten thousand times too coarse. Even the electron microscopes of guide 2, magnificent as they are, mostly give you a picture. To pin down the lattice as numbers — which structure, how big, how strained — the workhorse is not a microscope at all. It is X-ray diffraction, and it works by turning the crystal itself into a ruler.
The trick is a matter of scale matching. Ordinary light has a wavelength of 400 to 700 nm, thousands of times larger than the gap between atomic planes, so it strides right over the lattice without noticing it — like ocean swells rolling over a picket fence. X-rays are electromagnetic waves too, but with wavelengths near 0.1 nm, almost exactly the spacing between planes of atoms in a crystal. When a wave and the thing it meets are the same size, they interact strongly and interfere. So a crystal, with its planes of atoms stacked at just that spacing, behaves toward X-rays as a diffraction grating behaves toward visible light: it scatters the beam into a set of sharp, precisely-angled beams whose directions encode the geometry of the lattice.
Bragg's law: why crystals flash at just the right angle
Picture the X-ray beam striking not one plane of atoms but a whole deck of parallel planes, each a spacing d below the last. A little of the beam bounces off the top plane, a little more off the next plane down, and so on through hundreds of planes. Each reflected ray from a deeper plane has travelled a slightly longer path — down to that plane and back up. Waves that arrive back in step (crests on crests) add up into a strong beam; waves that arrive out of step cancel to nothing. So the crystal is silent at almost every angle, and then, at a few special angles, it flashes brightly. The condition for a flash is the single most important equation in this guide, Bragg's law.
BRAGG'S LAW : n(lambda) = 2 d sin(theta) (constructive interference)
incoming rays scattered rays
\ \ / /
\ theta \ / theta /
--------o---------o------------------o--------o-------- plane 1
\ \ / /
\ \ extra path / /
\ \ = 2 d sin(t)/ / d = spacing between planes
------------o---------o------------o--------o---------- plane 2 (d below)
The lower ray travels an EXTRA distance 2 d sin(theta).
* If that extra path = a whole number of wavelengths (n lambda),
the two rays return in step and ADD -> a bright diffraction peak.
* At any other angle they fall out of step and CANCEL -> darkness.
Bigger d (planes further apart) -> peak at a SMALLER angle theta.
So the pattern of angles is a direct read-out of the plane spacings.Let us actually use it, because the numbers are friendly. A very common lab source is a copper target, giving X-rays of wavelength lambda = 0.154 nm. Take the (110) planes of BCC iron, whose spacing turns out to be d = 0.203 nm. Bragg's law with n = 1 gives sin(theta) = lambda / (2d) = 0.154 / (2 x 0.203) = 0.154 / 0.406 = 0.379, so theta = 22.3 degrees. The detector, though, is set at the angle between the incoming and outgoing beams, which is 2 x theta, so the peak lands at 2-theta = 44.7 degrees — and if you run a real piece of ordinary steel, there indeed sits its strongest peak, right at 44.7 degrees. Run the arithmetic backwards and you see the payoff: measure the angle of a peak, read off d, and from d and the known geometry of the cell you recover the lattice parameter a to four decimal places. That is how we know the numbers we spent earlier rungs quoting.
Reading a powder pattern: phases, structure, stress
A single crystal, sitting still, would flash only when one plane happened to be at the right angle — you would have to rotate it hunting for peaks. The everyday trick is to grind the sample to a fine powder (or use a polycrystalline solid, which is already millions of tiny crystals). With grains pointing every which way, some crystallite is always oriented correctly for every set of planes at once, so you sweep the detector through 2-theta and collect the whole family of peaks in one scan. The chart of intensity versus 2-theta is a fingerprint: every crystalline phase has its own unique set of plane spacings, hence its own unique pattern of peak positions and heights, catalogued in databases of hundreds of thousands of known materials.
- Grind to a fine random powder (or mount a flat polycrystalline piece) so that crystallites face every direction, guaranteeing a peak for each plane family.
- Scan the detector slowly through 2-theta, recording intensity at each angle; sharp peaks rise wherever Bragg's law is satisfied.
- Convert each peak's angle to a plane spacing with d = lambda / (2 sin(theta)), turning a list of angles into a list of d-values.
- Match the set of d-values and relative intensities against a database — the fingerprint names the phase or phases present.
- Index the peaks by their (hkl) planes and refine the lattice parameter a; then watch it for shifts that reveal composition or residual stress.
The pattern does more than name a compound — it tells you the crystal system, because the geometry of the cell decides which reflections are even allowed. In a body-centred cubic metal the extra atom at the cell centre makes certain reflections interfere themselves to death: only planes with h+k+l even survive, so the first peaks are (110), (200), (211). In an FCC metal the rule is different — h, k, l must be all even or all odd — so the first peaks are (111), (200), (220). This is genuinely useful, not a curiosity: austenite is FCC and ferrite is BCC, so their patterns differ, and XRD can measure how much of a quenched steel transformed to martensite versus how much austenite is retained, straight from the peaks. The pattern reads out structure, not just identity.
Two honest limits, and two more free readings. First, XRD only sees things that are crystalline: a truly amorphous solid — window glass, many polymers — has no repeating planes, so it gives no sharp peaks at all, only a broad, gentle hump. That absence is itself a measurement (you can tell glassy from crystalline), but you cannot ask a glass for a lattice parameter, because it does not have one. Second, a phase present below roughly one or two percent by volume can hide under the noise entirely. Now the bonuses: peaks are not infinitely sharp, and their width carries information. Very small crystallites, a few tens of nanometres across, cannot interfere over enough planes to make a razor-thin peak, so the peak broadens — the Scherrer relation turns that width into a crystallite size. And if the whole lattice is squeezed by a compressive residual stress, every d shrinks a hair and every peak shifts to a slightly higher angle; measure the shift and you have measured the locked-in stress, which is exactly how engineers verify that shot-peening left a protective compressive skin.
Spectroscopy: asking atoms what they are, not where
Diffraction answers a geometry question — how far apart, in what arrangement — but it is nearly blind to a chemistry question. Two different oxides with the same crystal structure can give almost the same pattern; a peak position tells you a spacing, not which elements sit there. For chemistry we switch to spectroscopy, a whole family of tools that probe the energy levels of atoms and bonds. The idea is universal: fire in energy (an electron beam, X-rays, a laser, infrared light), and the atoms answer at energies that are their own signature, because electron shells and chemical bonds absorb and emit only at their characteristic energies. You met one already in guide 2 — energy-dispersive spectroscopy, the EDS on an SEM, which reads the characteristic X-rays each element gives off and so maps composition. That is the archetype; the others each listen to a different conversation.
For the outermost skin of a material there is X-ray photoelectron spectroscopy (XPS) and its cousin Auger spectroscopy. Shine soft X-rays on a surface and electrons are knocked out with energies that reveal not only which element they came from but its oxidation state — whether an iron atom is metallic Fe or bound as Fe2O3, whether chromium is metal or oxide. Because only electrons from the top few nanometres escape, XPS is exquisitely surface-sensitive, and that is its superpower. Recall the corrosion rung: stainless steel resists only because of an invisibly thin chromium-oxide passive film. XPS is the tool that actually proves that film is there, measures that it is chromium-rich oxide, and watches it break down — a claim you could not settle with bulk XRD, which sees micrometres deep and would drown a two-nanometre film in the signal of the metal beneath.
Where XPS reads elements and oxidation states, vibrational spectroscopy reads bonds. Infrared (FTIR) and Raman spectroscopy watch how molecular bonds stretch and wag: each kind of bond — C=O, C-H, a benzene ring — vibrates at its own frequency, so the spectrum is a bar-code of the chemistry present. This is the day-to-day identification tool for polymers, telling polyethylene from PVC from nylon at a glance, and it can gauge how far a thermoset has crosslinked. Raman is also the connoisseur of carbon: a sharp line near 1332 wavenumbers means diamond, a band near 1580 means well-ordered graphite, a strong 'D' band near 1350 flags defects and disorder, and a particular sharp 2D band is the signature the community uses to confirm a flake really is graphene. None of this is a lattice measurement — it is a chemistry measurement, complementary to diffraction rather than a substitute for it.
What each tool resolves — and its blind spots
Step back and the toolbox sorts itself by the question it answers. If you ask where — where are the grains, the crack, the second phase — you reach for a microscope: light for grains, the SEM for surfaces and fracture faces, the TEM for dislocations and nanometre detail. If you ask what arrangement — which crystal structure, what lattice parameter, how much of each crystalline phase, how much residual stress — you reach for X-ray diffraction. If you ask what element and what bond — composition, oxidation state, molecular identity, band gap — you reach for a spectroscopy: EDS for bulk composition in the SEM, XPS or Auger for the outermost surface, FTIR or Raman for bonds and polymers, UV-Vis for optical and electronic structure. Real characterisation almost never uses one alone; the power is in triangulation. XRD says the phase is chromium carbide, EDS confirms the chromium and carbon are there, and the SEM shows it sitting on the grain boundaries — three tools, one conclusion.
And every one of these tools has a blind spot worth naming out loud, because misreading them is where beginners come unstuck. XRD gives a statistical average over a spot maybe a millimetre wide and micrometres deep, so it will never show you a single dislocation or one lonely inclusion — that is a job for the TEM. Its peak position gives a spacing, not a composition, so on its own it cannot tell you which elements are present; you infer the phase by matching, and pair it with EDS to be sure. A scanning-probe microscope can trace a surface atom by atom but sees nothing beneath it. XPS and Auger see only the top few nanometres, so they answer a genuinely different question from bulk XRD, and the two can even disagree about the 'same' sample because they are looking at different depths. The habit to build is to ask, before trusting any single number, three questions: how deep did this tool look, how large an area did it average over, and did it measure geometry, composition, or bonding?
Notice, too, what none of this rung's tools has touched yet: none of them heats the sample and watches it change. XRD, XPS, Raman all photograph a material as it is, at one temperature, one moment. But a great deal of what a material does — a polymer softening at its glass transition, a solder melting, an alloy releasing heat as it transforms, a plastic burning off its filler — only shows up when you take it through a temperature ramp and record what happens. That is the subject of guide 4, thermal analysis, where DSC and TGA measure the heat and the weight of a sample as you cook it, and read out transitions that no still picture, however sharp, could ever reveal.