Two ways to see the unseeable
Across this rung we have made bold claims: atoms sit on a periodic lattice, a metal is a mosaic of grains, a glass has short-range but no long-range order. A fair beginner's question is simply: says who? Nobody has ever held an atom up to the light. The whole business of learning structure from real materials is called structure characterization, and it rests on a hard physical fact.
You cannot see something much smaller than the wavelength you look with. Atoms sit about 1 to 2 angstrom apart (1 angstrom = 10^-10 metre). Visible light has a wavelength around 5000 angstrom — thousands of times too coarse, like trying to feel the grooves of a vinyl record with a beach ball. To resolve atoms you need a probe whose wavelength is itself about an angstrom. Two such probes exist and are handled in two completely different ways, which is why there are two great families of tools.
The first family, diffraction, does not make a picture at all. It shines a wave of the right wavelength through the material and photographs the interference pattern that comes out, then reasons backwards to the arrangement. The second family, microscopy, tries to form an actual magnified image — of atomic columns, of a surface, or of the grains and phases we met in the last guide. Diffraction gives you numbers of exquisite precision but no direct picture; microscopy gives you a picture but only of a small, often thin, patch. The two are partners, not rivals.
Diffraction: photographing the crystal's shadow
Here is the whole idea in one picture. Send an X-ray wave at a crystal and each plane of atoms acts like a faint mirror. Reflections from evenly spaced planes are like echoes bouncing off a row of evenly spaced cliff walls: at most angles the echoes arrive out of step and cancel, but at a few special angles they arrive perfectly in step and reinforce into a bright beam. That condition is Bragg's law, n times lambda = 2 times d times sin(theta), where d is the spacing between the planes and theta is the glancing angle.
Put in real numbers. Copper K-alpha X-rays have lambda = 1.54 angstrom; for a set of planes spaced d = 2 angstrom, the first bright beam (n = 1) sits at theta = arcsin(1.54 / (2 times 2)) = arcsin(0.385) = 22.6 degrees. Every such angle is a fingerprint of one plane spacing, so the collection of angles maps out the whole reciprocal lattice — the crystal's shadow, in which each spot at distance 1/d points along the normal to a family of planes. The deep rule to carry forward: peak positions tell you the size and shape of the unit cell, while peak intensities tell you what sits inside it — the motif.
X-rays, electrons, neutrons: choosing your probe
Three probes have angstrom-scale wavelengths, and each 'feels' a different part of the atom. X-rays scatter off the atom's cloud of electrons, so heavy atoms shout and light atoms whisper — hydrogen is nearly invisible. Electrons interact with charge far more strongly, perhaps ten thousand times, which is a double-edged sword: you can work with a tiny speck of material, but the sample must be shaved ultra-thin and the scattering is so strong the wave can bounce more than once (dynamical scattering) instead of the clean single bounce X-rays enjoy. Neutrons scatter off the atomic nuclei and off unpaired electron spins, so they see light atoms clearly and, uniquely, can map magnetic order.
WHAT YOU WANT TO KNOW GO-TO TOOL WHAT IT GIVES BACK -------------------- ---------- ------------------ unit cell + which phases powder X-ray diffraction peak positions -> cell; ID the crystal full 3D atom positions single-crystal XRD the whole motif (once phases are solved) light atoms / magnetism neutron diffraction H and Li, plus magnetic order an image of atom columns TEM / HRTEM real-space picture of a thin slice surface atoms one-by-one STM / AFM the top layer, mapped in real space grains + orientation SEM + EBSD microstructure map + texture
There is also a choice of sample form. If you can grow one good single crystal, single-crystal diffraction spreads its reflections out as separate spots and gives the richest data. Usually you cannot, so you grind the material to a fine powder of millions of tiny randomly-oriented crystals; then powder X-ray diffraction collapses those spots into rings, read out as a one-dimensional trace of peaks versus angle. Powder is fast and forgiving — perfect for identifying which phases are present — but overlapping peaks make solving a brand-new structure harder.
Microscopy: a real-space picture (and its catch)
The other family forms an actual image. A transmission electron microscope fires electrons through a wafer-thin slice and, in its high-resolution mode, can show rows of atomic columns directly — the closest thing to a photograph of a lattice. A scanning electron microscope rasters a beam across a bulk surface to picture grains and fracture faces; add electron backscatter diffraction and it also reports the crystal orientation at every point, which is how the texture of a whole sheet of metal gets mapped. Scanning probes (STM and AFM) drag an atom-sharp tip across a surface and trace out its individual atoms.
But a picture has its own honest catch, and it is the mirror image of the phase problem. A single micrograph is a two-dimensional section through a three-dimensional material — one slice through the loaf of bread. The grains you see cut across their true shapes at random, so the circles in the image are smaller than the real grains. Recovering true 3D sizes from 2D slices is a whole discipline called stereology: it is how the grain size of guide 4 is actually measured, and it is the reason you never trust a single field of view.
The mindset: no single tool tells the whole story
Step back and the map has a simple logic that ties this whole rung together. Diffraction lives in reciprocal space and averages over billions of unit cells, so it is superb at the atomic and crystal scale — precise cell edges, precise atom positions — but it blurs any one-off feature into the average. Microscopy lives in real space and shows you individual features — this grain, this boundary, this dislocation — but only locally, and often only near a surface or in a thin slice. The length scales of guide 2 quietly decide which tool answers your question.
- Name the length scale of your question — atomic spacing, nanometre feature, grain, or bulk? (recall guide 2).
- Ask order or disorder: a sharp diffraction pattern means periodic crystal; broad halos mean amorphous (recall guide 3).
- Choose the probe by what it feels — X-rays for the average cell and motif, neutrons for light atoms and magnetism, electrons for a tiny or thin sample.
- Cross-check with an image: let microscopy show the real features that the averaging diffraction pattern hides.