Light fine enough to see an atom
The whole diffraction rung rested on one demand you should carry into this one: to read a structure whose features are spaced a couple of angstrom apart, your probe wave must have a wavelength of about that same size. That is the diffraction wavelength requirement, and it is why we cannot photograph atoms with a light microscope. Visible light has a wavelength near 5000 angstrom — two thousand times too coarse. Trying to resolve atoms with it is like trying to feel the grooves of a vinyl record while wearing oven mitts: the probe is simply far bigger than the thing you want to touch.
X-rays sit in exactly the right band. A photon of a few thousand electron-volts carries a wavelength of about 1 angstrom — the same scale as an atomic bond, as a Bragg plane spacing, as the unit cell itself. Feed such a wave into a crystal and it does not just reflect; it diffracts, scattering off the electron clouds of every atom and interfering constructively only at the special angles the last rung taught you to expect. So the first practical question of this whole rung is blunt: where do we get a steady, controllable beam of one-angstrom light? The answer, for the ordinary laboratory, is a device barely larger than a soda can — the sealed X-ray tube.
Keep straight what the beam actually touches. X-rays are scattered by electrons, not by the nucleus, so a heavy atom (many electrons) scatters far more strongly than a light one — that fact is bottled up in the atomic scattering factor, and it is why hydrogen is nearly invisible to X-rays while lead shouts. Hold that thought: the source sets the wavelength and the brightness, but what scatters and how much is a separate story, the one that eventually gives us peak intensities and the atomic motif.
Inside the sealed tube: how a copper atom rings
A sealed X-ray tube is almost comically simple. A heated filament boils off electrons; a voltage of around 40 kilovolts flings them across a vacuum gap and slams them into a water-cooled metal block, the anode, most often pure copper. It is a dreadful energy converter — about 99 percent of the beam's power turns straight into heat (which is why the anode is water-cooled, and why high-power labs spin it on a rotating drum), and barely 1 percent leaves as X-rays. Those X-rays come in two utterly different flavours, mixed together in the same beam, and telling them apart is the key to everything.
The first flavour is a smear. As each electron plunges into the copper and is braked by its atoms, it sheds energy as a photon — a little braking each time, a lot occasionally — so the tube pours out a broad, smooth CONTINUOUS spectrum of every wavelength above a sharp short-wavelength limit set by the tube voltage. This continuum is the white radiation (also called bremsstrahlung, German for braking radiation), and it is a mess: many colours at once. On its own it is the wrong tool for ordinary crystallography, though we will see it has one classic use.
The second flavour is a set of razor-sharp spikes, and this is the prize. Now and then an incoming electron does not merely brake — it scores a direct hit, knocking a tightly bound electron clean out of a copper atom's innermost K-shell. That leaves a hole, and an outer electron drops in to fill it, releasing the energy difference as a photon of one exact wavelength. When an L-shell electron makes the jump we get the characteristic line called Cu K-alpha, at 1.5418 angstrom; a jump from the M-shell gives the fainter, shorter Cu K-beta at about 1.392 angstrom. These wavelengths are fixed by copper's own energy levels — not by the voltage, not by the current — which is exactly why they are called characteristic. Change the anode metal and you change the colour: molybdenum rings at a much harder 0.71 angstrom, cobalt at 1.79, chromium at 2.29.
OUTPUT of a copper X-ray tube (intensity vs wavelength, angstrom)
I | Ka (1.5418 A)
| ||
| || <- sharp CHARACTERISTIC
| || line, set by copper
| Kb ||
| (1.392 A) ||
| | ||
| ___..---'''|'''----...||...___
| _.-'' | || ''--..__
| .-' continuous "white" bremsstrahlung '-.
+---+----+------------+------------+------------+---------> lambda
0.5 swl 1.0 1.5 2.0
swl = short-wavelength limit, pushed left by higher tube voltage
the two spikes ride ON TOP of the smooth continuum; we want only Ka
White light, one clean colour: monochromating the beam
Most crystallography wants monochromatic radiation — a single wavelength — because Bragg's law ties angle to wavelength, and a beam full of colours would smear every reflection into a rainbow. So we must strip the continuous hump and the K-beta spike away and keep only Cu K-alpha. There are two standard ways to do it, and both are cheap tricks of physics you already understand.
- The filter. Slip a thin foil of the element one place below the anode in the periodic table — nickel for a copper tube. Nickel's own K absorption edge sits at 1.488 angstrom, tucked neatly BETWEEN Cu K-beta (1.392) and Cu K-alpha (1.542). Wavelengths just short of 1.488 are gobbled up, so the foil devours K-beta and the short white tail while letting K-alpha stream through. Simple, cheap, but it only suppresses K-beta — it does not give you a truly single line.
- The monochromator. For a genuinely clean single wavelength, bounce the beam off a good crystal (graphite, silicon, germanium) set at the precise Bragg angle for K-alpha. That crystal obeys the very same Bragg's law as your sample: only the wanted wavelength reflects at that angle, and everything else misses. It costs some intensity but delivers a beam pure enough for careful work.
Be honest about how monochromatic 'monochromatic' really is. Even Cu K-alpha is not one wavelength but a close doublet — K-alpha-1 at 1.5406 and K-alpha-2 at 1.5444 angstrom, in a roughly 2-to-1 intensity ratio, because the L-shell it drops from is itself split. At low diffraction angles the two overlap into one peak, but push to high angle and every reflection visibly splits in two. Good refinement software models the doublet explicitly. Meanwhile the opposite choice — keeping ALL the colours, the raw white beam — is not useless at all: flood a single stationary crystal with polychromatic radiation and each plane family simply picks out the wavelength that satisfies Bragg's law for it, producing the classic Laue photograph. White light is the wrong tool for measuring spacings but the right tool for reading a crystal's symmetry at a glance.
The synchrotron: the brightest torch we own
The sealed tube's braking-radiation idea, scaled up to the sublime, becomes the synchrotron. Steer a beam of electrons up to nearly the speed of light around a storage ring hundreds of metres across, and every time a magnet bends their path they radiate X-rays — the same physics as an electron braking in copper, but from particles moving so fast and turning so hard that the emission is staggering. A synchrotron beam can be a billion times brighter than a lab tube, is finely collimated into a pencil, and above all is TUNABLE: instead of being stuck with copper's fixed 1.54 angstrom, you dial the wavelength to whatever your experiment needs.
That tunability and brilliance unlock things a benchtop tube simply cannot do. A tiny or weakly scattering crystal — a protein, a pressed grain of a new mineral, a thin film only nanometres deep — gives a usable pattern in seconds instead of days. You can park the wavelength right at an element's absorption edge to make that one element stand out (anomalous scattering, a lever that helps crack the phase problem for proteins). You can focus to a micron-wide spot to map how structure varies across a real part, or fire pulses to watch a reaction unfold in real time. The synchrotron is the tube's dream self.
From a beam to a structure: the road through this rung
With a clean one-angstrom beam in hand, the rest of the rung splits into two experimental styles, and the next guide is devoted to their contrast. In powder diffraction you grind the material into millions of randomly oriented tiny crystals, so that at any moment SOME of them are correctly angled for each plane family; the reflections that would be spots from a single grain merge into smooth cones, recorded as a one-dimensional trace of intensity against angle, the diffractogram. In single-crystal diffraction you keep one good crystal and rotate it in the beam, harvesting the full three-dimensional constellation of separate reflections — the weighted reciprocal lattice itself, mapped point by point.
Either way, three separate things about each peak carry three separate messages, and the guides ahead take them one at a time. A peak's POSITION (its angle) pins down the d-spacings, and hence the size and shape of the unit cell. A peak's INTENSITY encodes the motif — which atoms sit where inside that cell — through the structure factor you met last rung. And a peak's WIDTH betrays how big and how strained the crystallites are. Position from the lattice, intensity from the motif: it is the lattice-is-not-the-crystal split from the very start of this ladder, speaking now through a diffractometer.