The Reciprocal Lattice as a Fourier Transform

A quick recap, then the big reveal

So far this rung has done a lot with a little. Guide 1 argued we need a reciprocal lattice because diffraction is clumsy to picture in real space; guide 2 forged the vectors a*, b*, c* from the direct cell by cross products divided by the cell volume; and guide 3 delivered the payoff sentence — every dot (hkl) is a proxy for a whole family of parallel planes, reached by the vector g_hkl = h a* + k b* + l c*, which points along the plane normal with length |g_hkl| = 1/d_hkl. Hold all of that; this guide does not rebuild it.

Now the reveal. All of that machinery has a single, deeper source: the reciprocal lattice is the Fourier transform of the direct lattice. That one sentence upgrades the shadow world from a convenient trick into a statement about how waves interact with periodic matter. Everything below unpacks it — and in return it hands you three things for free: why real and reciprocal sizes are inverses, why peak positions report the unit cell while peak intensities report the motif, and why a diffraction pattern is not a coded message about the crystal but a literal picture of reciprocal space.

One honesty before we start, to head off a natural confusion. The reciprocal lattice is not a thing sitting inside the crystal — it lives in reciprocal space and its axes are measured in 1/length (per angstrom), not length. Yet it is not merely abstract either: a diffraction camera makes it visible. And keep the rung-1 distinction sharp — the reciprocal lattice here is the transform of the lattice (the bare grid of points). The motif has not vanished; it will re-enter shortly as a set of weights, and that is where the crystal's real atoms finally speak.

What a Fourier transform actually does

Strip away the mathematics and a Fourier transform does one homely thing: it breaks a complicated pattern into a sum of pure waves, and tells you how strong each wave must be. Your ear does exactly this when it splits a piano chord into its separate notes; a prism does it when it fans white light into colours. Hand the transform a picture of the crystal's electron density — the hills where atoms sit and the valleys between — and it answers with a recipe: add up these particular ripples, each at this particular strength, and you rebuild the crystal.

Each pure wave carries two facts. First a direction — which way its crests run. Second a spatial frequency — how many crests per angstrom, i.e. how finely ribbed it is. Bundle those into one arrow: a wavevector that points along the wave's travel and whose length equals the spatial frequency. The consequence is already the inverse-size relationship in disguise: fine detail (small features, small spacings) demands finely ribbed waves — high frequency — a long arrow far from the origin, while coarse, gently varying features need a short arrow near the origin.

Now the crux. A crystal's density repeats with the lattice's period, so out of all conceivable waves only those whose ribs line up perfectly with that repeat can survive — any other wave would drift out of step with the atoms and average away to nothing over the endless crystal. The wavevectors that DO survive are exactly the reciprocal lattice points. Read a dot (hkl) this way: it is the one density wave whose crests lie parallel to the (hkl) planes and repeat every d_hkl — so its arrow points along the plane normal with length 1/d_hkl. That is precisely the g_hkl of guide 3, now re-derived as a Fourier component rather than a geometric proxy.

The transform of a lattice is a lattice — and it flips size

Here is the theorem that makes the whole scheme click: the Fourier transform of a periodic array of points is again a periodic array of points — sharp spikes, and flat nothing in between. Transform the direct lattice and out comes the reciprocal lattice; transform that in turn and you land back on the direct lattice. They are a matched pair — a Fourier dual. That mutual mirroring is why 'reciprocal' is the honest word for it: neither is more fundamental, each is simply the other seen in frequency space.

The transform also turns size inside out — a fact you glimpsed in guide 2, now with its reason. Stretch the real cell and every wave that must fit inside it lengthens too, so its spatial frequency drops and its reciprocal point creeps toward the origin. Quantitatively this is the construction you already have, a* = (b x c) / V with V the direct-cell volume, and it pins down the reciprocal cell volume exactly: V* = 1/V. A large real cell forces a small, tightly packed reciprocal unit cell, and vice versa.

Put numbers on it. A cubic crystal with edge a = 4.00 angstrom has a* = b* = c* = 1/4 = 0.25 per angstrom, so its reciprocal cell is a tiny cube 0.25 per angstrom on a side; the direct volume V = 64 angstrom^3 gives V* = 1/64 = 0.0156 per angstrom^3. Check one dot: g_200 has length 2 a* = 0.50 per angstrom, which is 1/d_200 = 1/2.00 angstrom — consistent. This is why a protein crystal, whose real cell can be 100 angstrom across, throws a reciprocal lattice so dense the diffraction spots almost touch, while a simple metal with a 4-angstrom cell spreads its spots out generously.

Because a crystal is lattice PLUS motif

Recall the founding mantra of this whole ladder: a crystal is a lattice plus a motif, built by stamping the same motif at every lattice point. In transform language, 'stamp a motif at every point of a lattice' is precisely a convolution. And the convolution theorem — the second great gift of Fourier analysis — says the transform of a convolution is just the ordinary product of the two separate transforms. So FT(crystal) = FT(lattice) times FT(motif): the reciprocal lattice grid, multiplied point by point by the motif's own transform.

This factorisation is the honest engine behind a rule you have already met. The grid part fixes WHERE spots may appear — the reciprocal lattice, and hence the unit cell. The motif's transform fixes HOW BRIGHT each one is. Attach the brightness |F_hkl|^2 to each reciprocal point and you get a weighted reciprocal lattice, where F_hkl is the structure factor — the motif's transform sampled at that dot. In one line: peak POSITIONS report the cell, peak INTENSITIES report the motif. When the motif's transform happens to vanish at a grid point, that spot goes dark — a systematic absence, the fingerprint by which you spot centring like FCC or BCC.

Reciprocal-lattice row along a*      (crystal = lattice x motif)

  index hkl :  000    100    200    300    400    500
  position  :   0     1a*    2a*    3a*    4a*    5a*      a* = 1/a
  spot      :   O      o      .      O      .      o
  |F_hkl|^2 : bright   med   ABSENT  big   ABSENT small

  GRID SPACING  a* = 1/a    <- set by the LATTICE (the unit cell)
  SPOT BRIGHT   |F_hkl|^2   <- set by the MOTIF  (structure factor)
  a blank ( . ) = systematic absence: motif transform is zero there

A diffraction pattern is a photograph of reciprocal space

Now bolt the physics on. To first approximation — weak, single scattering, the so-called kinematic regime — the amplitude an X-ray scatters into a given direction is proportional to the Fourier transform of the electron density, evaluated at the scattering vector s = k_out - k_in. Read that slowly: the detector, direction by direction, is sampling the crystal's transform. It is not decoding a hidden message; it is literally photographing the reciprocal lattice. That is the deep meaning of a diffraction pattern being the reciprocal lattice.

A spot flares only when the scattering vector s lands exactly on a reciprocal point, s = g_hkl. That is the Laue condition — and it is just Bragg's law wearing reciprocal-space clothes: the two say the same thing, one as a vector equation, the other as lambda = 2 d sin(theta). The geometry that decides which reciprocal points a given wavelength can actually reach is the Ewald sphere, and it is worth a quick first look before guide 5 gives it the full stage.

1. Draw the reciprocal lattice, and mark its origin (the 000 point) — this is the map every incoming beam will explore.
2. Draw the incident wavevector k_in, of length 1/lambda, arriving so that its tip lands on that origin.
3. Swing a sphere of radius 1/lambda centred on the tail of k_in — this is the Ewald sphere, the sphere of reflection.
4. Any reciprocal point that happens to sit ON that sphere satisfies s = g_hkl and diffracts; every point off the sphere stays dark. Rotate the crystal and you sweep new points onto the sphere.

Be honest about the fine print of 'diffraction = Fourier transform': it holds while scattering stays weak. Electrons interact with matter roughly 10^4 times more strongly than X-rays, so a beam ploughing through even a thin foil is re-scattered again and again — dynamical scattering — and the simple 'intensity = |F|^2' bookkeeping breaks down. That is why a transmission electron microscope needs ultrathin specimens and still demands more careful theory, whereas X-rays and neutrons on small crystals sit comfortably in the clean kinematic regime.

A first look at the Brillouin zone, and why we live in reciprocal space

One more inhabitant of reciprocal space deserves a first sighting. Back in the lattice rung you built the Wigner-Seitz cell — the primitive cell you carve out by claiming all the space closer to one lattice point than to any other. Run that exact same construction in reciprocal space, around one reciprocal point, and the region you carve out is the Brillouin zone. Its walls are the perpendicular bisector planes of the g-vectors — which are precisely the Bragg planes — so the zone boundary is exactly the surface where a wave meets the diffraction condition and is turned back.

Why physicists care so much: electrons and lattice vibrations inside a crystal are themselves waves, and their natural home is reciprocal space. The Brillouin zone is the stage on which electron energy bands and phonon dispersion curves are drawn — the whole language of solid-state physics is written on it. For this rung, just pocket the one-line definition; guide 5 opens both the Ewald sphere and the Brillouin zone properly and shows them doing real work.

Stand back and the reason we live in reciprocal space is plain. Because a crystal is periodic, everything wavelike about it — diffraction spots, electron bands, phonon branches — is simplest as a short, tidy list of reciprocal points, instead of an infinite tangle of real atoms. The reciprocal lattice is the crystal's Fourier shadow, and diffraction is the lamp that casts it onto film. Carry three honest boundaries forward into guide 5 and the diffraction rung beyond: the reciprocal lattice is the transform of the LATTICE, weighted by the motif's transform; a measurement keeps |F|^2 but loses the phase; and the clean 'diffraction equals Fourier transform' picture assumes weak, kinematic scattering. Everything in reciprocal space flows from those three.