Somewhere behind the plastic panel of every microwave oven sits a short rectangular metal tube. It connects the magnetron, which generates the microwaves, to the cooking chamber. It has no wire inside it. No conductor runs down its centre. Yet it delivers hundreds of watts of electromagnetic energy with remarkably little loss. That empty metal box is a waveguide, and the way it works seems almost paradoxical at first.

We usually think of electrical energy flowing through wires. Maxwell's equations tell a different story. In any circuit, the energy actually travels in the electromagnetic field surrounding the conductors, and the wires merely steer it. A waveguide takes this idea to its logical conclusion. It removes the inner wire entirely and lets the field travel on its own, held in place by the walls.

Radar systems, satellite ground stations, particle accelerators and radio telescopes all rely on these hollow pipes. To understand them, we need three ideas. The first is how metal walls trap a wave. The second is why a pipe refuses to carry frequencies that are too low. The third is how the field arranges itself into distinct patterns called modes.

Start with what a good conductor does to an electric field. Inside a metal, free electrons respond almost instantly to any applied field. If an electric field lies along the surface of the metal, the electrons shift until they cancel it. The result is a strict boundary condition: at the surface of a perfect conductor, the tangential electric field must be zero. Only the component pointing straight into the wall can survive.

Now send an electromagnetic wave into a rectangular metal tube. Whenever the wave strikes a wall, it drives currents in the metal's surface. Those currents radiate a new wave that exactly cancels the tangential field at the wall, and that reflected wave heads back into the interior. The energy cannot leak out through the walls, so it has only one place to go: along the axis of the pipe.

A helpful way to picture this is to treat the guided wave as two ordinary plane waves crossing at an angle. Each one bounces between the side walls in a zigzag pattern, and where they overlap, they interfere. Across the width of the guide, they form a standing wave with zero electric field at the walls. Along the length, they combine into a travelling wave that carries energy forward.

The zigzag model explains a curious feature. The pattern of crests moves along the guide faster than light, a quantity called the phase velocity. But no information or energy travels at that speed. The energy follows the zigzag path, so it advances along the axis more slowly than light, at the group velocity. The two speeds are linked: their product equals the square of the speed of light. Nothing in the pipe breaks relativity. The fast-moving crests are just geometry, like the point where an ocean wave meets a beach at a shallow angle and seems to race along the shore.

Takeaway

A waveguide does not push energy through metal; it uses metal as a mirror. The wave travels through empty space, and the walls simply forbid it from going anywhere except forward.

The zigzag picture also reveals the waveguide's most important limitation. As the frequency drops, the wavelength grows. To keep the field at zero on both side walls, the two plane waves must strike the walls at steeper and steeper angles. Eventually they bounce straight across the guide, back and forth, making no forward progress at all. The frequency at which this happens is the cutoff frequency.

For the simplest field pattern in a rectangular guide, the condition is easy to state: half a wavelength must fit across the wider dimension. If the width is a, the cutoff frequency equals the speed of light divided by 2a. A standard X-band waveguide is about 22.9 millimetres wide, which gives a cutoff near 6.6 gigahertz. Above that frequency, waves travel freely. Below it, they do not travel at all.

A wave below cutoff is not reflected in the ordinary sense. The mathematics produces a propagation constant that is imaginary rather than real. Instead of oscillating along the guide, the field decays exponentially with distance. This is called an evanescent field. Well below cutoff, the field shrinks by a factor of e, roughly 2.7, over each distance equal to the guide width divided by π. A pipe just a few widths long therefore blocks the signal almost completely.

You can see this physics on the front of your microwave oven. The door contains a metal mesh with holes a few millimetres across, and each hole acts as a tiny waveguide. A circular hole 3 millimetres wide has a cutoff near 60 gigahertz. The oven's 2.45 gigahertz radiation, with a wavelength of about 12 centimetres, sits far below that threshold and cannot pass. Visible light, at hundreds of terahertz, passes straight through. The same holes are a wall for one wave and a window for another.

Takeaway

Size sets the frequency floor. A waveguide is a high-pass filter built from geometry alone, and any opening much smaller than half a wavelength is effectively a barrier to that wave.

Fitting half a wavelength across the guide is only the first solution. One full wavelength also fits, as do one and a half, and the field pattern can vary across the narrow dimension too. Each allowed field arrangement is called a mode, and each has its own cutoff frequency. Modes are the waveguide equivalent of the harmonics on a guitar string: the boundaries permit only certain shapes.

Hollow guides support two families of modes. In transverse electric (TE) modes, the electric field lies entirely across the guide, with no component pointing along the axis. In transverse magnetic (TM) modes, the magnetic field is the one that stays purely transverse. A single hollow conductor cannot support the TEM mode used in coaxial cable, where both fields are transverse. That mode needs two separate conductors. This is why a waveguide has a cutoff frequency and a coaxial cable does not.

The lowest-frequency mode in a rectangular guide is called TE10. Its electric field points between the broad walls, strongest at the centre and falling to zero at the sides. Higher modes, such as TE20 and TE01, only begin to propagate at higher frequencies. Engineers make the guide roughly twice as wide as it is tall. This gives TE10 a frequency band up to an octave wide in which it is the only mode that can propagate.

Why insist on a single mode? Because each mode travels at a different group velocity. If a signal splits its energy across several modes, the pieces arrive at different times and the pulse smears out, a problem called modal dispersion. Bends, joints and imperfections can also shift energy from one mode to another in unpredictable ways. Running a single mode keeps the field pattern known and the timing clean. Single-mode optical fibre, which carries long-distance internet traffic, follows exactly the same logic at a far smaller scale.

Takeaway

Boundaries do more than contain a wave; they decide which shapes it is allowed to take. Good design often means choosing dimensions so that only one shape is possible.

A waveguide is Maxwell's equations made tangible. The metal walls impose one simple rule: no tangential electric field at the surface. Everything else follows from that single constraint, including confinement, the zigzag of reflections, a sharp cutoff frequency and a discrete family of modes.

The deeper lesson is that energy travels in fields, not in matter. Wires, pipes and fibres are not conduits for the energy itself. They are boundary conditions that shape where the field can exist and how it moves.

Seen this way, a microwave oven door, a radar feed and an undersea optical cable are versions of the same idea. Choose the boundaries carefully, and you can steer energy wherever you need it to go.