Skip to main content

How to build a Mach–Zehnder interferometer

Task-oriented. Best attempted after you've succeeded with the Michelson, since the alignment skills carry over.

The Mach–Zehnder interferometer does the same fundamental thing as the Michelson: split a beam, send the halves on different journeys, recombine them and interfere. But with one practical difference: the two paths are completely separate loops instead of back-and-forth arms. That gives you a clear, open path where you can slide a sample into one beam only, which is what makes it useful for looking at transparent objects.

The Interferometer fully assembled acts as a very senstive seismic detector

In the end you will build something that looks like this

What you need​

  • Laser diode (650 nm) + fiber
  • Two beam-splitter cubes (one to split, one to recombine)
  • One kinematic mirror, one static mirror cubes both set to 45°
  • A converging lens (beam conditioning)
  • ~8 cubes and base plate
  • The 1.5 mm hex screwdriver
  • Screen
  • Optional: camera
Laser safety

Keep the beam horizontal and below eye level, wear the goggles, and remove reflective jewellery. Trace where every split beam goes — a Mach–Zehnder has more stray reflections than a Michelson. The laser is general thought to be eye-safe, but better be safe than sorry!

The layout: a rectangle of light​

The beam travels around a rectangle:

  1. Setup the illumination - let's place the fiber-coupled laser on the base plate

  1. First beam splitter — divides the laser into two beams (call them the upper and lower path).

  1. Prepare Mirrors - open the mirrors and set them to 45°, Place them on the grid; Each path has a mirror that turns it 90° so the two paths run parallel, then head toward the far corner.

  1. Reunite the beams - add the second beam splitter + the screen

  1. Add a polarizing filter: to tune the two contributing intensity of one of the beams. In the end they should have the same intensity to have best contrast (see Michelson Contrast)

Rotating the polarizer will dimm down one of the intensities in one arm. Ensure both arms have roughly the same intensity on the screen (obscure both sequentially and watch by eye)

  1. Adjust the kinematic mirror: match the two beams on the screen so that dust that sits in front of the first beamsplitter is reunited on the scree. This is a lot more involved than the michelson interfereomter since a parallel shift of one of the mirror w.r.t. the cube's center can move the coherence volume to the side a lot. Also polarization plays a role since the beamsplitter are not non-polarizing

A good interferogramm looks like this

  1. Screen or camera catches the pattern. You can place the raspberry pi camera in place of the screen. With this it's a lot easier to align the interferometer since we see fine stripes on the ~pixel scale (several micrometer) and can iterate on the screws much faster. With the openUC2 Off Axis Holo APP running on the ImSwitch SD Card you can also observe the peaks in Fourier space and eventually reconstrct an off-axis hologramm:

Aligning it​

The principle is the same as the Michelson alignment guide: get the two output spots to overlap using the kinematic mirror screws, one screw at a time, then expand the beam to see fringes.

The extra challenge is that you have two mirrors and two beam splitters, so there are more things to get square. Work methodically:

  • First make each path hit the centre of its next component.
  • Then use the kinematic mirror for the final overlap at BS2.
  • Keep the whole rectangle rigid — every cube on puzzle pieces, on a solid surface.

Add a sample: see phase (highly advanced)​

Once you have stable fringes, gently slide a thin transparent object (a coverslip, a wisp of clear tape, a drop of liquid) into one arm. Watch the fringes bend or shift.

That shift is the sample changing the phase of the light in that arm — light travels a hair slower through glass than through air. The Mach–Zehnder has just turned an invisible, perfectly transparent object into a visible, measurable fringe shift. That's the basic idea behind quantitative phase imaging and off-axis holography, the more advanced techniques this geometry leads into.

Michelson vs Mach–Zehnder — which when?

Use the Michelson to see interference and measure tiny distances (fringe counting). Use the Mach–Zehnder when you want to put a sample in one beam and study how it changes the light — it has a clean, accessible path the Michelson lacks.