JOVANA
Library Glossary Getting Started Three Levels Fields How it works Mission
Join the mission
All guides

The Josephson Junction

Almost every superconducting quantum chip is built around one strange little circuit element. Here is what it does, why it is special, and how two of them in a loop give you a tunable qubit.

The one element that bends but does not waste

To make a qubit out of an electrical circuit, you need two ingredients that almost never come together. First, the circuit must store and release energy without leaking it away as heat — any loss blurs the delicate quantum state. Second, it must be nonlinear: its response has to bend, not run in a straight line. Ordinary inductors and capacitors are lossless but perfectly straight; resistors bend things but burn energy. The Josephson junction is the one component that does both at once, and that is why it sits at the heart of nearly every superconducting qubit.

Physically it is almost nothing: two superconductors with an insulating gap between them, so thin — a couple of nanometres, a handful of atoms — that paired electrons quietly tunnel across without resistance. The whole device is smaller than a speck of dust, yet getting that gap right to within a few atoms is one of the hardest things in the entire field.

Two rules, in plain words

Everything the junction does follows from two short relations. They look like equations, but each says something you can picture. The key character is phi (the Greek letter, written 'phi' here), the phase difference between the two superconductors — think of it as how far one side's quantum wave is shifted relative to the other.

Josephson relations (plain symbols):

  I = Ic * sin(phi)              <- current
  V = (hbar / 2e) * dphi/dt      <- voltage

  phi  = phase difference across the gap
  I    = current flowing through the junction
  Ic   = critical current (the most it can carry)
  V    = voltage across the junction
  hbar = Planck's constant / 2*pi
  e    = the electron's charge
The two Josephson relations and every symbol spelled out. Current follows a sine of the phase; any voltage makes the phase wind up over time.

The first rule, I = Ic sin(phi), is the important one. In a plain wire the current grows in a straight line as you push harder. Here it follows a sine instead, so near zero it behaves like an inductor — but an inductor whose value changes depending on how much current is already flowing. A phase-dependent, nonlinear inductance: that bend is the whole point. The second rule just says that if you put a voltage across the junction, phi keeps winding up over time, like a dial that never stops turning while you hold a battery to it.

Two junctions in a loop: a tunable knob

A single junction has a fixed critical current Ic, so the qubit you build from it has a fixed frequency. Often you want to change that frequency on the fly. The trick is to put two junctions in a small superconducting loop. This pairing is called a DC SQUID, and threading magnetic flux through the loop lets you dial its combined strength up and down.

Single junction (an 'X' marks the thin gap):

        ---[ X ]---

DC SQUID = two junctions sharing a loop:

          +--[ X ]--+
   -------+         +-------
          +--[ X ]--+
             ^^^^
      magnetic flux threads
      the loop -> tunes Ic
Top: one junction, the 'X' is the insulating gap. Bottom: two junctions form a loop; flux through the loop tunes the effective critical current.

It works by interference. The two junctions offer the current two paths around the loop. Magnetic flux shifts the relative quantum phase between those paths. When they add in phase, the loop acts like one strong junction; add half a flux quantum and they fight, and the effective Ic shrinks toward zero. So a tiny coil or a nearby current-carrying wire becomes a frequency knob — the same idea behind a tunable coupler that switches the link between two qubits on and off.

Why this one element is also the hard part

The qubit's frequency is set by Ic, and Ic is set by how thick the oxide barrier is — down to a few atoms. Two junctions meant to be identical will come out slightly different, and on a chip with many qubits those frequencies start to collide. That clash even has a name: frequency crowding. Across a whole wafer, making junctions accurate and repeatable is one of the central open problems in scaling up.

  1. Picture a small batch of qubit chips fresh from the fab, each one carrying many junctions.
  2. Because each junction's oxide is a hair thicker or thinner than intended, its frequency lands a little off target.
  3. On a crowded chip, two neighbours that drift too close can no longer be told apart by their control pulses.
  4. So engineers measure, trim, and bin junctions — and lean on the tunable SQUID trick to nudge frequencies back apart after fabrication.

None of this is solved, and being honest about it matters. The Josephson junction is the elegant heart of the superconducting quantum processor, but it is also the place where atom-scale fabrication wobble turns into chip-scale headaches. Today's machines stay small and noisy in large part because of exactly this element — and improving it, one wafer at a time, is much of what quantum chip design is actually about.