A junction plus one big capacitor
If you have read about the junction already, the transmon is almost disappointingly simple to draw. Take a single Josephson junction — the one lossless, nonlinear element you need — and wire a large capacitor straight across it, in parallel. That is the whole circuit. The junction supplies the nonlinearity that bends the energy ladder so you can pick out two levels; the capacitor is there to calm everything down. Almost every chip you read about in the news is a grid of exactly this little pairing, repeated and wired together.
A transmon = one junction shunted by a big capacitor:
+----[ X ]----+
| |
-----+ +-----
| |
+---| |---| |--+
C (large)
[ X ] = the Josephson junction (the thin gap)
C = the shunt capacitor, deliberately big
Both sit across the same two nodes (in parallel).Why bolt on a big capacitor at all, when it adds nothing nonlinear? Because of charge. A bare junction is painfully sensitive to stray electric charge drifting onto it from the surrounding materials — and that charge wanders unpredictably, smearing the qubit's frequency from one moment to the next. The capacitor's job is to make the qubit's energy depend on charge as little as possible. The name even records the trick: transmon is short for *transmission-line shunted plasma oscillation qubit* — a junction whose plasma oscillation has been shunted, or calmed, by that extra capacitance.
One knob: the E_J/E_C ratio
A transmon really has just two energies that matter, and the design comes down to their ratio. One, written E_J, is the junction energy — how strongly the junction wants to keep current flowing smoothly; you set it by choosing the junction's size. The other, E_C, is the charging energy — how much it costs to push one extra unit of charge onto the island; the big capacitor is what makes E_C small. The entire transmon design problem is choosing the ratio between these two.
The transmon design knob (plain symbols): E_C = e^2 / (2 * C) <- charging energy ratio = E_J / E_C ~ 50 to 100 (transmon regime) e = the electron's charge C = the shunt capacitance (bigger C -> smaller E_C) E_J = junction energy (set by junction size) E_C = charging energy (set by the capacitor) Big capacitor -> small E_C -> large E_J/E_C ratio.
Making the ratio large is the whole point of the transmon, and it buys you one prized property: near-total immunity to charge noise. When E_J/E_C is up around 50 to 100, the wandering stray charge that tortures a bare junction barely moves the qubit's frequency at all. That single choice is most of why transmons became the default — they hold a steady frequency long enough to actually compute with, where earlier charge-based qubits jittered hopelessly.
The price of robustness: small anharmonicity
Nothing is free. The same large E_J/E_C ratio that flattens out charge noise also flattens out the very nonlinearity you went to the junction for in the first place. The leftover bend in the energy ladder is called the [[qubit-anharmonicity|anharmonicity]] — the gap between how far apart the bottom two rungs sit versus the next pair up. In a typical transmon it is only about 200 MHz, a few percent of the qubit's own frequency. The more charge-robust you make the qubit, the smaller this number gets. That tug-of-war is the central tradeoff of transmon design.
Why does such a small number matter so much? Because anharmonicity sets the speed limit on your gates. A control pulse meant to flip the qubit between its two working levels is never perfectly clean — it spreads a little energy onto nearby frequencies. If the third level sits only 200 MHz away, a fast, sharp pulse will spill some population up into it, and the qubit leaks out of the two-level space you are trying to compute in. So small anharmonicity forces you to drive more gently and therefore more slowly, or to shape pulses cleverly. It is a genuine, unavoidable engineering constraint.
The other road: fluxonium
If small anharmonicity is the transmon's price, you might ask whether some other circuit pays a different one. It does. The fluxonium is a cousin qubit that adds a third ingredient — a long chain of many junctions acting as a large inductor — alongside the usual junction. That extra piece reshapes the energy ladder so the lowest two levels are pushed far from everything above them. Its anharmonicity can be many times a transmon's, sometimes large enough that leakage almost stops being a worry.
So why isn't everyone using fluxonium? Because it moves the cost rather than removing it. That big inductor is a chain of dozens or hundreds of junctions, every one of which must be made well — harder to fabricate and more ways to go wrong. Its levels often sit at awkward frequencies that demand more elaborate control electronics and more careful operation. The transmon's appeal was always its plainness: a junction and a capacitor, easy to make in quantity. Fluxonium trades that simplicity for anharmonicity, and which trade is right depends on what the rest of the chip can support.