two-level-system (TLS) defect

Imagine that here and there inside a quantum chip there is a single atom that has two slightly different spots it could sit in, and it keeps quietly flipping between them. That tiny flippable thing behaves like its own little two-level system, an accidental, unwanted cousin of the qubit you actually built. There are thousands of them, hiding in the thin native oxide on a metal surface, in the amorphous glass under a junction, and at the interfaces between materials. They matter because each one can soak up energy at a particular frequency, and a quantum chip is exquisitely sensitive to losing even one photon's worth of energy.

The trouble starts when one of these defects happens to sit at the same frequency as a qubit or its readout resonator. Then they resonate together, and the qubit's energy leaks into the defect instead of staying where you want it, which shows up as a shorter coherence time. This is why TLS are the dominant coherence-limiter precisely under the conditions that matter most: very low microwave power, around the single-photon level, and millikelvin temperatures, where other loss channels have mostly gone quiet. Their effect is also frustratingly drifty. A defect can wander in frequency over hours or days, so a qubit that looked clean on Monday can develop a sudden bad spot on Tuesday as a TLS swings into resonance.

There is no clean way to remove them yet, only ways to have fewer and to dodge the worst. Engineers grow cleaner films, strip away lossy surface oxides, shrink the amount of amorphous dielectric that the qubit's electric field touches, and pick materials like tantalum or titanium nitride that happen to host fewer harmful defects. Tunable qubits can also be parked away from a known defect's frequency. But TLS are a property of disordered, glassy matter itself, so as of today they remain one of the hardest and least glamorous obstacles between small noisy chips and large reliable ones.