T2: Dephasing Time
T2 measures how long a qubit keeps a well-defined phase; it is bounded by T1 and usually the tighter limit on quantum computation.
Loss of phase
A superposition like (|0> + |1|)/sqrt(2) carries a relative phase that evolves in time. T2, the transverse or dephasing time, measures how long that phase stays coherent. Even if the qubit never loses energy, random fluctuations in its transition frequency scramble the phase, destroying interference. Because phase coherence is what quantum algorithms exploit, T2 is often the more binding constraint.
T2, T2*, and echo
- T2* is the raw dephasing time from a Ramsey experiment, sensitive to slow noise and drift
- T2 (Hahn echo) inserts a refocusing pulse that cancels slow noise, giving a longer time
- Dynamical decoupling with many pulses extends coherence further, up to the 2*T1 ceiling
The relationship to T1
Energy decay also destroys phase, so T2 can never exceed 2*T1. The general relation is 1/T2 = 1/(2*T1) + 1/T_phi, where T_phi captures pure dephasing from frequency noise. When T_phi is short, dephasing dominates; when the qubit is engineered to a noise-insensitive sweet spot, T2 can approach the 2*T1 limit.
Sources of dephasing
Charge noise, flux noise, fluctuating two-level-system defects, magnetic field drift, and control-line noise all jitter the qubit frequency. Operating at first-order-insensitive bias points (sweet spots) and applying echo or decoupling sequences are the main mitigations.
Reporting both T1 and T2 characterizes a qubit's coherence: T1 is how long it holds energy, T2 is how long it holds phase, and useful computation needs both to be long compared with gate times.