Coherence-Limiting Mechanisms
Qubit coherence is bounded by energy relaxation and dephasing, each driven by a distinct family of physical mechanisms that hardware design tries to suppress.
Two Clocks
Qubit coherence is captured by two timescales. The relaxation time, T1, measures how long an excited qubit takes to lose its energy and fall to the ground state. The dephasing time, T2, measures how long the qubit's quantum phase stays well-defined. They are related by 1/T2 = 1/(2 T1) + 1/T_phi, where T_phi is the pure dephasing time from frequency fluctuations. A good qubit needs both to be long compared with gate times.
Relaxation Mechanisms
- Dielectric loss from two-level systems in oxides and interfaces absorbing qubit energy.
- Purcell decay through the readout resonator into the output line.
- Quasiparticle tunneling across Josephson junctions.
- Radiative loss into spurious modes and coupling to lossy materials.
Dephasing Mechanisms
Pure dephasing comes from anything that makes the qubit frequency fluctuate. Flux noise dominates for flux-tunable qubits away from their sweet spots. Charge noise matters for charge-sensitive designs, which is why the transmon was engineered to be charge-insensitive. Thermal photons in the readout resonator cause dephasing through the fluctuating dispersive shift they impose. Critical-current noise in the junction shifts the qubit frequency as well.
Measuring and Separating Them
T1 is measured by exciting the qubit and watching it decay. T2 is measured by Ramsey interferometry, which is sensitive to low-frequency dephasing, while a spin-echo measurement refocuses slow drifts and reports a longer echo time, T2-echo. Comparing these separates fast, irreversible processes from slow, refocusable ones, guiding which mechanism to attack.
Design Response
Because the mechanisms are distinct, so are the fixes: better materials and interfaces for two-level-system loss, Purcell filters for readout decay, quasiparticle traps and shielding for tunneling, sweet-spot operation and dynamical decoupling for low-frequency dephasing, and thermalization of the readout mode for photon-induced dephasing. Understanding which mechanism limits a given device is the first step in every coherence-improvement effort, and it feeds directly into the error budget.