Ask an engineer how good a qubit is and, sooner or later, two numbers come up: T1 and T2. They sound like model names, but they are timers. Each measures how long a qubit can hold onto a specific piece of information before the world erases it. Nearly everything else a quantum computer does, from gate speed to error rates, is a negotiation with these two clocks.
Energy leaks: the T1 clock
A qubit stores information in two states, usually labeled 0 and 1. In a superconducting chip, the 1 state carries a little more energy than the 0 state, and nature dislikes holding extra energy. Given the chance, the qubit dumps that energy into its surroundings and slumps back down to 0. This is relaxation, and T1 measures how fast it happens.
To read T1, you excite the qubit to 1, wait a set amount of time, and check whether it is still there. Repeat with longer and longer waits, and you get a curve that decays exponentially. T1 is the time it takes for roughly 63 percent of that excitation to have leaked away. A modern superconducting qubit might post a T1 of a couple hundred microseconds; trapped ions and neutral atoms, which store information in stable internal states rather than fragile energy levels, can hold on far longer.
What steals the energy? Stray infrared photons, lossy dielectric materials near the qubit, and microscopic defects in the metal and its oxide layers all offer a place for the energy to escape. Chasing longer T1 is largely a materials problem, which is why fabrication teams fuss over cleaner films, better junctions, and packaging that keeps warm radiation out.
Phase scrambles: the T2 clock
T2 is subtler, and it is where the quantum weirdness lives. A qubit doesn't have to be fully 0 or fully 1. It can sit in a superposition, a blend of both, and the useful part of that blend is the phase, a kind of internal angle that says how the two components line up. Quantum algorithms encode their answers in these phases and in the interference between them.
The trouble is that the phase advances at a rate set by the qubit's frequency, and that frequency wobbles. A drifting magnetic field, a fluctuating charge trapped in the substrate, or a slow shift in the control electronics all nudge the frequency up and down. Each nudge advances the phase a little too fast or too slow. After enough random jostling, the qubit forgets which angle it was supposed to hold. That loss is called dephasing, and T2 measures how long the phase survives.
Crucially, T2 can never exceed twice T1. Losing energy also destroys phase, so relaxation sets a hard ceiling. In practice T2 is usually shorter than that limit because dephasing piles on top of relaxation. When you hear that a qubit's T2 is close to its T1 ceiling, that is a sign the noise environment is unusually quiet.
Why the numbers rule the machine
These two clocks set the pace for everything. A logic gate takes time, tens to hundreds of nanoseconds on a superconducting device. Divide the coherence time by the gate time and you get a rough sense of how many operations you can perform before the qubit forgets. That ratio, more than the raw qubit count, decides whether a real algorithm can finish before noise wins.
Engineers fight back on both fronts. Spin-echo and dynamical-decoupling pulses flip the qubit mid-computation to cancel slow frequency drifts, effectively stretching T2 without touching T1. Better shielding, purer materials, and colder fridges push T1 higher. Some qubit designs, like fluxonium, are deliberately shaped to be less sensitive to the noise that scrambles phase.
There is also a reporting subtlety worth knowing. A single quoted T2 often comes from a Ramsey measurement, which is sensitive to slow drift, while an echo measurement gives a longer, more flattering number. When you compare machines, it pays to ask which T2 someone means.
Error correction eventually promises to hide these clocks behind logical qubits that outlive their physical parts. But even a fault-tolerant machine is built on physical qubits, and the longer their T1 and T2, the fewer of them you need to protect a single logical bit. The two timers never really go away. They just get buried deeper in the design.
