Hardware & Errors
Coherence Time
The duration a qubit can maintain its quantum state before decoherence causes it to lose quantum information.
Definition
Coherence time describes how long a qubit can preserve its quantum properties — superposition and phase relationships — before environmental noise causes it to decay into a classical-like state. Longer coherence times allow more operations to be performed before errors accumulate.
In depth
Coherence time is how long a qubit can hold a quantum state before noise corrupts it. It is usually given by T2, the time over which a superposition keeps its phase, together with T1, the energy relaxation time. Coherence times differ enormously between platforms. Superconducting qubits reach tens to hundreds of microseconds in processors and more than a millisecond in laboratory devices. Trapped ions hold hyperfine states for seconds and, with refocusing pulses, for minutes. Neutral atoms reach seconds, nitrogen-vacancy centers reach milliseconds at room temperature, and nuclear spins of donors in purified silicon have kept coherence for minutes to hours in special conditions.
A raw coherence time does not tell you how powerful a machine is. What matters is the number of operations that fit inside it, roughly the coherence time divided by the gate time. A superconducting qubit with a 100-microsecond T2 and 50-nanosecond gates fits about 2,000 gates, and a trapped ion with a one-second T2 and 100-microsecond gates fits about 10,000. Slow gates on long-lived qubits and fast gates on short-lived ones end up with similar budgets, though control accuracy sets the actual error.
Coherence times are not fixed. They drift because of fluctuating defects in materials, temperature changes, and cosmic-ray events that disrupt many qubits at once. Fabrication improvements have raised superconducting coherence by roughly a factor of a thousand since 1999, from nanoseconds to over a millisecond in the best devices. Dynamical decoupling and error correction can extend the effective coherence of stored information far beyond the bare qubit's, and the goal of fault-tolerant machines is to make the logical coherence time effectively unlimited.
Technical Definition
Two specific coherence times are commonly measured: T₁ (relaxation time, how long before a qubit in |1⟩ decays toward |0⟩) and T₂ (dephasing time, how long phase information is preserved). T₂ is always less than or equal to twice T₁ in standard models.
Visual Explanation: An Analogy
Think of coherence time like how long a spinning top keeps spinning before friction and air resistance slow it down and it falls over — the longer it spins, the more 'useful time' you have before it loses its special spinning state.
Key facts
| Item | Detail |
|---|---|
| Superconducting qubits | About 100 to 300 microseconds in processors; above 1 ms in lab devices |
| Trapped ions | Seconds, and minutes with dynamical decoupling |
| Neutral atoms | Seconds for nuclear spin states |
| NV centers | Milliseconds at room temperature in purified diamond |
| Useful ratio | Coherence time divided by gate time gives roughly the number of gates available |
Where the idea comes from
Measuring coherence times began in nuclear magnetic resonance in the 1940s and 1950s. Quantum computing made them a headline metric, and superconducting qubit coherence improved from about a nanosecond in 1999 to hundreds of microseconds by the 2010s and over a millisecond in the 2020s.
Real-World Use Cases
- A key benchmark for comparing different quantum hardware platforms
- Determines how many gate operations can realistically be performed in a circuit
- Trapped-ion and neutral atom systems generally report longer coherence times than superconducting qubits
How it differs from related ideas
- T1 is the energy relaxation time, and T2 is the phase coherence time, which is at most twice T1.
- Coherence time describes a physical qubit, while logical qubits can have much longer effective lifetimes after error correction.
- Gate time determines how much computation fits inside the coherence time.
Frequently asked questions
What is coherence time?
The time a qubit can keep its quantum state, specifically its superposition phase, before noise destroys it.
How long do qubits stay coherent?
From microseconds to hundreds of microseconds for superconducting qubits and from seconds to minutes for trapped ions.
Is longer coherence always better?
Yes, but only relative to the gate speed and error rates. A long coherence time with slow, noisy gates may not give a better computer.
Related Terms
Keep exploring
- Algorithm: Quantum Error-Correcting Codes (Shor Code, Surface Code)
- Hardware: IBM Heron (r1, r2, and r3)
- Hardware: Quantinuum H2