Short answer: Mid-circuit measurement reads out chosen qubits while a circuit is still running, without destroying the qubits that are not measured. In trapped-ion machines the readout uses laser-induced fluorescence, and stray light can disturb neighbouring ions. Published papers report a micromotion technique that cuts this decoherence by over an order of magnitude, and a benchmarking protocol that found and removed previously undetected crosstalk on a 20-qubit device. These are research results from specific machines.
What it is and why it matters
A 2025 Nature Communications paper describes high-fidelity mid-circuit measurements as reading out the state of specific qubits in a multiqubit processor without destroying them or disrupting their neighbours. The authors call them a critical component for useful quantum computing, enabling fault-tolerant quantum error correction and dynamic circuits. They also note that few methods exist to assess these measurements comprehensively.
The crosstalk problem in trapped ions
A 2021 Physical Review A paper explains the difficulty. Measurement and reset need strong interactions with selected qubits while the neighbours stay isolated. In trapped-ion systems, measurement uses laser-induced fluorescence, and stray light from the detection beam, plus fluorescence from the measured ions, can be significant sources of decoherence for unmeasured qubits.
Micromotion as a fix
The same paper presents a technique that uses ion micromotion to reduce these sources of decoherence by over an order of magnitude. The authors benchmark it with a method based on randomized benchmarking, which estimates crosstalk errors on nearby qubits. Using the Honeywell System Model H0, they report measurement and reset on selected qubits with low crosstalk errors on neighbouring qubits.
Benchmarking the errors
The 2025 paper introduces what its authors call the first randomized benchmarking protocol that measures the rate at which mid-circuit measurements induce errors in many-qubit circuits. They report using it to detect and eliminate previously undetected measurement-induced crosstalk in a 20-qubit trapped-ion computer, a Quantinuum H1-1 system. They then apply the same protocol to a 27-qubit IBM Q processor and quantify how much of the error is removed by dynamical decoupling. The results are specific to the devices tested.
Using fewer qubits
A 2023 Physical Review X paper describes a different benefit. The authors say trapped-ion and superconducting machines can now perform mid-circuit measurements and resets, which can reduce the number of qubits needed to run many algorithms by measuring qubits early, resetting them and reusing them. They introduce qubit-reuse compilation, with an exact constraint programming model and a greedy heuristic, and show that optimal reuse needs the same number of qubits as the circuit's dual. They test it on circuits such as time-evolution circuits and the quantum adiabatic optimization algorithm. These are numerical benchmarks.
Limits
These are technical papers on small systems. They describe measurement quality and compilation, not completed error-corrected computation. Figures such as the order-of-magnitude reduction belong to the specific setups in each paper.
FAQ
What is mid-circuit measurement?
Reading out specific qubits during a computation without destroying them or disrupting their neighbours.
Why is it hard in trapped ions?
Laser-induced fluorescence readout can scatter stray light and fluorescence onto unmeasured ions, which the 2021 paper names as sources of decoherence.
How was crosstalk reduced?
The 2021 paper reports an ion micromotion technique that cut these sources of decoherence by over an order of magnitude.
How do researchers measure the errors?
A 2025 randomized benchmarking protocol measures the error rate that mid-circuit measurements induce in many-qubit circuits.
Can it save qubits?
A 2023 paper says measuring, resetting and reusing qubits can reduce the number of qubits an algorithm needs, and offers compilation methods to do it.
All field notes