Google tests dynamic surface codes on Willow - what changed inside its quantum chip

Key details

Google's experiment implements three dynamic surface code circuits on Willow: hexagonal, walking, and iSWAP. Each circuit alters the pattern of detecting regions that localize physical errors in space and time, instead of repeating a single static circuit each error correction cycle.

• Hexagonal circuit
  • The hexagonal circuit uses three couplers per qubit by alternating two cycle types within the error correction sequence.
  • One coupler is used twice per cycle, producing overlapping detecting regions that still allow error triangulation.
  • On Willow's square lattice, unused couplers were disabled to emulate hexagonal connectivity while running this circuit.
  • As code distance increased from three to five, the logical error rate improved by a factor of 2.15.
  • Google reports that this performance matched a traditional static surface code circuit operated on the same hardware.
  • Simulations showed that a hexagonal lattice with three couplers per qubit reduced frequency optimization complexity and improved the error suppression factor by about 15 percent compared with a square lattice.
• Walking circuit
  • The walking circuit periodically swaps data and measure qubit roles, allowing standard leakage reset to apply to every qubit.
  • Measurements showed that walking reduced long-range time correlations between detectors by more than an order of magnitude.
  • With walking enabled, the remaining correlations matched those from a static surface code using the DQLR leakage-removal technique.
• iSWAP circuit
  • The iSWAP circuit uses iSWAP gates instead of controlled-Z (CZ) gates to entangle data and measure qubits, building on our prior theory work on iSWAP-based surface code circuits.
  • Google notes that the iSWAP gate does not rely on non-computational states and produces fewer leakage-induced correlated errors.
  • On Willow, the dynamic iSWAP surface code achieved an error suppression factor of 1.56, slightly below the performance of the device's optimized CZ-based circuit.

Background context

Google Quantum AI's processors use superconducting circuits as physical qubits, which are subject to noise and state leakage. Quantum error correction groups many physical qubits into logical qubits that can detect and correct errors during operation. Surface codes are the quantum error correction scheme used in Google's Willow experiments.

In December 2024, Google reported that error correction on Willow had reached below-threshold operation for a logical qubit, in a milestone experiment that used a static surface code circuit with the same set of gates repeated each cycle. The new dynamic surface code experiments let the circuit pattern and detecting-region tiling change from one cycle to the next.

Google describes three dynamic tilings in the latest work: hexagonal, walking, and iSWAP-based configurations. The team states that dynamic circuits help address hardware layout constraints, leakage, and qubit or coupler dropouts. Google links these surface code advances with recent color code operation and a verifiable quantum advantage experiment on Willow.

Earlier posts such as Making quantum error correction work and Overcoming leakage on error corrected quantum processors describe the static surface code implementations and leakage-removal tools that underpin the new dynamic circuits.

Google Quantum AI's public roadmap sets a next milestone of a long-lived logical qubit with fewer than one error per million cycles. The dynamic surface code demonstrations are presented as steps toward that target. All reported metrics and configurations are based on experiments or simulations on the Willow architecture.

Source citations

All statements here are drawn from Google's official materials and cited research outputs. These sources document the dynamic surface code experiments, the Willow architecture, and related quantum error correction work.

• Google Research blog: Dynamic surface codes open new avenues for quantum error correction, January 13, 2026
• Nature Physics article: "Demonstration of dynamic surface codes" (Paper)
• Nature article on the Willow architecture and the original below-threshold milestone experiment: Willow architecture study
• Google Research blog: Making quantum error correction work
• Google Research blog: Overcoming leakage on error corrected quantum processors
• Google Research blog: A verifiable quantum advantage
• arXiv preprint on iSWAP-based surface code circuits: arXiv:2302.02192
• Google Research blog: A colorful quantum future
• Google Quantum AI roadmap: Roadmap to a useful quantum computer