Dynamically protected erasure qubit via low-frequency charge driving

In the race to build a practical quantum computer, scientists face a stubborn paradox. To make calculations, they must control delicate quantum bits, or qubits, with extreme precision. Yet, the very act of controlling them often introduces the noise that destroys the information they hold. This noise often comes from the environment in the form of low-frequency electrical fluctuations, a background hum that is nearly impossible to eliminate in solid-state materials. For years, researchers have tried to shield their qubits from this interference, often by designing them to be less sensitive to charge. However, this approach usually comes at a cost: it makes the qubits slower to control or harder to connect to one another. The challenge has been to find a way to protect the quantum information from this low-frequency noise without sacrificing the speed and connectivity required for complex computing.

A team of researchers at the University of California, Berkeley, and collaborating institutions has found a clever solution by changing how they drive the system. Instead of trying to block the noise or make the qubits completely immune, they used a strong, rapid electrical drive to effectively lock the qubits in place, a technique known as spin-locking. They demonstrated this using a specific type of superconducting circuit called a Kerr oscillator, which acts as a quantum bit. By applying a charge drive at a frequency just below one gigahertz, they discovered a unique property: the circuit becomes almost blind to the slow, low-frequency noise that usually causes errors, while remaining highly responsive to the fast drive signals needed for operations. This approach allowed them to reset their quantum bits in just 82 nanoseconds, cooling them to a state with less than 0.7 percent of the unwanted energy that typically lingers in these systems. Furthermore, by encoding a single logical bit across two of these oscillators, they created a dual-rail qubit that is naturally biased toward a specific type of error. This means that when an error does occur, it is much more likely to be a detectable erasure rather than a silent corruption of data, a distinction that makes correcting mistakes significantly easier.

The core of this breakthrough lies in how the frequency of the drive signal interacts with the circuit. The researchers found that the sensitivity of their quantum oscillator to charge noise does not stay constant; instead, it grows rapidly as the frequency of the drive increases. Specifically, the sensitivity scales with the square of both the drive's frequency and its strength. This relationship is crucial because the harmful noise in their environment follows a pattern where low frequencies are much stronger than high frequencies. By driving the system at a frequency near one gigahertz, the researchers pushed the system into a regime where it is strongly coupled to their control signals but remains remarkably insensitive to the low-frequency noise that usually plagues these devices. They verified this by measuring how the energy levels of the oscillator shifted under different drive conditions, observing that the shift increased quadratically with the drive parameters. This confirmed that the system was effectively filtering out the slow, disruptive fluctuations while staying open to the fast, useful commands.

To prove this concept works in practice, the team first focused on the fundamental task of resetting the quantum bit. In quantum computing, starting with a clean slate is essential, but thermal energy often leaves the system in a mixed state. Using the low-frequency charge drive, they were able to rapidly swap the energy from the quantum bit into a lossy resonator, effectively dumping the heat. This process took only 82 nanoseconds and reduced the population of excited states to below 0.7 percent. This result was significant because it was lower than the natural thermal equilibrium of the system, which sat at 2.5 percent. The speed and efficiency of this reset demonstrated that the drive could manipulate the system with high fidelity without introducing the excess errors that often accompany such rapid operations.

Building on this, the researchers encoded a logical qubit using two of these oscillators, creating what is known as a dual-rail qubit. In this setup, the information is stored in the presence of a single photon shared between the two circuits. If one of the circuits loses its photon, the system falls into a ground state that is easily detectable, rather than scrambling the information into an unknown state. This property, called an erasure bias, is highly desirable because it turns a hard-to-detect error into a known failure that can be discarded or corrected. The team showed that their dynamical protection scheme enhanced this bias by a factor of four. When they performed random sequences of logical operations, known as Clifford gates, the error rate per gate was measured at 5.6 × 10⁻⁴. After filtering out the runs where an erasure was detected, the error rate dropped even further to 1.5 × 10⁻⁴. These numbers were achieved with gate times of just 25 nanoseconds, indicating that the protection mechanism does not slow down the computation.

The success of this experiment suggests that the trade-off between noise protection and operational speed can be overcome by exploiting the frequency dependence of the system's response. The researchers noted that while the raw error rate is currently limited by the natural lifetime of the oscillators, the potential for improvement is substantial. They pointed out that the current setup uses only a single check at the end of a sequence to detect erasures. By adding a third oscillator to act as an ancilla for mid-circuit checks, they expect the error rates to drop even further, potentially resolving events where the system decays and then heats up again. This work provides a clear path forward for using sub-gigahertz charge drives to protect quantum information, offering a robust method for building high-fidelity quantum processors that can operate effectively in the noisy environment of a solid-state chip. The findings confirm that dynamical protection is not just a theoretical possibility but a practical tool that can be integrated into existing superconducting architectures to enhance their performance and reliability.