Real-Time FPGA-Based Multi-Parameter Feedback Stabilization of a Silicon Double Quantum Dot

Imagine you are trying to balance a stack of Jenga blocks on a table that is constantly shaking. If the table vibrates too much, the tower collapses. Now, imagine that instead of wooden blocks, you are trying to balance a single electron inside a tiny, invisible cage made of electricity. This is the world of quantum computing, specifically using "spin qubits"—particles that act like tiny magnets to store information. The problem is that the world around these particles is noisy. Tiny, invisible charges in the materials nearby jitter and shift, creating a low-frequency "hum" (known as $1/f$ noise) that pushes the electron out of its perfect spot. If the electron drifts, the computer loses its memory. To build a useful quantum computer, scientists need to keep these electrons perfectly still, even while the world around them is shaking. They need a way to sense the drift and instantly push the electron back to the center, faster than the noise can move it.

This is exactly what the researchers at UCLA set out to do. They built a high-speed "autopilot" for a silicon double quantum dot—a device that traps two electrons side-by-side. Previous attempts to stabilize these dots were like trying to steer a race car using a map that updates once every few minutes; the car would have already crashed by the time the map changed. The team in this paper replaced that slow map with a super-fast, real-time GPS. By using a special chip called an FPGA (Field-Programmable Gate Array) and a technique called RF-reflectometry (which listens to the electron's "voice" using radio waves instead of waiting for a slow electrical current), they created a feedback loop that is thousands of times faster than before. They found that this new system can suppress noise up to a frequency of 5 kHz, effectively silencing the "hum" that usually ruins the experiment. This doesn't mean they solved quantum computing yet, but it proves that we can now keep the electron's home stable long enough to potentially do complex calculations in the future.

The Jenga Tower on a Shaking Table

Think of a silicon double quantum dot as a very delicate, two-room house built for electrons. Inside this house, the walls are made of invisible electric fields. For the house to work as a quantum computer, the electrons need to stay in a very specific configuration, like a perfect Jenga tower. But the ground beneath this house is shaky. Tiny, random charges in the surrounding materials are constantly shifting, creating a low-frequency "charge noise." This noise is like a slow, invisible hand that keeps nudging the walls of the house, causing the electrons to drift away from their perfect spots. If they drift too far, the quantum information they hold gets scrambled and lost.

To fix this, scientists use a technique called "feedback stabilization." It's like having a robot that constantly checks the position of the Jenga tower and makes tiny adjustments to the walls to keep it balanced. In the past, this robot was slow. It would check the tower, calculate a correction, and then apply it, but the process took so long (about 0.3 Hz, or once every few seconds) that the tower would have already wobbled significantly by the time the robot acted. The researchers in this paper wanted to build a robot that could move at the speed of thought.

The Super-Fast Autopilot

The team, led by Johnathan Bryan, Tim J. Wilson, and Hong-Wen Jiang, built a new kind of autopilot using a device called an OPX by Quantum Machines. This device contains a powerful chip (an FPGA) that can do math incredibly fast. Instead of waiting for a slow electrical current to tell them where the electrons are, they used a method called RF-reflectometry.

Imagine you are in a dark room trying to find a specific object. Instead of walking over and touching it (which is slow), you throw a ball at it and listen to the echo. If the object moves, the echo changes. The researchers did this with radio waves. They bounced microwaves off the quantum dot and listened to the "echo" (the reflected signal). This allowed them to sense the position of the electrons almost instantly.

The process works like a game of "hot and cold," but played at lightning speed:

1. Listen: The system measures the "echo" to see where the electrons are.
2. Nudge: It gently pushes one of the electric walls (a gate voltage) by a tiny amount to see how the echo changes.
3. Calculate: It figures out which direction to push to get the electrons back to the center.
4. Correct: It instantly adjusts the voltage to push the electrons back.

They did this for two different "walls" (two parameters) at the same time, creating a multi-parameter feedback loop.

The Results: Silencing the Hum

The team tested their new system by introducing a fake, rhythmic shaking (a sinusoidal perturbation) to see if the autopilot could stop it. They found that their system was incredibly effective.

• Speed: They could perform a full cycle of measuring, calculating, and correcting in just 81.8 microseconds (that's 0.0000818 seconds).
• Noise Suppression: The system successfully reduced the noise by 6 dB (a significant drop in volume) for frequencies up to 5 kHz.
• Comparison: Previous methods using slow current measurements could only handle noise up to 0.3 Hz. The new method is roughly 16,000 times faster in terms of the frequency range it can handle.

The paper explicitly notes that while this is a huge improvement, it is not a magic wand that solves everything yet. The system is currently limited by how fast they can measure the signal. If they could make the measurements even clearer (better signal-to-noise ratio), they could potentially go even faster. The theoretical limit for their current setup is around 6.1 kHz, based on the laws of physics known as the Nyquist-Shannon sampling theorem.

Why This Matters

The authors are careful to point out that they cannot run a quantum computer while this feedback is active. The act of constantly adjusting the voltages to stabilize the dot would destroy the delicate quantum state needed for computing. Instead, this system acts like a "parking brake" or a "stabilizer" that keeps the device in a perfect, ready state. Once the device is stabilized, the feedback can be paused, and the quantum computation can happen.

This work suggests that we are getting closer to the day when we can build large arrays of quantum dots that stay stable for long periods. By taming the low-frequency noise that has plagued these systems for years, this high-speed feedback loop is a crucial step toward making quantum computers that don't just work for a split second, but can run complex calculations reliably. The paper concludes that the next step is to test how this "interleaved" approach (stabilizing, then computing, then stabilizing again) affects the actual quality of the quantum information, but for now, they have proven that the "autopilot" works, and it works fast.