🔬 mesoscale physics

Reservoir- and Measurement-free Microwave Initialization of Semiconductor Spin Qubits

This paper demonstrates a scalable, reservoir- and measurement-free method for initializing silicon spin qubits to a singlet state with 99.4% fidelity using a fixed sequence of microwave and baseband pulses, offering a practical control primitive for large-scale quantum processors.

Original authors: Leon C. Camenzind, Ik Kyeong Jin, Akito Noiri, Kenta Takeda, Takashi Nakajima, Takashi Kobayashi, Seigo Tarucha

Published 2026-10-01

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Leon C. Camenzind, Ik Kyeong Jin, Akito Noiri, Kenta Takeda, Takashi Nakajima, Takashi Kobayashi, Seigo Tarucha

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Building a quantum computer is like trying to build a city where every single house needs its own private power plant and a dedicated security guard just to turn on the lights. For the most promising type of quantum processor, made from tiny silicon chips, this requirement has been a major bottleneck. These processors rely on particles called electrons, specifically their "spin," which acts like a tiny internal compass pointing either up or down. To perform calculations, these compasses must be reset to a known starting position, usually pointing down, before every single operation. Until now, doing this reset required a complex network of wires connecting each tiny electron to a reservoir of extra electrons, or a sensitive sensor to measure the electron's state and tell a computer when to flip it. As scientists try to pack more and more of these qubits onto a chip to make them powerful, running all those extra wires and sensors to every single spot becomes physically impossible. The infrastructure simply takes up too much space and becomes too tangled to manage.

A team of researchers at RIKEN in Japan has found a way to reset these silicon spins without needing those extra wires or sensors. They demonstrated a method that uses only a fixed sequence of microwave pulses and electrical signals already present in the device. Think of the electron spins as a pair of people who can either face the same direction or opposite directions. The researchers wanted to force them to face opposite directions, a state that is essential for starting calculations. In the past, if the pair happened to be facing the same way, the system would get stuck. The new method acts like a clever trick that gently nudges the stuck pair, using microwaves to shake them into a different configuration where they can naturally settle into the correct starting position. This process happens repeatedly and automatically, turning a messy mix of directions into a clean, uniform starting state.

The experiment took place in a device made of silicon and a related material called germanium, fabricated by Intel. The researchers focused on a pair of electrons trapped in a tiny double-well structure. They did not use any external reservoir to swap electrons in or out, nor did they use a sensor to check the result after every attempt. Instead, they relied on the natural physics of how these spins interact with electric charges. When the spins are in a "blocked" state—meaning they are facing the same way and cannot easily move—they are driven by a burst of microwave energy. This energy pushes them into a temporary state where they can mix with other states. Once they mix, they naturally relax into a stable "singlet" state, which is a specific pairing where the spins are opposite. Because the microwaves are tuned to only affect the "wrong" states, the correct state remains untouched. By repeating this cycle of nudging and waiting, the researchers could accumulate the electrons into the desired starting state with a success rate of 99.4 percent.

The team verified their results by checking the final state of the spins using a technique called exchange spectroscopy. This method confirmed that the electrons were not just in the right charge configuration, but were actually in the specific operational state needed for calculations, with one spin up and the other down. The process took about 12 microseconds to complete, which is fast enough for modern quantum computing needs. The researchers also identified the specific physical limits of their current device. They found that the speed of the reset is limited not by the microwave pulses themselves, which are very fast, but by how quickly the electrons can settle into the final state after being nudged. They calculated that with slight improvements to the device, such as better magnetic gradients, this time could be reduced to less than one microsecond.

This work is significant because it removes a major architectural hurdle for scaling up quantum computers. By showing that initialization can be done locally with a fixed sequence of pulses, the researchers proved that qubits deep inside a dense chip do not need their own dedicated reservoir or sensor. This opens the door to packing millions of qubits onto a single chip without the wiring becoming unmanageable. The method relies on fundamental physics that is present in many semiconductor systems, suggesting it could be adapted for other types of silicon-based quantum processors. The researchers have shown that a complex, infrastructure-heavy problem can be solved with a simple, repeatable control sequence, making the path toward large-scale quantum processors much clearer.

Technical Summary: Reservoir- and Measurement-free Microwave Initialization of Semiconductor Spin Qubits

Problem Statement
Scalable semiconductor quantum processors require repeated qubit initialization across large arrays. Current initialization methods face significant architectural constraints:

Reservoir-based initialization: Relies on energy-selective tunneling to a local electron reservoir. While fast, this requires dedicated reservoir access and control gates at every qubit site, consuming routing resources and area, which becomes prohibitive in dense two-dimensional arrays.
Measurement-based feedback: Requires real-time charge sensing and conditional logic to flip spins if the parity is even. This necessitates local charge sensors and fast measurement infrastructure at every qubit, similarly constraining device layout.
Intrinsic relaxation: Spin relaxation times (milliseconds to seconds) are too slow for microsecond-scale cycle times.

The authors identify a need for a "fixed-sequence local reset" that operates without local reservoirs or measurement feedback, enabling initialization in both dense and sparse architectures.

Methodology
The authors demonstrate a microwave-based initialization protocol on a silicon spin-qubit pair (Q2 and Q3) within an industrially fabricated Si/SiGe triple quantum dot device. The protocol utilizes existing microwave and baseband controls, avoiding electron tunneling to reservoirs or measurement-conditioned feedback.

Physical Mechanism: The method exploits the spin-charge structure near the (1, 3)–(0, 4) charge transition.
- Odd parity states ( $|\uparrow\downarrow\rangle, |\downarrow\uparrow\rangle$ ) naturally relax to the singlet charge state $|(0, 4)S\rangle$ via Pauli-spin-blockade (PSB) mediated singlet-triplet mixing.
- Even parity states ( $|\uparrow\uparrow\rangle, |\downarrow\downarrow\rangle$ ) are blocked in the (1, 3) configuration. The protocol pulses the system deep into the (0, 4) charge state, where these blocked states map to polarized triplet states ( $|(0, 4)T_-\rangle, |(0, 4)T_+\rangle$ ).
- Microwave Driving: A resonant microwave pulse drives transitions from the polarized triplets to the $T_0$ -like manifold ( $|(0, 4)T_0\rangle$ ).
- Conversion: The $T_0$ -like population is then converted to the singlet ground state $|(0, 4)S\rangle$ through singlet-triplet mixing and charge hybridization during a mixing dwell time.
- Accumulation: Since the singlet state is no longer resonantly driven by the microwave pulse, repeated cycles accumulate population in the reset state. Finally, the system is diabatically/adiabatically mapped from $|(0, 4)S\rangle$ to the operational state $|\uparrow\downarrow\rangle$ .
Characterization: The authors use a nearby charge sensor (CS) solely for characterization and verification, not for the initialization feedback loop. They employ an eight-state spin-charge Hamiltonian to model the spectrum and a driven three-state triplet model to analyze transfer dynamics.

Key Results

High-Fidelity Initialization: Repeated cycles of the microwave and baseband sequence produce the singlet-associated charge outcome with a median probability of 99.4% across sampled preparation states (standard deviation 0.37%).
Target State Verification: Exchange spectroscopy (Conditional Rotation, CROT) confirms that the initialized state is the specific operational state $|\uparrow\downarrow\rangle$ , rather than a mixture of odd-parity states. No resolved branch corresponding to the opposite odd state ( $|\downarrow\uparrow\rangle$ ) was observed, with hidden resonance contributions limited to <1.08% at 95% confidence.
Dynamics and Limits:
- Dark State Limit: Single-cycle transfer is limited by a "dark state" in the triplet manifold. A coherent resonant pulse can transfer at most 50% of the population from a polarized triplet to the $T_0$ pathway.
- Time Scale: The microwave transfer occurs in tens of nanoseconds, but the subsequent blockade-lifting (conversion of $T_0$ -like states to the singlet) sets the time scale. The effective single-cycle conversion time is approximately 1 µs ( $\tau_{mix} \approx 0.97 \pm 0.05$ µs).
- Total Time: The demonstrated pumping sequence uses approximately 12 µs of microwave bursts and mixing dwells to reach 99% fidelity.
Scalability Projections: The authors project that under improved device conditions (specifically increased Zeeman-frequency difference and improved microwave transfer), sub-microsecond initialization is achievable.

Significance and Claims
The paper establishes fixed-sequence microwave initialization as a scalable control primitive for semiconductor spin-qubit processors. Its primary significance lies in:

Architectural Decoupling: It removes the requirement to colocate every initialization site with a local reservoir or charge sensor, addressing a critical bottleneck for scaling to large, dense two-dimensional arrays.
Platform Independence: The principle relies on singlet-triplet physics within Pauli-spin-blockade transitions, making it adaptable to other semiconductor platforms (e.g., SiMOS electron-spin and hole-spin systems) with analogous transitions.
Operational Efficiency: It achieves high-fidelity initialization (>99%) using only fixed sequences of microwave and baseband pulses, without the overhead of real-time measurement and feedback logic.

The authors note that while the current implementation takes ~12 µs, the physics supports a route to sub-microsecond initialization, and the method provides a flexible alternative to shuttling-based architectures by allowing local reset at operation zones.

More like this