Simultaneous operation of an 18-qubit modular array in germanium

This paper demonstrates the simultaneous initialization, control, and readout of an 18-qubit modular array in germanium with high-fidelity single-qubit gates and entangling operations, establishing a scalable architecture for utility-scale semiconductor quantum processors.

The Gist

Imagine you are trying to build a massive orchestra to play a symphony of the future. In the world of quantum computing, each musician is a qubit (a quantum bit). The problem is that for a long time, building an orchestra with more than a handful of musicians was a nightmare. If you added too many, the instruments started interfering with each other, the conductor couldn't hear everyone, and the music fell apart.

This paper is like a breakthrough announcement from a group of engineers who have finally built a 18-musician orchestra that plays perfectly in sync, all while using a new, highly efficient design.

Here is the story of how they did it, explained simply:

1. The Problem: The "Cable Chaos"

In the past, trying to control a large group of quantum bits was like trying to conduct an orchestra where every single musician needs their own dedicated cable running from the conductor's podium to their instrument. As you add more musicians, you get a tangled mess of wires that is impossible to manage. This is called the "wiring bottleneck."

2. The Solution: The "Modular Lego" Design

The researchers (from Groove Quantum and QuTech in the Netherlands) didn't try to build one giant, complicated machine. Instead, they built using modules.

Think of their design like LEGO bricks.

• They created a standard "unit cell" (a small brick) that holds 6 qubits (musicians).
• This brick has its own built-in microphone (a charge sensor) to listen to the musicians.
• They stacked three of these bricks together to make their 18-qubit device.

Because the bricks are identical and modular, if they want to build a 100-qubit orchestra in the future, they just need to snap more bricks onto the end. They don't have to redesign the whole thing.

3. The Material: "Silicon's Cool Cousin"

Most quantum computers use silicon (like your phone's chip). This team used Germanium, which is like silicon's cooler, more flexible cousin.

• The Analogy: Imagine trying to balance a spinning top on a rough table (silicon) versus a perfectly smooth, frictionless glass table (germanium). The Germanium allows the quantum "spins" (the tops) to spin longer and more stably without wobbling out of control.

4. The Magic Trick: "Parallel Play"

Usually, in these small quantum experiments, you have to tune one musician, then the next, then the next. It's like a conductor saying, "Okay, violinist, play. Now, stop. Flutist, play." This is slow.

This team achieved simultaneous operation.

• The Analogy: They turned on the lights for the entire orchestra at once. They initialized (waked up), controlled (gave instructions), and read out (listened to) all 18 musicians at the exact same time.
• Why it matters: In the past, adding more musicians made the setup time explode. Here, adding more musicians doesn't make the setup slower because every "brick" does its own job in parallel.

5. The Results: A Perfect Performance

The team didn't just get the musicians to play; they played beautifully.

• High Fidelity: They achieved a 99.8% accuracy rate for single-note instructions. In the quantum world, this is like hitting a note perfectly 998 times out of 1,000.
• The "GHZ" State: They managed to entangle three qubits (create a "Greenberger-Horne-Zeilinger" state).
  • The Analogy: Imagine three musicians who, once they start playing, are so magically connected that if you change the note of one, the others change instantly, even if they are on opposite sides of the room. They proved they could create this spooky connection across their array.

6. The Big Picture: Why This Matters

This paper is a blueprint for the future. It proves that we can scale up quantum computers from tiny experiments (a few qubits) to utility-scale machines (thousands of qubits) without losing control.

By using Germanium (the smooth table) and a Modular Design (the LEGO bricks), they have shown a clear path to building a quantum computer that is big enough to solve real-world problems, like designing new medicines or cracking complex codes, without getting tangled in a mess of wires.

In short: They built a scalable, modular, high-fidelity quantum orchestra using Germanium, proving that we can finally start conducting the symphony of the future.

Technical Summary

1. Problem Statement

The realization of utility-scale quantum computing requires hardware architectures capable of scaling to thousands of qubits while efficiently managing wiring and control overhead. While semiconductor spin qubits are a leading candidate due to their compatibility with existing industrial manufacturing, scaling them has proven difficult. Previous demonstrations in silicon and germanium were largely limited to:

• Small-scale arrays (typically 4–12 qubits).
• Sequential, individual qubit operations rather than parallel control.
• Non-modular, bespoke device layouts that do not easily extend to larger grids.
• A lack of experimental proof that high-fidelity control and readout can be maintained simultaneously across a large, integrated two-dimensional grid.

2. Methodology

The authors developed a scalable, modular quantum processing unit (QPU) based on germanium (Ge) spin qubits embedded in a Ge/SiGe heterostructure.

• Architecture: The device features a 2×N extendable architecture. The 18-qubit array is composed of three repeated 2×3-qubit unit cells. Each unit cell includes a dedicated charge sensor (single-electron transistor) capable of reading out all quantum dots within that specific cell.
• Device Layout: The qubits are arranged in a planar grid with a locally one-dimensional gate fan-out, allowing the architecture to be extended to arbitrary lengths. The device utilizes overlapping gate stacks (ohmic contacts, screening, plunger, and barrier gates) and operates at ~14 mK.
• Control Strategy:
  • Parallelism: Initialization, control, and readout are performed simultaneously across all unit cells. Within a cell, operations are sequential to avoid crosstalk between adjacent vertical pairs, but the three cells operate in parallel.
  • Gate Virtualization: A four-layer virtualization scheme is employed to mitigate control crosstalk, compensating for sensor influence, orthogonalizing plunger gates, normalizing charging energies, and compensating for barrier gate cross-talk.
  • Magnetic Field: The device operates at a specific magnetic field orientation ($B_x=0.83$ mT, $B_y=50$ mT, $B_z=10$ mT) to place qubits near the "sweet spot" for minimal hyperfine interaction.
• Measurement Protocol:
  • Readout: Utilizes Pauli Spin Blockade (PSB) via charge sensors. The system reads out the parity of vertical double quantum dot (DQD) pairs.
  • Calibration: An automated calibration routine tracks daily fluctuations, performing sensor tuning, Rabi optimization, and frequency alignment.

3. Key Contributions

• Largest Spin-Qubit Array: Demonstrated the largest spin-qubit array to date (18 qubits) with simultaneous operation.
• Modular Scalability: Proved the viability of a modular, extendable unit-cell architecture where the State Preparation and Measurement (SPAM) overhead does not increase with system size.
• Simultaneous High-Fidelity Control: Achieved high-fidelity single-qubit gates and parallel readout across the entire 2D array without degrading performance.
• Two-Qubit Entanglement: Successfully implemented controlled-Z (CZ) gates and generated a three-qubit Greenberger-Horne-Zeilinger (GHZ) state, demonstrating connectivity in both dimensions of the array.

4. Key Results

• Single-Qubit Performance:
  • Gate Fidelity: Achieved an average single-qubit gate fidelity of 99.8% and a median of 99.9% across all 18 qubits.
  • Coherence: Measured average dephasing time ($T_2^*$) of 6.2 µs and coherence time with dynamical decoupling ($T_2^{CPMG}$) of 0.47 ms.
  • Uniformity: The array showed consistent g-factors and driving efficiencies, though some spread was attributed to strain fluctuations.
• Parallel Operation:
  • Successfully initialized, controlled, and read out all 18 qubits simultaneously.
  • Rabi oscillations showed high visibility in parallel operation, comparable to individual qubit operation, with only minor reductions attributed to charge-sensor peak broadening.
  • Crosstalk analysis confirmed low interference between qubits during simultaneous X-gate application.
• Two-Qubit Gates & Entanglement:
  • Exchange Coupling: Characterized nearest-neighbor exchange couplings with an average tunability of 19 ± 2.9 mV/decade.
  • CZ Gate: Implemented a high-fidelity Controlled-Z gate between qubits Q9 and Q10 using adiabatic voltage pulses at the zero detuning point.
  • GHZ State: Generated a three-qubit GHZ state using qubits Q7, Q9, and Q10 (spanning both horizontal and vertical directions), verified by parity oscillation measurements.

5. Significance

This work represents a critical milestone in the path toward fault-tolerant quantum computing using semiconductor spin qubits.

• Scalability Blueprint: It establishes a blueprint for scaling 2D semiconductor quantum processors by proving that modular unit cells can be tiled to create larger arrays without sacrificing control fidelity or increasing measurement overhead.
• Industry Compatibility: By leveraging the mature infrastructure of the semiconductor industry (Ge/SiGe heterostructures), this approach offers a viable pathway to mass-produce large-scale quantum processors.
• Performance Benchmark: The combination of high gate fidelities (>99.8%) and simultaneous multi-qubit operation in a 2D array sets a new benchmark for spin-qubit systems, moving beyond small-scale demonstrations to utility-scale architectures.

The authors conclude that by combining this modular architecture with advanced manufacturing and cryogenic control electronics, a viable path to fault-tolerant quantum computation based on semiconductor spin qubits has been established.