Millikelvin digital-to-analog converter for superconducting quantum processors

This paper demonstrates a superconducting digital-to-analog converter (DAC) operating at millikelvin temperatures that uses single-flux-quantum pulses to provide deterministic, in-situ tuning of high-coherence fluxonium qubits, thereby addressing wiring and heat load constraints in scaling superconducting quantum processors.

The Gist

The Big Problem: The "Cable Mess"

Imagine you are trying to control a massive orchestra of 100,000 musicians (quantum bits, or "qubits") who are playing in a room so cold that it's colder than outer space (millikelvin temperature).

Right now, to control each musician, you have to run a separate, thick cable from the warm control room outside all the way down to the cold room. If you have 100,000 musicians, you need 100,000 cables.

• The Heat Issue: All those cables carry heat. If you plug in 100,000 cables, the cold room gets too warm, and the musicians stop playing.
• The Space Issue: There simply isn't enough room in the fridge to fit that many cables.

The Solution: A "Digital Remote Control" Inside the Fridge

The researchers built a new device called a Millikelvin Digital-to-Analog Converter (DAC). Think of this as a tiny, super-fast "remote control" that lives inside the cold room, right next to the musicians.

Instead of running a new cable from the outside for every single adjustment, you send a stream of digital "clicks" (called SFQ pulses) down a single wire. The remote control inside catches these clicks and translates them into a smooth, steady signal to tune the musicians.

How It Works: The "Staircase" Analogy

The device works like a digital staircase that stays put without needing electricity to hold it there.

1. The Digital Clicks (SFQ Pulses): The control room sends a digital signal. Imagine this as a person tapping a button.
2. The Translation: Inside the cold device, each tap moves a "step" on a staircase. This staircase is made of superconducting loops (circuits with zero resistance).
3. The Persistent Signal: Once you tap the button to move up a step, the staircase stays there. It doesn't need power to hold its position. It creates a steady, invisible magnetic force (flux) that gently pushes the qubit to the exact frequency it needs to be.
4. The Result: You can tune the qubit precisely using just a few digital wires, rather than hundreds of heavy analog cables.

The Experiment: Testing the "Remote"

The team tested this by attaching their new "remote" to a specific type of quantum musician called a fluxonium qubit.

• The Test: They used the remote to tune the qubit up and down, checking if the qubit could still hold its tune (coherence) while being controlled this way.
• The Result: It worked perfectly. The qubit didn't get "noisy" or lose its memory. The digital remote was just as gentle and precise as the old, heavy cables.
• The Benefit: They proved they could tune the qubit without needing a dedicated cable from the outside for every single adjustment.

Why This Matters for the Future

Currently, building a quantum computer with millions of qubits is impossible because we can't fit the cables.

This new device is like a universal adapter. It allows engineers to:

1. Tune the qubits locally inside the fridge using digital signals.
2. Fix manufacturing errors: Just like you might tune a guitar string slightly tighter or looser to match the others, this device can adjust each qubit individually to make them all act the same, even if they were built slightly differently.
3. Scale up: Because you don't need a million cables, you can eventually build quantum computers with millions of qubits without the fridge overheating or running out of space.

In short: They built a tiny, digital "dial" that lives inside the super-cold computer, allowing them to tune the quantum bits precisely without needing a massive, heat-carrying mess of wires from the outside.

Technical Summary

1. Problem Statement

Scaling superconducting quantum processors to the fault-tolerant level (requiring >1 million qubits) faces a critical bottleneck known as the "wiring bottleneck."

• Thermal Load & Complexity: In conventional architectures, every qubit requires dedicated control lines running from room temperature (RT) electronics to the millikelvin (mK) stage. As qubit counts increase, the sheer number of cables creates an unsustainable heat load that exceeds the cooling power of dilution refrigerators and creates spatial congestion.
• Static Parameter Variability: Superconducting qubits (specifically flux-tunable ones like fluxonium) suffer from fabrication-induced variations in frequency and coupling. Precise static tuning (flux biasing) is required to homogenize qubit parameters, but current methods rely on individual RT DC bias lines, exacerbating the wiring problem.
• Need for Cryogenic Control: There is an urgent need for control hardware that operates directly at the mK stage to reduce wiring complexity and heat load, while maintaining compatibility with high-coherence qubits.

2. Methodology

The authors propose and demonstrate a Superconducting Digital-to-Analog Converter (DAC) integrated directly into the quantum processor architecture.

• Architecture:

  • SFQ-Programmable: The DAC is programmed using Single-Flux-Quantum (SFQ) pulses. SFQ logic is ideal for cryogenic control due to its ultrafast switching, low power dissipation, and natural compatibility with superconducting circuits.
  • Mechanism: The DAC utilizes an rf-SQUID (Superconducting Quantum Interference Device) with a storage inductor ($L$) and a Josephson junction ($I_c$). SFQ pulses are injected into the SQUID loop, inducing a quantized change in magnetic flux ($\Phi_0$). This creates a persistent current that biases the qubit without requiring continuous power or external signals.
  • Multi-Chip Module (MCM): To optimize fabrication, the device is split into two chips:
    1. Qubit Chip: Contains high-coherence fluxonium qubits fabricated on sapphire.
    2. Control Chip: Contains the SFQ logic, dc/SFQ converters, and DAC circuits fabricated on silicon.
    • These are integrated via flip-chip bonding with a $\sim$10 $\mu$m gap for inductive coupling.

• Operation:

  • A global bias line provides a coarse flux offset.
  • SFQ pulses are routed through a demultiplexing network to specific DACs.
  • Each pulse shifts the DAC state by one flux quantum, allowing deterministic, stepwise tuning of the qubit's operating point.

3. Key Contributions

1. First Integration of SFQ DACs with Fluxonium Qubits: The paper demonstrates the first successful integration of a programmable, persistent-flux DAC with high-coherence fluxonium qubits in an MCM architecture.
2. Digital Interface for Static Control: It establishes a digital interface (SFQ pulses) for static analog control (flux bias), bridging the gap between dynamic gate operations (also SFQ-driven) and static parameter tuning.
3. Elimination of RT DC Bias Lines: The system eliminates the need for individual room-temperature DC bias lines for each qubit, replacing them with a shared global bias line and a digital SFQ routing network.
4. Non-Volatile Control: The DAC state is stored persistently in the superconducting loop, meaning no power is dissipated to maintain the qubit's bias point, only during the infrequent programming events.

4. Key Results

• DAC Performance:
  • Resolution: The DAC achieves a minimum programmable step of 4.8 $\pm$ 0.6 m$\Phi_0$ (milli-flux quanta).
  • Range: The device covers a range of approximately 0.7 $\Phi_0$, corresponding to ~146 discrete steps.
  • Determinism: The DAC responds linearly to SFQ pulses, with each pulse adding or subtracting exactly one flux quantum.
• Qubit Coherence Preservation:
  • Dephasing ($T_2$): Measurements of Ramsey and Echo dephasing times showed no measurable degradation when the qubit was biased via the DAC compared to conventional RT bias lines.
  • Noise Levels: Extracted flux noise amplitudes ($A_\Phi \approx 6.75 - 10.47 \, \mu\Phi_0/\sqrt{\text{Hz}}$) are consistent with state-of-the-art fluxonium qubits without on-chip DACs.
  • Relaxation ($T_1$): Energy relaxation times remained stable ($\sim$82 $\mu$s) across different DAC states, confirming that the DAC does not introduce new dissipation channels.
• Scalability: The architecture supports a demultiplexing tree where $N$ DACs can be addressed using only $\log_2(N)$ control lines, drastically reducing wiring complexity.

5. Significance and Outlook

• Scalability: This work provides a fundamental building block for digitally controlled superconducting quantum processors. By moving static control to the cryogenic stage, it solves the wiring and heat load issues that currently limit scaling to millions of qubits.
• Parameter Homogenization: The ability to locally and programmatically tune qubit frequencies allows for the correction of fabrication variations. This is essential for implementing parallel and multiplexed gate operations, where many qubits must be driven simultaneously with identical parameters.
• Unified Control Stack: The combination of SFQ-based dynamic gate control and DAC-based static configuration paves the way for a fully digital, cryogenic control stack. This unifies fast operations and slow calibration within a single superconducting electronics platform.
• Future Applications: The authors suggest this technology can be extended to control microwave drive amplitudes via tunable couplers, further enhancing the flexibility of in-situ calibration for large-scale quantum systems.

In summary, this paper demonstrates a critical step toward scalable quantum computing by replacing bulky, heat-generating room-temperature bias lines with a compact, non-volatile, and digitally programmable superconducting DAC that operates seamlessly at millikelvin temperatures without compromising qubit coherence.