⚛️ quantum physics

A Compact Broadband Purcell Filter for Superconducting Quantum Circuits in a 3D Flip-Chip Architecture

This paper presents a compact, fabrication-tolerant four-pole broadband Purcell filter implemented on a 3D flip-chip platform that enables high-fidelity multiplexed qubit readout by providing strong resonator-feedline coupling while suppressing qubit decay, supported by an analytical model for rapid design optimization.

Original authors: Zhen Luo, Lea Richard, Ivan Tsitsilin, Anirban Bhattacharjee, Christian M. F. Schneider, Stefan Filipp, Amelie Hagelauer

Published 2026-07-02

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Zhen Luo, Lea Richard, Ivan Tsitsilin, Anirban Bhattacharjee, Christian M. F. Schneider, Stefan Filipp, Amelie Hagelauer

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

Imagine you are trying to listen to a very quiet whisper (a quantum bit, or "qubit") in a room that is also filled with a roaring vacuum cleaner (the measurement equipment).

The Problem: The "Leaky" Whisper
In the world of superconducting quantum computers, scientists need to "read" the state of a qubit very quickly. To do this, they connect the qubit to a listening device (a feedline) through a special tunnel (a resonator).

The Catch: To hear the whisper clearly, the tunnel needs to be wide open. But if the tunnel is too wide, the whisper doesn't just get heard; it leaks out and disappears before it can be measured. This leakage is called the Purcell effect. It's like trying to listen to a secret in a room with the windows wide open; the sound escapes too fast, and the qubit "dies" (loses its information) prematurely.

The Solution: The "Smart Soundproof Door"
The authors of this paper built a special device called a Purcell Filter. Think of this filter as a smart, soundproof door placed between the whispering qubit and the roaring vacuum cleaner.

How it works: This door is designed to be wide open only for the specific frequency of the "whisper" (the readout signal), allowing it to pass through clearly. However, for the "leakage" frequency (where the qubit would decay), the door slams shut, blocking the sound completely.
The Result: The qubit can be read quickly and accurately without losing its energy to the outside world.

The Innovation: A "Highway" for Multiple Whispers
Previous versions of these filters were like single-lane roads. They worked well but were narrow, meaning you could only listen to one qubit at a time, or you had to pack them very tightly together, which caused traffic jams (crosstalk).

The New Design: The team created a four-lane highway (a four-pole broadband filter).
- Wide Bandwidth: This highway has a massive "passband" (a flat, open lane) of 1 GHz. This is huge in the quantum world. It allows them to fit six different qubits (represented by six floating readout resonators) on this single highway, all talking at once without bumping into each other.
- Strong Protection: Even with this wide highway, the filter is incredibly effective at blocking the "leakage" noise. It suppresses unwanted decay by more than 45 dB (which is like turning a shout into a whisper).

The "3D Flip-Chip" Trick
Instead of drawing all these circuits on a flat piece of paper (a 2D chip), the team used a 3D flip-chip approach.

The Analogy: Imagine building a house with two floors. The "ground floor" (bottom chip) holds the main highway (feedline), and the "second floor" (top chip) holds the houses (resonators). They are stacked directly on top of each other, connected by tiny metal bumps (like pillars).
Why it helps: This stacking makes the whole device much more compact. It's like folding a large map into a small pocket. This saves space, which is crucial when you want to build a quantum computer with thousands of qubits.

The "Magic Blueprint" (Analytical Model)
Designing these circuits usually requires running massive, slow computer simulations for every tiny change.

The Breakthrough: The authors created a mathematical blueprint (an analytical model). This is like having a formula that can predict exactly how the sound will travel through the door just by looking at the dimensions of the wood and hinges, without needing to build a full-scale prototype first.
Benefit: This allows engineers to design and optimize these filters much faster. They can also use this model to predict how the qubits will behave just by looking at the physical shape of the chip.

Real-World Testing
The team built a physical chip using a special metal (Niobium) and tested it in a super-cold fridge (at 20 millikelvin, colder than outer space).

The Outcome: The experiment worked. The filter successfully let six signals pass through clearly while blocking the noise that would kill the qubits.
The "Glitch": They noticed the signals were slightly lower in pitch than expected. They traced this to the "kinetic inductance" of the metal (a subtle quantum property of the superconducting material), which they successfully accounted for in their math.

Summary
This paper presents a compact, high-speed "soundproof door" for quantum computers. It solves the trade-off between reading qubits fast and keeping them alive long enough to be read. By using a 3D stacking method and a wide "highway" design, it paves the way for building larger, more complex quantum processors that can handle many qubits at once without them interfering with each other.

Technical Summary: A Compact Broadband Purcell Filter for Superconducting Quantum Circuits in a 3D Flip-Chip Architecture

Problem Statement
Superconducting quantum processors face a fundamental design trade-off between achieving fast, high-fidelity qubit readout and maintaining long qubit lifetimes ( $T_1$ ). Rapid state discrimination requires strong coupling between the readout resonator and the feedline; however, this strong coupling inevitably enhances qubit decay via the Purcell effect. While Purcell filters are standard solutions to suppress this decay, existing implementations for multiplexed readout architectures face significant limitations. Traditional approaches often rely on individual half-wavelength filters for each resonator, demanding precise frequency matching and occupying large footprints that hinder scalability. Alternatively, common single-pole filters integrated into the feedline offer protection but suffer from narrow passbands (e.g., ~200 MHz), restricting the number of multiplexed channels and forcing tight frequency spacing that risks crosstalk. Furthermore, previous multi-pole broadband designs have either required large footprints or lacked experimental verification in three-dimensional (3D) architectures, which are increasingly envisioned for large-scale quantum processors.

Methodology and Design
To address these challenges, the authors present a four-pole broadband Purcell filter implemented on a 3D flip-chip platform. The design utilizes a compact $10 \times 6 \text{ mm}^2$ chip fabricated with a 150 nm Niobium (Nb) thin-film process. The physical architecture consists of two bulk silicon substrates separated by a 10 µm gap defined by SU-8 spacers, with metal layers patterned on both chips.

The filter topology comprises four coupled resonators:

Input/Output Coupling: Compact spiral resonators act as the input and output coupling elements, tapped to a 50 $\Omega$ feedline.
Inner Resonators: Two unfolded quarter-wavelength ( $\lambda/4$ ) coplanar waveguide (CPW) transmission-line resonators form the core of the filter.
Coupling Mechanism: The spiral and CPW resonators are coupled via "over-the-air" parallel-plate capacitance at their open ends. The two inner CPW resonators are inductively coupled through a short section of parallel-coupled lines.
Multiplexed Readout: Six floating readout resonators are coupled within the filter's passband to the inner CPW resonators, emulating a realistic multiplexed readout system.

The authors developed a comprehensive analytical model to guide the design. This model treats the system as a network of coupled parallel LC resonators, represented by a coupling matrix. It allows for the extraction of self-coupling coefficients (based on resonator geometry and kinetic inductance), mutual coupling coefficients (derived from even-odd mode analysis of coupled lines and parallel-plate capacitance), and external coupling coefficients (based on the tapped feedline position). This framework enables rapid circuit synthesis and the estimation of resonance frequencies and external quality factors ( $Q_e$ ) directly from physical geometry.

Key Results
The fabricated device was characterized at 20 mK in a cryogenic setup. The experimental results demonstrate:

Broadband Performance: The filter achieves a flat 1 GHz passband centered at 7.68 GHz (fractional bandwidth of 13.2%).
Purcell Suppression: The device provides more than 45 dB of suppression at typical qubit frequencies (~4.5 GHz), effectively protecting qubits from radiative decay.
Multiplexed Readout: Six floating readout resonators were successfully coupled within the passband with approximately 100 MHz spacing. The measured transmission spectrum ( $S_{21}$ ) closely tracks both the analytical model and full-wave electromagnetic simulations (Ansys HFSS) after compensating for system losses.
Parameter Extraction: The resonance frequencies and external quality factors extracted from the measurement data align with the model and simulations, with minor deviations attributed to the kinetic inductance of the Nb thin film and fabrication tolerances in the chip spacing.

Significance and Claims
The paper claims that this work provides a practical, compact, and fabrication-tolerant solution for large-scale superconducting quantum processors. By integrating a multi-pole broadband filter directly into a 3D flip-chip architecture, the design overcomes the bandwidth limitations of single-pole filters and the footprint constraints of individual filters. The authors emphasize that the "over-the-air" capacitive coupling strategy offers high tolerance to fabrication variations, particularly chip-spacing nonuniformities inherent in the bonding process.

The development of the analytical model is highlighted as a key contribution, enabling rapid design iteration and optimization without relying solely on time-consuming full-wave simulations. While the current work focuses on the filter and readout resonators without a fully integrated qubit, the authors state that the design serves as a scalable and robust foundation for suppressing Purcell decay in multiplexed readout architectures. Future work is identified as integrating this filter with actual qubit devices to validate performance in realistic quantum processor environments.

More like this