🔬 materials science

Substrate-metal interface engineering enhances TaN/Ta thin film superconducting resonator performance

This study demonstrates that inserting a thin tantalum buffer layer between the silicon substrate and a tantalum nitride seed layer significantly enhances the internal quality factor of superconducting resonators by suppressing nitrogen accumulation and structural disorder at the interface, thereby reducing two-level system losses.

Original authors: Moritz Singer, Harsh Gupta, Benedikt Schoof, Elena Willinger, Anton Orekhov, Alexandra Schewski, Samuil Kostadinov, Philip Constantin Schneider, Marc Tornow

Published 2026-07-27

📖 3 min read☕ Coffee break read

CC BY 4.0

Original authors: Moritz Singer, Harsh Gupta, Benedikt Schoof, Elena Willinger, Anton Orekhov, Alexandra Schewski, Samuil Kostadinov, Philip Constantin Schneider, Marc Tornow

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 build a super-fast, super-quiet library where the books are made of light. In this library, the "books" are tiny packets of energy called photons, and the "shelves" are circuits made of special metals that can conduct electricity without any resistance at all. This is the world of quantum computing, where scientists are building "artificial atoms" to solve problems that would take normal computers forever. To make these artificial atoms work, they need a very quiet environment. If the shelves (the circuits) are too rough or dirty, the light-books get lost or confused, and the computer stops working. Scientists call these tiny glitches "two-level system losses," or TLS for short. Think of TLS as tiny, invisible ghosts in the walls of the library that whisper to the light, causing it to scatter. The goal of this research is to build the smoothest, quietest shelves possible so the ghosts can't hide.

The scientists in this paper were investigating a specific type of metal called Tantalum Nitride (TaN). Imagine Tantalum as a very strong, reliable brick. When you mix it with Nitrogen (a gas), you get Tantalum Nitride, which can be a bit like a shape-shifter: sometimes it acts like a super-conductor (letting electricity flow perfectly), and sometimes it doesn't. The researchers wanted to see if they could use this shape-shifting material to build better shelves for their quantum library. They tested different ways of stacking layers of Tantalum and Tantalum Nitride on top of a silicon base (the floor of the library) to see which combination made the quietest, most efficient shelves.

Here is what they found: When they built a shelf using just Tantalum Nitride, or when they put a layer of Tantalum on top of a thin Tantalum Nitride base, the shelves were decent but still had a few ghosts. The "quality factor" (a score measuring how well the shelf holds the light) was about 1.5 out of 1,000,000. This was okay, but not great. However, when they added a secret ingredient—a thin, 10-nanometer layer of pure Tantalum acting as a buffer between the floor and the Tantalum Nitride—the performance skyrocketed. The quality factor jumped to nearly 5.9 out of 1,000,000. That is almost four times better!

Why did this happen? The team used powerful microscopes to look at the floor where the metal met the silicon. They discovered that when Tantalum Nitride touches the silicon directly, the meeting point is messy and chaotic, like a construction site where the bricks and the floor have smashed together in a disorderly way. This mess creates the "ghosts" (the TLS losses) that ruin the performance. But when they put that thin Tantalum buffer layer in between, it acted like a peacekeeper. It stopped the Tantalum Nitride from touching the silicon directly, resulting in a clean, orderly interface. The paper suggests that by keeping the nitrogen-containing metal away from the silicon floor, they eliminated the source of the chaos. While they didn't prove exactly how the ghosts work at a microscopic level, the evidence strongly points to the idea that a clean, well-engineered boundary between the metal and the floor is the key to building better quantum computers.

Technical Summary: Substrate-Metal Interface Engineering in TaN/Ta Superconducting Resonators

Problem Statement
While tantalum (Ta) has emerged as a promising material for superconducting qubits, comparatively little attention has been paid to tantalum nitride (TaN). Tantalum nitride exhibits a wide range of stoichiometries, offering both superconducting (delta-TaN) and non-superconducting phases, which makes it versatile for use as seed layers, base materials, or barrier layers in Josephson junctions. However, the performance of superconducting CPW resonators made from TaN, or Ta grown on TaN seed layers, has not been thoroughly investigated. A critical unknown is how the interface between these metal films and the silicon substrate affects the internal quality factor ( $Q_i$ ), particularly regarding two-level system (TLS) losses which dominate in the single-photon regime.

Methodology
The study employed a systematic approach to deposit and characterize thin film stacks on high-resistivity silicon (100) substrates using RF-magnetron sputtering. Three specific stack configurations were fabricated:

Standalone TaN: A 200 nm thick TaN film.
TaN Seed with Ta: A 5 nm TaN seed layer followed by 200 nm of alpha-phase Ta.
Buffered Stack: A 10 nm Ta buffer layer, followed by a 5 nm TaN seed layer, and finally 200 nm of alpha-phase Ta.

The films were characterized using:

Surface Morphology: Atomic Force Microscopy (AFM) to measure root mean square roughness ( $S_q$ ).
Crystallography: Grazing Incidence X-ray Diffraction (GI-XRD) to identify crystal phases (alpha-Ta vs. cubic delta-TaN).
Electrical Properties: Four-point resistivity measurements and critical temperature ( $T_c$ ) sweeps in a cryostat.
Structural/Compositional Analysis: Cross-sectional Transmission Electron Microscopy (TEM) and Electron Energy Loss Spectroscopy (EELS) to examine interface disorder and elemental distribution (specifically nitrogen presence).
RF Performance: Coplanar Waveguide (CPW) resonators were fabricated and measured at 100 mK. Internal quality factors ( $Q_i$ ) were extracted via circle fits of S21 transmission data across varying input powers to model TLS losses.

Key Results

Material Properties: All stacks exhibited superconducting transitions. The standalone TaN film showed a $T_c$ of ~5 K, while the Ta-on-TaN stacks showed a $T_c$ of ~4.1 K. The TaN film exhibited high disorder (RRR < 1), whereas the Ta-containing stacks showed lower disorder. Surface roughness was low for all samples (0.4–0.9 nm).
Resonator Performance:
- Standalone TaN and TaN/Ta: Both configurations yielded similar internal quality factors in the single-photon regime ( $Q_i \approx 1.5 \times 10^5$ ). The TaN/Ta stack performed similarly to the standalone TaN despite the bulk Ta layer being alpha-phase, suggesting the loss mechanism is not determined by the bulk superconducting film.
- Buffered Stack (Ta/TaN/Ta): The introduction of a 10 nm Ta buffer layer between the Si substrate and the TaN seed layer resulted in a significant performance enhancement. The $Q_i$ increased to $5.9 \times 10^5$ in the single-photon regime, representing an approximately four-fold improvement over the other stacks.
Interface Analysis:
- TaN/Si Interface (No Buffer): TEM and EELS revealed structural disorder, lattice defects, and a direct extension of nitrogen to the silicon interface. This correlates with the lower $Q_i$ values.
- Ta/Si Interface (With Buffer): The Ta buffer layer successfully separated the nitrogen-containing TaN layer from the silicon. TEM images showed an undisturbed interface lattice structure with no visible disorder or defects at the Si boundary. This correlates with the significantly higher $Q_i$ .

Key Contributions

Performance Benchmarking: The study establishes that standalone superconducting TaN films achieve $Q_i$ values of $\sim 1.5 \times 10^5$ , and that Ta grown on TaN seed layers yields comparable performance, indicating that the bulk properties of the top Ta layer are not the limiting factor.
Interface Engineering: The primary contribution is the demonstration that inserting a thin (10 nm) Ta buffer layer between the silicon substrate and the TaN seed layer drastically improves resonator performance.
Mechanism Identification: By correlating RF performance with structural analysis, the authors identify that the dominant loss mechanism in TaN-based stacks is likely TLSs originating from structural disorder and nitrogen-related defects at the metal-substrate interface. The buffer layer suppresses these defects by preventing direct contact between reactive TaN and the silicon substrate.

Significance
The paper concludes that controlling the chemical composition and structural integrity of the metal-substrate interface is pivotal for the performance of superconducting qubit circuitry. The findings suggest that for transition-metal nitrides like TaN, the interface with the substrate is a critical source of TLS losses. By engineering this interface—specifically by preventing nitrogen incorporation at the silicon boundary via a buffer layer—researchers can achieve substantial improvements in the internal quality factor of superconducting resonators. This supports the hypothesis that interface engineering is a necessary step for optimizing superconducting devices based on tantalum nitride.

Technical Summary: Substrate-Metal Interface Engineering in TaN/Ta Superconducting Resonators

More like this