Low-loss Nb on Si superconducting resonators from a dual-use spintronics deposition chamber and with acid-free post-processing

This paper demonstrates that high-quality, low-loss niobium superconducting resonators can be fabricated in a dual-use chamber shared with magnetic materials by employing an acid-free resist strip process that achieves internal quality factors near one million, thereby enabling the integration of superconducting and magnetic systems without compromising device performance.

The Gist

Imagine you are trying to build a super-sensitive radio receiver that operates at the very edge of reality, where electricity flows without any resistance at all. This is the world of superconducting circuits, which are the brains behind the most advanced quantum computers.

To make these circuits work perfectly, scientists usually need a pristine, "sterile" factory. They are terrified of magnetic impurities (tiny bits of magnetic dust like iron or nickel) because, much like a speck of dust ruining a diamond, these impurities can ruin the superconducting "magic," causing the circuit to lose energy and fail.

For a long time, the rule was: Never use a machine that has ever made magnetic materials to build these delicate superconducting circuits. You needed a dedicated, separate factory just for the superconductors.

The Big Discovery
This paper tells the story of a team that broke that rule and found out it didn't matter. They took a machine that had been used for over 20 years to make magnetic materials (like the magnets in hard drives) and used it to build high-quality superconducting circuits.

Here is how they did it, explained through simple analogies:

1. The "Shared Kitchen" Analogy

Think of the deposition chamber (the machine that sprays the metal onto the silicon) as a kitchen.

• The Old Rule: If you cooked spicy, smelly curry (magnetic materials) in a kitchen, you could never use that same kitchen to bake a delicate, flavorless soufflé (superconductors) because the smell would ruin the taste.
• The New Approach: The team decided to clean the kitchen very thoroughly. They scrubbed the walls, swapped out the utensils, and even "seasoned" the oven with a layer of the new ingredient before starting.
• The Result: They proved that even though the kitchen had cooked "curry" for decades, after a good cleaning, they could bake a soufflé that tasted just as good as one baked in a brand-new, dedicated kitchen. Their measurements showed no detectable magnetic dust in the final product.

2. The "Surface Cleaning" Secret Sauce

While the machine cleaning was impressive, the real magic happened in how they cleaned the surface of the metal after they built the circuit.

Imagine you just painted a wall. You have to peel off the tape (the "resist") used to keep the paint in the right shape.

• The Old Way: They used a standard solvent (called "1165") to peel off the tape. But this left behind a sticky, invisible residue (like chlorine) that made the wall rough. To fix this, they had to use a strong acid (Hydrofluoric acid, or HF) to scrub the wall clean.
• The New Way: They tried a different, specialized solvent (called "AZ"). This solvent was like a "magic eraser" that didn't just peel the tape off; it also dissolved the sticky residue and dirt while it was peeling.
• The "Acid-Free" Breakthrough: Because the "magic eraser" solvent did such a good job, they didn't need the harsh acid scrub at the end. This is huge because:
  • Some materials (like the special junctions in quantum computers) get eaten away by acid.
  • Acid is dangerous and creates environmental hazards.
  • By skipping the acid, they got results that were just as good, or even better, than the acid method.

3. The "Silicon Prep" Experiment

The team also tried three different ways to prepare the silicon "floor" before painting it:

1. BOE: A quick chemical dip.
2. Anneal: Heating the floor to 700°C to make it smooth.
3. Thermal: A complex process of growing and removing a layer of oxide.

The Surprise: It didn't matter which floor preparation they used. The quality of the final circuit was almost identical across all three. This suggests that the cleaning of the surface (the solvent choice) is far more important than how you prepare the floor.

The Bottom Line

This paper proves two major things:

1. Shared Tools Work: You don't need a billion-dollar, dedicated factory to make top-tier quantum circuits. You can use a "dual-use" machine that also makes magnetic materials, provided you clean it well. This makes quantum research cheaper and more accessible.
2. Skip the Acid: By choosing the right cleaning solvent, you can make these circuits without using dangerous acids, which is safer for the environment and allows for using materials that acid would destroy.

In short, they showed that with the right cleaning routine, you can build world-class quantum computers in a "shared kitchen" without ruining the recipe.

Technical Summary

Technical Summary: Low-loss Nb on Si Superconducting Resonators from a Dual-use Spintronics Deposition Chamber and with Acid-free Post-processing

Problem Statement
Magnetic impurities are known to degrade superconductivity by suppressing critical temperatures, introducing flux noise, and inducing sub-gap density of states that lead to internal losses. Consequently, physical vapor deposition chambers previously used for magnetic materials are typically avoided for fabricating high-quality superconducting resonators. Furthermore, the specific environmental threshold at which magnetic contamination impacts device performance has not been established. Additionally, standard post-fabrication processing for Niobium (Nb) often involves acid treatments (e.g., HF) which can damage other materials in hybrid systems, such as conventional Josephson junctions, and pose health and environmental risks regarding fluorine-based byproducts.

Methodology
The authors investigated the feasibility of using a high-vacuum magnetron sputtering chamber, primarily used for magnetic depositions (Fe, Co, Ni) for over 20 years, to synthesize high-quality Nb films on high-resistivity Si(100) substrates.

• Chamber Hygiene: Two deposition procedures were tested:
  • DP1: The Nb target was co-loaded only with other superconducting targets.
  • DP2: The Nb target was co-loaded with magnetic targets (permalloy, Co, CoFeB) that had been recently used.
  • Both procedures involved rigorous cleaning, including bead-blasting, solvent sonication, and "seasoning" the chamber with Nb and Ti to coat internal components.
• Substrate Preparation: Three methods were evaluated to optimize the Si-Nb interface:
  1. BOE: Buffered Oxide Etch dip immediately prior to loading.
  2. Anneal: BOE dip followed by an in-situ flash anneal at 700°C.
  3. Thermal: Removal of the top ~100 nm of the wafer via thermal oxide growth and subsequent etching, followed by the flash anneal.
• Fabrication: 60 nm Nb films were sputtered at 2.4 Å/s. Devices were patterned using photolithography and chlorine-based dry etching.
• Post-Processing & Stripping: A critical variable was the resist strip bath. Two baths were compared:
  • 1165: MICROPOSIT 1165 resist remover.
  • AZ: Integrated Micro Materials AZ 300T stripper.
  • Post-fabrication treatments included dips in 2% HF (18–60 seconds) or 10:1 BOE (20 minutes), as well as aging studies.
• Characterization: Films were analyzed via Secondary-Ion Mass Spectrometry (SIMS), Atomic Force Microscopy (AFM), and X-ray Photoelectron Spectroscopy (XPS). Microwave resonators (3 µm gap CPW) were characterized for internal quality factors ($Q_i$) and loss ($\delta_i$) at single-photon powers across six different cryogenic setups.

Key Contributions and Results

1. Dual-Use Chamber Viability:

   • SIMS analysis revealed no detectable magnetic impurities (Fe, Co, Ni, Cr) in the bulk of the Nb films or at the substrate-metal interface, even for films deposited under DP2 (co-loaded with magnetic targets).
   • Trace amounts of Fe and Cr were detected only at the metal-air interface, likely due to surface oxide or handling, and were not consistent across all films.
   • Devices fabricated in the "magnetically contaminated" chamber achieved low-power internal quality factors ($Q_i$) near one million, comparable to state-of-the-art devices.

2. Impact of Resist Strip Baths:

   • 1165 Bath: Required a post-fabrication HF dip to achieve state-of-the-art performance. Without HF, losses were ~15 times higher. XPS indicated that 1165 left significant Chlorine (Cl) and Nitrogen residues on the Si surface.
   • AZ Bath: Successfully removed Cl and other residues and reduced surface carbon content without requiring an HF dip.
   • Acid-Free Performance: Using the AZ bath without any post-fabrication acid treatment yielded devices with median low-power losses ($\delta_{LP}$) of $1.52 \times 10^{-6}$ (DP1 and DP2 combined). This performance is comparable to 1165-stripped devices that did receive an HF dip.
   • Optimized Acid-Free: A 20-minute post-BOE treatment on AZ-stripped samples further reduced losses to a median of $0.48 \times 10^{-6}$, outperforming standard HF treatments. However, aging studies showed these losses increased over 4–6 weeks, returning to the range of untreated samples, suggesting regrowth of oxides or contamination.

3. Substrate Preparation:

   • The three substrate preparation methods (BOE, Anneal, Thermal) did not yield clearly distinct differences in device quality. The authors suggest the metal-substrate interface is a subdominant source of low-power loss for these devices, or that the Nb-Si interface is robust against these specific preparation variations.

4. Material Analysis:

   • Nb films exhibited superconducting transition temperatures ($T_c$) > 9.1 K and Residual Resistance Ratios (RRR) > 4.5.
   • XPS analysis showed that while Nb sub-oxides were present, the AZ bath resulted in higher Nb pentoxide content compared to other methods, yet still achieved superior performance, indicating that Cl residue is a more critical factor than relative oxide stoichiometry in this context.

Significance and Claims
The paper claims to demonstrate that:

• Chamber Constraints are Surmountable: High-quality superconducting resonators can be fabricated in deposition chambers shared with magnetic material processing, provided rigorous cleaning and seasoning protocols are followed. This allows for broader and more cost-effective materials exploration without dedicated superconducting-only systems.
• Acid-Free Processing is Viable: A specific resist strip bath (AZ300T) can provide a superior surface chemical template that eliminates the need for aggressive acid post-processing (HF) while maintaining or improving device performance. This is significant for hybrid systems where acids might damage other components (e.g., Josephson junctions) and for reducing health/environmental risks associated with fluorine-based nanofabrication.
• Surface Chemistry is Critical: The choice of solvent and the removal of specific residues (like Chlorine) are more impactful on resonator quality than the specific substrate preparation method or the presence of magnetic targets in the chamber.

The authors conclude that these results enable the community to innovate on superconducting materials for quantum information science using existing, multi-purpose equipment, while highlighting the importance of solvent selection in fabrication workflows.