Active-Learning Inspired $\textit{Ab Initio}$ Theory-Experiment Loop Approach for Management of Material Defects: Application to Superconducting Qubits

Abstract

Surface oxides are associated with two-level systems (TLSs) that degrade the performance of niobium-based superconducting quantum computing devices. To address this, we introduce a predictive framework for selecting metal capping layers that inhibit niobium oxide formation. Using DFT-calculated oxygen interstitial and vacancy energies as thermodynamic descriptors, we train a logistic regression model on a limited set of experimental outcomes to successfully predict the likelihood of oxide formation beneath different capping materials. This approach identifies Zr, Hf, and Ta as effective diffusion barriers. Our analysis further reveals that the oxide formation energy per oxygen atom serves as an excellent standalone descriptor for predicting barrier performance. By combining this new descriptor with lattice mismatch as a secondary criterion to promote structurally coherent interfaces, we identify Zr, Ta, and Sc as especially promising candidates. This closed-loop strategy integrates first-principles theory, machine learning, and limited experimental data to enable rational design of next-generation materials.

Figures (6)

• Figure 1: Diffusion barrier performance versus diffusion barrier energy: barrier energy (position along horizontal axis), performance (green=good, red=poor) for Au, Pd, W, Ta, Mo, Al, Pt, and Zr from left to right. Data show no clear correlation between barrier energies and performance.
• Figure 2: Oxide vacancy energy versus metal interstitial energy for a set of metals. Darker green regions indicate increasingly favorable thermodynamic barriers to oxygen diffusion
• Figure 3: Oxygen vacancy energy vs. metal interstitial energy across logistic regression iterations: first iteration (a), final iteration (b). Experimentally tested metals are outlined in green (success) or red (failure). Point colors reflect predicted oxide formation probabilities. The solid blue line shows the model’s decision boundary; the dashed line indicates constant oxide formation energy.
• Figure 4: Per-oxygen oxide formation energy plotted against oxygen vacancy and metal interstitial energies for all calculated metals. Point colors indicate predicted oxide formation probabilities from the logistic regression model. A best-fit plane is shown, viewed at an angle that nearly reduces it to a line.
• Figure 5: Per-oxygen oxide formation energy of various experimentally tested oxides, colored according to their success (green) or failure (red) to prevent oxidation of the niobium layer