Lattice-Renormalized Tunneling Models for Superconducting Qubit Materials

This paper introduces a lattice-renormalized formalism for configurational tunneling two-level systems that overcomes prior model limitations by capturing lattice distortions, thereby enabling accurate computation of tunnel splittings for hydrogen-based defects in bcc Nb and providing critical insights for mitigating decoherence in superconducting qubits.

The Gist

Here is an explanation of the paper "Lattice-Renormalized Tunneling Models for Superconducting Qubit Materials," translated into simple, everyday language with creative analogies.

The Big Picture: Why Are Quantum Computers "Noisy"?

Imagine you are trying to balance a spinning top on a table. You want it to spin perfectly for a long time so you can do math with it. But, the table is shaking, and there are tiny, invisible marbles rolling around underneath it, bumping into the top. Every time a marble hits the top, the top wobbles and loses its spin.

In the world of superconducting quantum computers (the kind used for the most advanced calculations), the "spinning top" is the qubit (the basic unit of information). The "shaking table" is the environment, and the "invisible marbles" are defects in the materials used to build the computer.

These defects are called Two-Level Systems (TLS). They are tiny atoms (usually Hydrogen) that get stuck in the metal lattice and can "tunnel" (jump) between two spots. When they jump, they make noise that ruins the computer's calculations.

The Old Way of Looking at the Problem

Scientists have been trying to predict how these Hydrogen atoms behave to figure out how to stop them. They used two main methods, but both had a major flaw:

1. The "Minimum Path" Method: Imagine trying to walk from one side of a mountain to another. The old method assumed the atom always takes the shortest, straightest path down the valley. But in quantum mechanics, atoms are weird; they don't always take the shortest path. They might take a shortcut through a tunnel.
2. The "Rigid Lattice" Method: This method assumed the metal atoms (Niobium) around the Hydrogen were frozen in place, like a statue. It treated the Hydrogen as a light particle bouncing in a rigid cage.

The Problem: Both methods were wrong because they ignored the fact that the metal cage isn't rigid. When the Hydrogen atom tries to jump, the metal atoms around it actually squish and stretch to help it. It's like trying to jump across a river while standing on a trampoline; the trampoline moves with you. The old models ignored the trampoline.

The New Solution: The "Lattice-Renormalized" Model

The authors of this paper (Pritchard and Rondinelli) built a new model that treats the metal cage and the Hydrogen atom as a team.

The Analogy: The Dance Floor
Imagine the Hydrogen atom is a dancer trying to switch from one side of a dance floor to the other.

• Old Model: The dancer jumps while the floor stays perfectly flat.
• New Model: The dancer jumps, but the floor is made of springs. As the dancer moves, the springs stretch and compress, creating a wave that actually helps the dancer move faster and easier.

The new model introduces a "Composite Phonon Coordinate." In plain English, this is a mathematical way of describing how the entire neighborhood of metal atoms moves together to help the Hydrogen atom tunnel.

What They Found

By using this new "Dance Floor" model on Hydrogen trapped in Niobium (a metal used in quantum computers), they discovered:

1. The Metal Helps the Jump: The movement of the metal atoms (the lattice) significantly changes how fast the Hydrogen jumps. The old models overestimated how hard it was to jump because they forgot the metal was helping.
2. It's Not Just One Jump: Sometimes, these defects aren't just two spots (Two-Level Systems). They can be clusters of three or four spots (Multi-Level Systems). The new model can handle these complex "dance parties" where the atom has many places to go.
3. Strain Matters: If you stretch or squeeze the metal (strain), it changes the energy of the jump. This means that how the metal is manufactured (how smooth or stressed it is) directly controls how much noise the quantum computer makes.

Why This Matters for You

This research is a roadmap for building better quantum computers.

• Better Materials: Now that we know the metal atoms move to help the Hydrogen jump, engineers can design materials that prevent this movement. For example, they might add other atoms to "glue" the metal lattice down so it can't stretch as easily.
• Quieter Computers: By understanding exactly how these defects work, scientists can reduce the "noise" (decoherence). This means quantum computers can stay stable longer and solve harder problems.
• Fixing the Math: The paper proves that previous calculations were wrong because they ignored the "trampoline effect" of the metal. Now, we have the correct math to predict how these materials will behave.

The Takeaway

Think of this paper as realizing that to stop a leak in a boat, you don't just look at the hole (the Hydrogen atom); you have to look at how the wood around the hole flexes (the metal lattice). By understanding that the wood flexes, you can build a boat that doesn't leak, leading to a quantum computer that actually works.

Technical Summary

Here is a detailed technical summary of the paper "Lattice-Renormalized Tunneling Models for Superconducting Qubit Materials" by P. G. Pritchard and J. M. Rondinelli.

1. Problem Statement

Superconducting qubits suffer from decoherence primarily due to parasitic interactions with configurational two-level systems (TLS). These TLS arise from nearly degenerate atomic configurations (e.g., hydrogen atoms trapped at defect sites) separated by small energy barriers.

• Limitations of Existing Models: Current methods for estimating TLS tunnel splittings rely on two flawed approaches:
  1. Minimum-Energy-Path (MEP): Uses the Nudged Elastic Band (NEB) algorithm and a 1D Schrödinger equation. It often underestimates transition frequencies because the MEP is not necessarily the most efficient tunneling path.
  2. Light-Particle Approximation: Treats the tunneling atom as a light particle in a rigid, symmetrized lattice. This approach is thermodynamically unstable; the symmetrized structures required for calculation often possess formation energies far exceeding the thermal energy available in quantum circuits (<100 mK), leading to unphysical results.
• The Gap: There is a lack of a rigorous model that incorporates lattice degrees of freedom (phonons) while maintaining thermodynamic stability, leading to inaccurate predictions of tunnel splittings and excitation spectra.

2. Methodology

The authors introduce a lattice-renormalized formalism derived directly from the nuclear Hamiltonian.

• Theoretical Framework:

  • The model reduces the Born-Oppenheimer nuclear Hamiltonian to essential degrees of freedom.
  • It introduces a composite phonon coordinate ($Q$) that continuously transforms the host lattice between adjacent degenerate configurations.
  • The potential energy surface $V(q, Q)$ is sampled on a multidimensional grid, where $q$ represents hydrogen phonon coordinates and $Q$ represents the collective lattice distortion.
  • This approach explicitly couples the hydrogen defect to a collective lattice mode, capturing anharmonic couplings that are neglected in rigid-lattice models.

• Numerical Implementation:

  • System: Applied to hydrogen (H) and deuterium (D) trapped by defects in body-centered cubic (bcc) Niobium (Nb). Specific defects studied include interstitial Oxygen (O), and substitutional Titanium (Ti), Zirconium (Zr), and Tantalum (Ta).
  • DFT Calculations: Used Density Functional Theory (DFT) with the PBE functional to compute the potential energy surface on a 4D grid ($q_x, q_y, q_z, Q$).
  • Solver: Solved the Schrödinger equation using a finite difference Hamiltonian on a subgrid, employing the Implicitly Restarted Lanczos Method to obtain eigenvalues and wavefunctions.
  • Validation: Results were benchmarked against experimental specific heat and inelastic neutron scattering data.

3. Key Contributions

1. Rigorous Formalism: Developed a 4D Hamiltonian model that avoids the thermodynamic instability of symmetrized structures by allowing the lattice to relax dynamically during tunneling.
2. Defect Site Identification: Systematically calculated formation energies to identify the thermodynamically favored trapping sites for H in Nb:
   • O-trapped H: Favors a specific 7th nearest-neighbor tetrahedral site (binding energy $\approx$ -72.5 meV).
   • Ti/Zr-trapped H: Favors the 1st nearest-neighbor tetrahedral site.
   • Ta-trapped H: Repels H, making it an unlikely TLS source.
3. Extension to Multi-Level Systems (MLS): Generalized the formalism to 5D Hamiltonians to model Four-Level Systems (FLS) arising from substitutional defects with high site degeneracy (24 sites), which are relevant for Zr-doped Nb.
4. Critique of Machine Learning Potentials (MLPs): Highlighted that while MLPs can reproduce experimental tunnel splittings, this agreement may be accidental due to error cancellation, as MLPs often fail to capture the subtle atomic displacements critical for tunneling.

4. Key Results

• Tunnel Splittings ($J$):
  • The 4D lattice-renormalized model produces tunnel splittings that bound experimental values from below.
  • For O-trapped H (c=1/54), the computed $J$ is 0.064 meV, compared to the experimental range of 0.17–0.23 meV.
  • In contrast, the rigid-lattice (3D) model overestimates $J$ (0.57 meV), confirming that neglecting lattice distortion leads to significant errors.
  • The inclusion of the composite coordinate $Q$ reduces the tunnel splitting by an order of magnitude compared to the light-particle approximation, bringing it closer to experimental reality.
• Dimensionality Effects:
  • Tunnel splitting is highly sensitive to anharmonic couplings. Reducing the model to 1D or 3D subspaces yields values significantly higher than the full 4D calculation.
  • The 4D model captures the "bending" of the tunneling path through the lattice, which lowers the effective barrier height.
• Mass and Strain Effects:
  • Mass Scaling: Tunnel splitting decreases significantly when moving from lighter (H) to heavier (D) isotopes or when the lattice mass is effectively increased (e.g., Ta substitution).
  • Strain Coupling: The model reveals that local strain fields (via elastic dipole tensors) strongly modulate TLS energetics. While strain can quench TLS in simple 2-level systems, Multi-Level Systems (MLS) remain active over a broader range of strain fields, posing a persistent decoherence risk.
• TLS Density: The estimated TLS density for interstitial H in Nb is orders of magnitude higher ($\sim$60 eV$^{-1}$ nm$^{-3}$) than in amorphous dielectrics, suggesting H is a dominant source of loss in Nb-based qubits.

5. Significance and Implications

• Materials Design: The study provides a predictive framework for designing superconducting qubit materials. It suggests that strain engineering and controlling film homogeneity are critical to suppressing TLS-related loss.
• Defect Mitigation: Identifies that while O-trapped H is a major concern, Ti and Zr impurities create complex multi-level systems that are harder to suppress via strain alone.
• Theoretical Advancement: Establishes a direct link between TLS dynamics and phonon-mediated strain interactions, moving beyond the static "light particle" view.
• Experimental Guidance: The finding that lattice-renormalized models provide a lower bound to experimental splittings suggests that further refinement (e.g., including more phonon modes) could achieve quantitative agreement, guiding future experimental and theoretical efforts.

In summary, this paper resolves long-standing discrepancies in TLS modeling by rigorously incorporating lattice dynamics, demonstrating that the interaction between tunneling atoms and lattice phonons is the dominant factor in determining tunnel splittings and decoherence rates in superconducting qubits.