🔬 physics

A Reproducible Workflow for Extracting Quantum Hamiltonians from Surface-Adsorbate Models

This paper presents an open-source, reproducible workflow that bridges density functional theory surface models and quantum algorithms by automatically extracting hydrogen-capped clusters, selecting active spaces, and constructing validated fermionic and qubit Hamiltonians for surface-catalysis systems like IrO2(110) without requiring high-performance computing.

Original authors: Manisha Malhotra, Fabio Rinaldi, Neil Ramarapu

Published 2026-07-10

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Manisha Malhotra, Fabio Rinaldi, Neil Ramarapu

Original paper licensed under CC BY 4.0 (https://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're trying to solve a giant, 3D puzzle of a chemical reaction happening on a metal surface. The pieces are atoms, and the picture is how energy moves when a hydrogen atom sticks to an Iridium oxide surface. For years, scientists have used a powerful tool called Density Functional Theory (DFT) to look at the whole puzzle. It's great for seeing the big picture, but it's like trying to play a video game using a blurry, low-resolution photo of the screen. You can see the shapes, but you can't see the individual pixels needed to run the game.

Now, a new team of researchers has built a "translation machine" that turns that blurry photo into a crisp, playable video game file for quantum computers. Their paper, "A Reproducible Workflow for Extracting Quantum Hamiltonians from Surface–Adsorbate Models," is all about building this bridge.

The Problem: The "Too Big" Puzzle

The main hurdle the authors tackle is a mismatch in languages. Real-world catalysts (like the Iridium oxide used to split water) are huge, periodic sheets of atoms. Quantum computers, however, are currently small and need a very specific, tiny input: a list of rules (called a Hamiltonian) describing just a few key electrons interacting in a small space.

Previously, there was no standard way to go from the giant, messy surface model to that tiny, clean list of rules. It was like having a map of the entire world but needing a single street address to send a letter, and nobody knew how to zoom in without losing the location.

The Solution: The "Zoom and Cap" Workflow

The authors created a step-by-step recipe (a workflow) to solve this. Think of it as a digital chef preparing a complex dish:

Start with the Whole Cake: They begin with a full, optimized model of the catalyst surface (the "slab") created by standard DFT software.
Cut a Slice: They use a tool called ASE to cut out a small, finite "cluster" of atoms right around the spot where the reaction happens (the active site).
Patch the Edges: Cutting the slice leaves some atoms with dangling bonds (like a cake with a jagged edge). To fix this, they stick little "hydrogen caps" onto the cut edges. This is like putting a decorative frosting border on the jagged slice so it doesn't crumble and acts like a complete, stable piece.
Pick the Stars: Not all atoms in the slice are equally important. The workflow automatically scans the slice to find the "frontier" orbitals—the specific electron paths involved in the reaction. They filter out the deep, boring core electrons and focus only on the active ones.
Translate to Quantum: Finally, they convert these electron rules into a format a quantum computer can read (a qubit Hamiltonian).

The Test Drive: Iridium Oxide

To prove their machine works, they tested it on Iridium oxide (IrO₂), a material used in hydrogen production. They looked at two specific spots on the surface where a hydrogen atom (or hydroxyl group) sticks.

The Result: The workflow successfully turned the complex surface models into 14-qubit Hamiltonians (think of this as a game file with 14 switches).
The Size: These files contained 3,382 Pauli terms (the specific rules for how the switches interact).
The Active Space: They focused on an (8e, 7o) active space, meaning they tracked 8 electrons moving across 7 specific orbitals.

Did It Work? (The "Taste Test")

The authors didn't just guess; they ran a rigorous check. They took the quantum "game file" they created and ran a simulation to see if it reproduced the original energy of the system.

The Match: The energy calculated from their new quantum file matched the original computer simulation energy to better than 10⁻⁶ Hartree. That is an incredibly tiny difference, proving their translation was accurate.
The Correlation: They measured the "static correlation energy" (a fancy way of saying how much the electrons wiggle and interact in complex ways).
- At the neutral site, the energy was −0.044 mHa (milli-Hartree).
- At the anionic (negatively charged) site, it was −2.49 mHa.
- The authors note that the anionic site has about 50 times more of this complex interaction, suggesting it's a more challenging, multi-faceted problem for a quantum computer to solve.

What They Explicitly Say "No" To

It is crucial to understand what this paper doesn't do, because the authors are very careful about their claims:

No "All-Electron" Trickery: They explicitly warn against trying to use the def2-SVP basis set (a standard math tool for atoms) on Iridium without its special "Effective Core Potential" (ECP). They argue that trying to treat all 77 electrons of an Iridium atom with this specific basis set is not just an approximation; it is ill-defined. It's like trying to build a skyscraper with a foundation designed for a shed; the math might give you a number, but the building is fundamentally broken.
No Magic Quantum Speedup: They do not claim to have solved the hydrogen production problem or found a new catalyst. They only built the tool to get the data ready for quantum computers.
No Future Hardware Claims: While they mention that their 14-qubit files could run on current quantum processors, they admit that actually running them on real hardware to resolve such tiny energy differences (milli-Hartree scale) is a challenge for future work due to "shot noise" (random errors in the machine).

The Bottom Line

This paper is a "how-to" guide for a very specific, difficult task. The authors have built a reproducible, open-source pipeline that takes a messy, real-world surface model and cleans it up into a format a quantum computer can actually use.

They proved that their method works by showing the numbers match perfectly (< 10⁻⁶ Hartree) and that the resulting files are small enough to run on standard computers without needing a supercomputer. They didn't discover a new law of physics or a new super-catalyst; instead, they handed the quantum computing community a set of keys to finally unlock the door to simulating real-world chemical surfaces.

The authors suggest that while their current results show small correlation energies (meaning the electrons aren't too chaotic in these specific spots), the anionic site shows signs of being more complex. This hints that as we get better at these simulations, we might need to expand the "active space" (look at more electrons) to fully capture the magic of these reactions. But for now, the bridge is built, and the path is clear.

Technical Summary: A Reproducible Workflow for Extracting Quantum Hamiltonians from Surface–Adsorbate Models

Problem Statement
A significant practical barrier exists in applying quantum algorithms to heterogeneous catalysis: the disconnect between realistic surface models generated by Density Functional Theory (DFT) and the specific inputs required by quantum algorithms. While DFT provides total energies and electron densities for periodic slab models, algorithms like the Variational Quantum Eigensolver (VQE) require explicit second-quantized fermionic or qubit Hamiltonians. Current surface science workflows typically output energies or charge distributions but lack a standard, reproducible pipeline to transform a periodic DFT slab into a localized active-space Hamiltonian suitable for quantum simulation. This gap is particularly acute for heavy transition metal catalysts, such as Iridium-based systems, where relativistic effects and variable oxidation states challenge standard DFT approximations.

Methodology
The authors present an open-source, end-to-end workflow that bridges this gap through a sequence of controlled transformations, implemented in Python using ASE, xTB, PySCF, and OpenFermion. The pipeline proceeds as follows:

Slab Optimization: A periodic DFT optimization of the catalyst surface (e.g., using Quantum ESPRESSO or VASP) yields a relaxed slab geometry.
Cluster Extraction: Using the Atomic Simulation Environment (ASE), a finite cluster containing the active site is extracted from the slab within a specific radial cutoff (e.g., 3–6 Å). To handle the severed bonds at the cluster boundary, dangling oxygen atoms are capped with hydrogen atoms (placed at 0.96 Å along the severed bond vector) rather than using fractional-charge pseudo-hydrogens.
Active Space Selection: A semiempirical tight-binding calculation (GFN2-xTB) is performed on the neutral and anionic cluster states to rank molecular orbitals. An automated, deterministic selection process identifies the active space by:
- Restricting candidates to a frontier energy window ( $w \approx 0.15$ Hartree) around the HOMO and LUMO.
- Filtering orbitals based on their fractional weight on the adsorbate and nearest-neighbor surface atoms.
- Selecting the top $n_{occ}$ occupied and $n_{virt}$ virtual orbitals.
Hamiltonian Construction: Restricted Hartree–Fock (RHF) calculations are performed on the selected active space using PySCF. Heavy-element cores (Iridium) are treated with an Effective Core Potential (ECP) (def2-SVP basis with associated ECP), while oxygen and hydrogen are treated all-electron. This yields one- and two-electron integrals defining the fermionic Hamiltonian.
Qubit Mapping: The fermionic Hamiltonian is mapped to a qubit Hamiltonian (using the Jordan–Wigner transformation) via OpenFermion and exported in JSON format.

Key Results
The workflow was applied to H*/OH* adsorption on a rutile IrO $_2$ (110) surface at two distinct sites (o69 and o47).

Active Spaces: The automated selection identified closed-shell (8e, 7o) active spaces for both sites, comprising adsorbate–surface bonding orbitals, neighboring Ir d-orbitals, and bridging O p-orbitals.
Hamiltonian Specifications: Each active space maps to a 14-qubit Hamiltonian containing 3,382 Pauli terms under the Jordan–Wigner transformation.
Validation: The pipeline includes an internal consistency check where the Hartree–Fock reference energy reconstructed from the qubit Hamiltonian matches the converged SCF energy to better than $10^{-6}$ Hartree.
Correlation Energies: Exact diagonalization yielded static correlation energies of $-0.044$ mHa (neutral o69) and $-2.49$ mHa (anionic o47). The anionic site exhibits roughly 50 times the correlation of the neutral site, consistent with greater multireference character in the reduced state.
Hardware Requirements: The entire active-space pipeline, from cluster extraction to validated qubit Hamiltonian generation, runs on commodity hardware without requiring high-performance computing (HPC).

Significance and Claims
The authors explicitly state that the primary contribution of this work is not a new quantum algorithm or a novel materials discovery, but rather a reproducible bridge between established surface science tools (periodic DFT, semiempirical screening, quantum chemistry integrals) and the input formats required by near-term quantum algorithms.

Key claims regarding the work's impact include:

Reproducibility: The workflow addresses the lack of a standard, open-source pipeline, enabling different research groups to reproduce results and construct active-space Hamiltonians for challenging catalytic sites.
Accessibility: By running on commodity hardware, the workflow lowers the barrier to entry for applying quantum algorithms to heterogeneous catalysis.
Methodological Caution: The paper highlights a critical technical nuance: the def2-SVP basis for Iridium is defined strictly for use with its ECP. Attempting an all-electron treatment with this basis is "ill-defined" rather than merely approximate, as the basis functions cannot accommodate the full electron count of the heavy atom.
Future Pathways: The validated pipeline serves as a foundation for future work, including the addition of dynamic correlation (e.g., via NEVPT2), the implementation of automated valence active space constructions (AVAS/RE), and the execution of VQE on current trapped-ion and superconducting hardware.