⚛️ quantum physics

Embedded quantum computing for many-body surface reaction

This paper introduces QC-DFET, a quantum-computing density-functional embedding framework that successfully predicts correlated surface-reaction energetics for complex catalytic processes on Cu(111) by mapping active spaces to environment-aware qubit Hamiltonians and validating results against experimental benchmarks.

Original authors: Dedong Wan, Xiaopeng Li, Yi Fan, Jie Liu, Xiongzhi Zeng, Zhenyu Li

Published 2026-07-30

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Dedong Wan, Xiaopeng Li, Yi Fan, Jie Liu, Xiongzhi Zeng, Zhenyu Li

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 predict how a tiny Lego brick will stick to a giant, shimmering metal wall. This isn't just about the brick; it's about how the whole wall wiggles and reacts when the brick touches it. In the world of chemistry, this is called a "surface reaction," and it's the secret sauce behind everything from making fuel to cleaning pollution. For decades, scientists have used a powerful tool called Density Functional Theory (DFT) to simulate these interactions. Think of DFT as a very fast, very good guesser. It's great for most things, but when it comes to the messy, chaotic dance of electrons during a chemical reaction, it sometimes gets the details wrong, like predicting a brick will stick to the wrong spot or calculating the wrong amount of effort needed to pull it off.

To get the perfect answer, scientists need a method that accounts for the complex "many-body" problem—where every electron is constantly reacting with every other electron. This is where quantum computers enter the story. Unlike regular computers that use bits (0s and 1s), quantum computers use "qubits," which can exist in many states at once, making them theoretically perfect for simulating these chaotic electron dances. However, current quantum computers are still in their "teenage years"—they are noisy and can't handle huge, complex systems on their own. The big question has been: Can we combine the speed of the old-school guesser (DFT) with the precision of the new-school quantum computer to solve real-world chemistry problems that were previously impossible?

This paper introduces a clever new framework called QC-DFET (Quantum-Computing Density-Functional Embedding Theory) that answers "yes." The researchers, led by Dedong Wan and colleagues at the University of Science and Technology of China, didn't just try to simulate a single molecule in a vacuum; they simulated reactions happening on a real metal surface. They created a hybrid workflow where the quantum computer acts as a specialized "microscope" focused only on the most critical part of the reaction, while the classical computer handles the rest of the giant metal wall.

The team tested this new tool on three famous "puzzles" involving a copper surface (Cu(111)). First, they looked at how hydrogen gas splits apart and sticks to the copper. Previous methods struggled to get the energy barriers right for both breaking the bond and reforming it, often being off by a significant margin. QC-DFET, however, nailed the numbers, predicting the energy barriers with an error of only 0.01 to 0.02 electron-volts, which is within the "chemical accuracy" range scientists dream of.

Next, they tackled the mystery of Carbon Monoxide (CO) adsorption. For years, standard simulations have incorrectly predicted that CO prefers to sit in the "hollow" spots between copper atoms, while experiments show it actually prefers the "top" spot, sitting right on a single atom. The new quantum-enhanced method corrected this mistake, correctly identifying the top spot as the favorite and explaining that it's due to specific orbital connections rather than just a simple charge transfer.

Finally, they examined a complex branching reaction involving formate (a key step in making methanol from CO2). Here, the reaction can go two different ways, and scientists needed to know which path is faster. The new method successfully distinguished between the two paths, showing that one route is kinetically favored, and it matched experimental data for the reverse barrier of the other route.

The paper explicitly argues against the idea that we need to wait for perfect, error-free quantum computers to do useful chemistry. Instead, they demonstrate that by using the quantum processor only to sample the most important electron configurations (a technique called Quantum-Selected Configuration Interaction) and then cleaning up the results with classical math, we can get highly accurate results right now on existing, noisy hardware. They also ruled out the idea that simple, isolated cluster models are enough; they showed that keeping the "environment" of the full metal surface in the simulation is crucial for getting the right answer.

In short, this work doesn't claim to have solved all of chemistry, but it proves that we can already use today's imperfect quantum computers as a powerful, targeted tool to fix the errors in our best chemical simulations. It's like using a quantum-powered magnifying glass to fix the blurry spots in a high-resolution photo, allowing us to see the true nature of how molecules interact with metal surfaces with unprecedented clarity.

Technical Summary: Embedded Quantum Computing for Many-Body Surface Reaction

Problem Statement
Predictive simulations of heterogeneous catalysis require accurate descriptions of correlated electronic structures at extended metal interfaces. Standard Density Functional Theory (DFT), while computationally tractable, suffers from self-interaction, delocalization, and density-driven errors. These approximations often fail to balance localized chemical transformations (bond breaking/forming) with the response of the delocalized metallic environment. This leads to significant errors in reaction barriers, adsorption site preferences, and branching thermodynamics for benchmark systems such as $H_2$ dissociation on Cu(111), CO adsorption, and formate hydrogenation. While wave-function-based methods offer systematic improvements, their direct application to periodic surfaces is limited by steep computational scaling. Embedding theories (e.g., DFET) mitigate this by treating a reactive cluster at a high level within a lower-cost environment, but the "active-space solver" remains a bottleneck for capturing strong static and dynamic correlation in these clusters.

Methodology: QC-DFET Framework
The authors introduce QC-DFET (Quantum-Computing Density-Functional Embedding Theory), a hybrid framework designed to map surface-reaction active spaces to compact, environment-aware qubit Hamiltonians. The workflow integrates four key components:

DFET Embedding: A periodic DFT description of the surface is partitioned into a chemically active cluster and an environment. An embedding potential ( $V_{emb}$ ) is optimized using the Wu-Yang density-matching formulation to ensure the cluster retains the electronic response of the extended metallic surface. This potential is fixed during subsequent correlated calculations.
Reaction-Consistent Active Space (MBECAS-SR): A "Many-Body Expansion-based Complete Active Space for Surface Reaction" protocol selects a compact active space. Crucially, this protocol ensures orbital continuity along the reaction coordinate. It identifies adsorbate and surface orbitals involved in bond rearrangement and coupling, ensuring the active space remains chemically consistent across initial, transition, and final states.
Quantum-Selected Configuration Interaction (QSCI): The embedded active-space Hamiltonian is mapped to qubits (up to 28 qubits in this study) and solved using hardware-sampled QSCI on the Zuchongzhi superconducting quantum processor.
- Low-depth circuits (derived from ADAPT-UCCGSD) prepare trial states.
- Quantum hardware measures bitstrings to sample important determinants.
- A classical subspace diagonalization is performed on the selected determinant space to obtain the variational energy.
- A configuration-recovery procedure is applied to mitigate hardware noise.
Dynamic Correlation Recovery: Residual dynamic correlation outside the active space is recovered using Strongly Contracted N-electron Valence State Perturbation Theory (SC-NEVPT2) applied to the quantum-computing wave functions.

Key Results
The framework was validated on a hierarchy of experimentally constrained challenges on the Cu(111) surface:

$H_2$ Dissociation/Desorption:
- Challenge: Requires balanced accuracy for forward (dissociation) and reverse (desorption) barriers. Standard DFT (PBE-D3BJ) underestimates both; previous DFET-ENEPRT2 methods improved the reverse barrier but still underestimated the forward barrier.
- Outcome: QC-DFET achieved absolute errors of only 0.01 eV (forward) and 0.02 eV (reverse) relative to experiment.
- Insight: The success relied on the reaction-consistent active space (CAS(12e,12o) mapped to 24 qubits) which maintained orbital continuity from separated H fragments to molecular $H_2$ . Additionally, the correlated treatment correctly captured charge redistribution from the Cu slab to the adsorbate at the transition state, which DFT-level embedding suppressed.
CO Adsorption:
- Challenge: Resolving the "CO puzzle" where semi-local DFT incorrectly favors hollow sites over the experimentally observed top site.
- Outcome: QC-DFET (CAS(14e,12o), 24 qubits) correctly predicted the top-site preference with an adsorption energy within the chemical accuracy window of experiment. The fcc-top energy separation was 0.30 eV.
- Insight: Orbital-resolved Mayer bond-order analysis revealed that the correction was not driven by net charge transfer but by the correction of local orbital channels. QC-DFET identified that the strongest C-Cu $\sigma$ coupling coincides with the top site, correcting the DFT error which over-stabilized hollow sites.
Formate ( $HCOO^*$ ) Hydrogenation:
- Challenge: Distinguishing between competing $C-H$ (forming $H_2COO^*$ ) and $O-H$ (forming $HCOOH^*$ ) bond formation branches.
- Outcome: For the $H_2COO^*$ branch, QC-DFET (CAS(16e,14o), 28 qubits) corrected the reverse barrier from a DFT underestimation of 0.46 eV to 0.62 eV (experiment: ~0.65 eV), bringing it within chemical accuracy. For the $HCOOH^*$ branch, the method predicted a lower forward barrier (0.80 eV) compared to the $H_2COO^*$ pathway, identifying $HCOOH^*$ formation as kinetically favored on ideal terraces.
- Insight: The method successfully distinguished the electronic evolution of the two pathways: $H_2COO^*$ formation involves enhanced adsorbate-surface coupling, while $HCOOH^*$ formation involves progressive electronic decoupling.

Significance and Claims
The paper claims that QC-DFET establishes a practical route to correlated surface-reaction energetics by integrating quantum hardware into an embedding framework. Its significance lies in three specific contributions:

Environment-Aware Quantum Simulation: Unlike previous proof-of-concept studies using isolated clusters, QC-DFET treats realistic, experimentally benchmarked surface reactions by retaining the extended metallic environment via DFET.
Hardware-Efficient Workflow: The method does not require the quantum processor to perform a fully converged variational optimization of the entire surface problem. Instead, it uses the quantum device for the probabilistic task of sampling important determinants (QSCI), while relying on classical diagonalization and perturbative corrections (SC-NEVPT2) for final energy evaluation. This design is well-matched to near-term quantum hardware capabilities.
Quantitative Accuracy: The framework demonstrates that embedded quantum computing can resolve long-standing discrepancies in surface chemistry (e.g., CO site preference, $H_2$ barrier balancing) that semi-local DFT cannot, achieving near-chemical accuracy without the prohibitive cost of full periodic high-level wave-function methods.

The authors modestly note that this work represents a step toward predictive quantum simulations rather than a demonstration of "quantum advantage" in the sense of outperforming all classical methods, as the active spaces (up to 28 qubits) remain within the regime where classical validation is possible. The long-term value depends on scaling to larger active spaces and more complex interfaces.

Technical Summary: Embedded Quantum Computing for Many-Body Surface Reaction

More like this