Simulation of additive binding energies in asphalt using quantum-selected configuration interaction (QSCI)

This paper demonstrates that a hybrid quantum-classical workflow called QuantumPave, utilizing quantum-selected configuration interaction (QSCI) on a 54-qubit quantum processor, can successfully compute chemically meaningful additive binding energies for asphalt binder models, proving the feasibility of quantum-centric supercomputing for industrially relevant materials science problems.

The Gist

Imagine trying to figure out why a specific ingredient makes a cake stay fresh longer. In the world of roads, that "ingredient" is an additive mixed into asphalt (the black stuff used to pave streets). Scientists want to know exactly how strongly this additive sticks to the asphalt to prevent the road from cracking and aging due to sun and weather.

This paper is about a new, high-tech way to measure that "stickiness" (called binding energy) using a mix of super-fast classical computers and a brand-new type of quantum computer.

Here is the story of what they did, explained simply:

1. The Problem: The "Glue" of the Road

Asphalt is a complex soup of molecules. To keep roads durable, engineers add chemicals to stop them from rotting. To understand how these additives work, scientists need to calculate the energy of the "handshake" between the additive and the asphalt.

• The Challenge: These handshakes are tiny and tricky. They involve electrons dancing around in complex ways that regular computers struggle to predict perfectly.
• The Test Subject: Instead of simulating the whole messy road, the researchers picked a tiny, representative model: a Pyridine-Phenol complex. Think of this as a "miniature handshake" between two molecules (one with a nitrogen ring, one with an oxygen ring) that mimics the real chemistry found in asphalt.

2. The New Tool: "Quantum-Centric Supercomputing"

The authors used a workflow they call QuantumPave. Imagine a team of two experts working together:

• The Classical Computer (The Librarian): It handles the heavy lifting of organizing data and doing the final math calculations.
• The Quantum Computer (The Sampler): Instead of calculating every single possibility (which would take forever), the quantum computer acts like a master chef tasting a soup. It quickly "samples" the most important flavors (electronic configurations) to see what the dish tastes like.

This specific method is called QSCI (Quantum-Selected Configuration Interaction). It's like saying, "We don't need to check every single grain of sand on the beach; let's just check the 10 most important grains that determine the shape of the dune."

3. The Experiment: A 54-Qubit Taste Test

The researchers ran their "miniature handshake" simulation on a real quantum computer (the IQM Emerald processor).

• The Setup: They focused on a small group of 10 electrons and 10 orbitals (the "active space") where the magic happens.
• The Twist: Usually, quantum computers are noisy (like a radio with static). You'd expect the noise to ruin the result. However, in this specific method, the noise actually helped! It was like the static on the radio accidentally helped the chef taste more of the soup, ensuring they didn't miss any important flavors.
• The Result: The quantum computer's result matched the "perfect" classical calculation exactly. They found the binding energy was -3.52 kcal/mol.

4. What the Numbers Mean

• The Match: The quantum computer and the classical "gold standard" agreed perfectly. This proves the new method works on real hardware without needing complex tricks to fix errors.
• The Gap: The result (-3.52) was a bit lower than the real-world experimental value (-6.25).
  • Why? The researchers explain that their "miniature model" (the active space) was too small to catch every tiny force involved in the handshake. It captured the strong hydrogen bond, but missed some of the weaker, long-range forces.
  • Analogy: It's like measuring the weight of a person by only weighing their head and torso. You get a good idea of their weight, but you miss the weight of their legs and arms. To get the exact real-world number, they would need to include more "body parts" (more electrons) in the calculation.

5. The Takeaway

This paper is a proof-of-concept. It shows that:

1. We can use current, noisy quantum computers to solve real chemistry problems related to road materials.
2. By letting the quantum computer just "sample" and the classical computer "diagonalize" (solve the math), we can get accurate results without needing perfect, error-free quantum machines yet.
3. This approach, called QuantumPave, is a promising step toward understanding how to make roads last longer, though the model used here was a simplified version of the real thing.

In short: They built a digital bridge between a quantum computer and a classical one to measure how well two molecules stick together. The test was successful, proving that this new hybrid method can handle the complex math of road chemistry, even with today's imperfect quantum hardware.

Technical Summary

Technical Summary: Simulation of Additive Binding Energies in Asphalt Using Quantum-Selected Configuration Interaction (QSCI)

Problem Statement
The long-term durability and oxidative ageing of asphalt binder, a critical component of road infrastructure, are governed by complex molecular interactions. Understanding how additives interfere with ageing pathways requires accurate calculation of binding energies between additives and key asphalt components. However, predicting these energies is challenging due to the need to account for strong correlation effects and dispersion interactions, which are often poorly described by standard mean-field approximations. The authors aim to demonstrate a "quantum-centric" workflow capable of computing these binding energies for an industrially relevant materials problem, specifically focusing on the interaction between asphalt components represented by a pyridine-phenol complex.

Methodology
The study introduces QuantumPave, a hybrid quantum-classical workflow designed to compute additive binding energies. The methodology integrates three complementary approaches:

1. System Selection: The authors selected a 24-atom pyridine-phenol complex ($C_{11}H_{11}NO$) as a representative model. This system features a strong O-H$\cdots$N hydrogen bond and $\pi$-conjugated electron systems, mirroring functional groups identified in molecular dynamics studies of asphalt ageing.
2. Geometry Optimization: Machine-learning interatomic potentials (ORB v3) were utilized for rapid geometry optimization, providing a balance between computational efficiency and chemical accuracy.
3. Binding Energy Calculation: Three methods were employed to calculate binding energies:
   • Machine Learning (ORB v3): Used for rapid screening and initial estimates.
   • Density Functional Theory (DFT): B3LYP-D3/def2-SVP calculations served as a classical reference, incorporating dispersion corrections for $\pi$-$\pi$ stacking and hydrogen bonding.
   • Quantum-Selected Configuration Interaction (QSCI): Also known as Sample-based Quantum Diagonalisation (SQD), this is the core quantum-inspired method. It employs an active space approximation (10 electrons in 10 orbitals, mapped to 20 qubits) to reduce complexity while targeting valence-dominated interactions.
4. Quantum Execution: The QSCI algorithm was executed on the 54-qubit IQM Emerald superconducting quantum processor. The workflow involves:
   • Sampling: The quantum processor samples dominant electronic configurations (bitstrings) from the prepared state.
   • Error Mitigation: Instead of zero-noise extrapolation, the method utilizes self-consistent configuration recovery. This iterative process repairs invalid configurations (those violating particle-number symmetry) using average orbital occupations from retained samples.
   • Diagonalisation: The classical High-Performance Computing (HPC) resources perform the diagonalisation within the selected subspace.
   • Dispersion Correction: Empirical D3BJ corrections were added to the QSCI results to account for long-range van der Waals interactions not captured by the active space electronic structure alone.

Key Results

• Hardware Accuracy: On the IQM Emerald processor, the SQD method reproduced the classical active-space Complete Active Space Configuration Interaction (CASCI) reference exactly. Both yielded a binding energy of -3.52 kcal/mol (-0.153 eV).
• Noise Robustness: The device noise, rather than degrading the result, broadened the sampling distribution to span the active space. Consequently, the classical diagonalisation recovered the exact active-space energy without requiring zero-noise extrapolation.
• Comparison to Experiment: The calculated active-space value (-3.52 kcal/mol) underbinds the experimental calorimetric enthalpy of approximately -6.25 kcal/mol. The authors attribute this discrepancy to the compact nature of the active space, which captures internal valence correlations but omits contributions from the wider orbital space involved in the electrostatic hydrogen bond.
• DFT Comparison: The B3LYP-D3 method overestimated the binding energy (-9.08 kcal/mol), consistent with known tendencies of hybrid functionals to overestimate moderate hydrogen bonds.

Key Contributions

• Proof-of-Concept for Quantum-Centric Supercomputing: The paper demonstrates a fully reproducible, small-scale instantiation of quantum-centric supercomputing, where a QPU samples configurations and classical HPC performs diagonalisation.
• Validation of QSCI/SQD on Hardware: It provides experimental evidence that QSCI can achieve numerical precision against its classical active-space limit on current noisy intermediate-scale quantum (NISQ) hardware.
• Noise-Tolerant Workflow: The study highlights a workflow where device noise aids convergence by broadening the sampled distribution, eliminating the need for complex zero-noise extrapolation techniques in this specific regime.
• Application to Materials Science: It successfully applies these quantum algorithms to a problem relevant to the asphalt industry, bridging the gap between quantum computing demonstrations and industrial material problems.

Significance and Claims
The paper claims that chemically meaningful binding energies for industrially relevant materials problems are attainable on current quantum hardware within a quantum-centric workflow. The significance lies in demonstrating that:

1. Feasibility: Real quantum processors can execute correlated electronic-structure calculations that match classical active-space references.
2. Practicality: The approach is robust against noise, suggesting that error mitigation strategies like zero-noise extrapolation may not always be necessary if the sampling strategy and classical post-processing are optimized.
3. Scalability: While the current active space is limited, the methodology offers a systematic path to improvement. Future work can utilize qubit-efficient formulations and embedding schemes (e.g., density matrix embedding) to address larger, more realistic condensed-phase environments like the asphalt matrix.

The authors conclude that the primary limitation is the size of the active space rather than the hardware itself, and that the approach provides a promising foundation for next-generation computational chemistry applications in materials science and catalysis.