Quantum Chemistry Simulation of Dibenzothiophene for Asphalt Aging Analysis

TL;DR

This work applies a quantum-chemistry pipeline to dibenzothiophene as a model for asphalt aging, using an (8e,8o) active space and Bravyi-Kitaev mapping with Z2 tapering to simulate with 14 qubits. It compares k-UpCCGSD-VQE and ADAPT-VQE on both simulators and IBM hardware, finding a substantial correlation-energy gain ($ $~9 Ha) over HF and superior performance over DFT and CASSCF within the chosen active space, illustrating a quantum advantage for strongly correlated oxidation chemistry. The study provides a reproducible workflow, error-mitigated results, and a clear hardware-aware analysis of circuit depth as a key bottleneck for near-term devices. Industrial relevance is highlighted through design principles for oxidation-resistant asphalt formulations, estimated durability improvements, and concrete pathways toward scalable quantum-assisted materials design. The work paves the way for larger, multi-molecule simulations and eventually fault-tolerant quantum chemistry in infrastructure materials research.

Abstract

This paper presents the execution and analysis of a comprehensive quantum chemistry pipeline for gathering actionable insight into asphalt aging mechanisms through the study of dibenzothiophene (DBT), a key sulfur-containing compound in asphalt binders. Using advanced quantum algorithms, specifically Variational Quantum Eigensolver (VQE) with k-UpCCGSD and ADAPT-VQE ansatze, we achieved ground state energy calculations with accuracies reaching -864.69 Ha. Our implementation demonstrates quantum advantages in handling strongly correlated electron systems while providing actionable insights for designing oxidation-resistant asphalt formulations. The work also establishes a scalable framework for quantum-enhanced materials design.

Figures (7)

• Figure 1: Molecular Orbital Analysis of Dibenzothiophene. Energy diagram showing frontier molecular orbitals with HOMO at -0.0729 Ha (red) featuring sulfur p-orbital character and LUMO at +0.0764 Ha (blue) displaying aromatic $\pi^*$ antibonding character. The HOMO-LUMO gap of 0.149 Ha (4.05 eV) indicates substantial kinetic stability against oxidative attack. Yellow shading highlights the (8e,8o) active space targeting orbitals 45-52.
• Figure 2: Circuit Depth Analysis for Near-term Implementation. Left: Linear scale showing k-UpCCGSD requiring 9,398 layers versus ADAPT-VQE's 41 layers. Right: Logarithmic scale emphasizing the 229-fold reduction in circuit depth achieved through adaptive operator selection. This dramatic difference illustrates that circuit depth, rather than qubit count, represents the primary limitation for NISQ device implementation.
• Figure 3: VQE Algorithm Convergence Comparison. Left: k-UpCCGSD VQE convergence achieving -864.69062 Ha final energy in 114 iterations (112.43 seconds). Right: ADAPT-VQE convergence reaching -855.5711 Ha in fewer iterations (30.63 seconds) but with incomplete correlation recovery. The structured k-UpCCGSD approach demonstrates superior accuracy while ADAPT-VQE offers faster convergence with hardware-compatible shallow circuits.
• Figure 4: Quantum Advantage in Correlation Energy Recovery. Comparison of correlation energy capture across different computational methods. VQE (k-UpCCGSD) achieves -9.08 Ha correlation energy, representing 2.5$\times$ improvement over DFT (B3LYP) at -3.62 Ha and 113$\times$ enhancement over classical CASSCF at -0.08 Ha. Hartree-Fock provides zero correlation energy by construction, serving as the reference baseline.
• Figure 5: Computational Scaling Comparison. Left: Linear scale comparison showing polynomial scaling advantages of quantum methods. Right: Logarithmic scale including exponential CASSCF scaling (O(2$^N$)). VQE exhibits favorable O(N$^2$) parameter scaling versus classical methods: HF O(N$^4$), DFT O(N$^3$), and exponentially scaling CASSCF. This quantum advantage becomes pronounced for larger molecular systems essential for industrial applications.