How quantum computing and advanced algorithms are revolutionizing our understanding of intermolecular potential energy surfaces
Imagine trying to understand every possible handshake between thousands of people—not just a firm grip, but the subtle variations in pressure, angle, and distance that make each interaction unique. Now shrink that world down to the molecular scale, where invisible forces dictate whether proteins fold correctly, drugs bind to their targets, or materials gain their unique properties. This is the challenge scientists face in mapping intermolecular interactions, the subtle yet powerful forces that occur between molecules without forming chemical bonds.
For decades, researchers have struggled to accurately capture the complete picture of these interactions through what are known as potential energy surfaces—complex maps describing how energy changes as molecules approach, orient, and interact with each other. But recent breakthroughs are revolutionizing this field, combining quantum computing with advanced algorithms to automatically compute these landscapes with unprecedented speed and accuracy 3 . These developments are opening new frontiers in drug discovery, materials science, and our fundamental understanding of the molecular world.
Key Insight: Non-covalent interactions may be individually weak, but their collective effect determines molecular structure, function, and reactivity across biology and materials science.
At its core, a potential energy surface (PES) is a conceptual map that describes the energy of a molecular system as a function of the positions of its atoms. Think of it as a mountainous landscape where every point on the ground represents a particular arrangement of atoms, and the height at that point represents the energy of that configuration 1 .
These are the valleys in the landscape, representing stable molecular configurations where molecules can reside comfortably.
These are mountain passes between valleys, representing transition states that molecules must pass through to rearrange themselves.
This conceptual diagram represents how energy varies with molecular configuration, showing minima (valleys), maxima (peaks), and transition states (passes).
For simple systems, these landscapes are relatively straightforward. A diatomic molecule, for instance, can be represented by a simple curve showing how energy changes with the distance between two atoms—the energy decreases as atoms approach their optimal bonding distance, then sharply increases if they get too close due to repulsive forces 6 .
However, the complexity explodes for larger systems. Each additional atom adds more dimensions to the landscape, creating a multidimensional surface that's impossible to visualize completely. A system with N atoms has (3N-6) dimensions for nonlinear molecules, creating a hyper-surface that can only be partially explored or projected onto key dimensions of interest 6 .
While chemical bonds (covalent bonds) represent the superhighways of chemistry, non-covalent interactions are the subtle footpaths that often determine molecular behavior. These weaker forces include:
Despite their individual weakness, the collective effect of these interactions is profound. They dictate how proteins fold into their functional shapes, how drugs recognize their biological targets, how DNA strands pair up, and how materials self-assemble into complex structures 3 .
Understanding these interactions requires knowing not just the most stable arrangements (the energy minima), but the complete landscape of possible interactions—including the higher-energy configurations that molecules briefly visit as they transition between stable states.
The fundamental challenge in mapping these landscapes lies in the sheer computational cost. Traditional quantum mechanical methods, while accurate, are prohibitively expensive for all but the smallest systems. As noted in recent research, "the most accurate approaches, achieving chemical accuracy, rely on quantum mechanical descriptions of non-covalent interactions, which limits their scalability" 3 .
This has forced researchers to make difficult trade-offs between accuracy and computational feasibility. Common approaches include:
Calculating only selected points on the surface and using interpolation to estimate the rest 1
Focusing only on the most chemically relevant degrees of freedom 6
Each compromise comes with limitations, particularly for non-covalent interactions where the energy differences are small—often just 1-5 kcal/mol—but critically important. Achieving "chemical accuracy" (within 1 kcal/mol of reality) has been the holy grail of computational chemistry, but elusive for all but the simplest systems.
In 2025, a groundbreaking study published in Communications Physics demonstrated a novel approach that combines quantum computing with classical high-performance computing to tackle this challenge. The research team developed what they call quantum-centric supercomputing (QCSC), a hybrid approach that uses quantum processors for specific subroutines while leveraging classical computers for other tasks 3 .
The researchers focused on two model systems: the water dimer (two water molecules connected by a hydrogen bond) and the methane dimer (two methane molecules held together by weak dispersion forces). These systems represent two important classes of non-covalent interactions 3 .
Using the Automated Virtual Atomic Site (AVAS) method to identify which molecular orbitals are most relevant to the interactions
Employing quantum circuits with 27-54 qubits to sample important electronic configurations from the wavefunction of the molecular system
Using classical computing resources to process quantum measurements and recover electronic configurations that might be corrupted by quantum noise
Solving the Schrödinger equation in the subspace spanned by the recovered configurations to determine accurate energy levels
Using Hamiltonian variance extrapolation to estimate the exact energy from calculations with varying levels of approximation
This approach, called Sample-based Quantum Diagonalization (SQD), allowed the team to simulate systems with up to 54 qubits—significantly larger than previous quantum simulations of molecular systems 3 .
The quantum-centric approach achieved remarkable accuracy, with deviations from the gold-standard coupled-cluster (CCSD(T)) method within 1.000 kcal/mol in the equilibrium region of the potential energy surface. This places it within the coveted realm of "chemical accuracy" for these non-covalent interactions 3 .
| Method | Accuracy | Scalability | Key Strengths |
|---|---|---|---|
| Quantum-Centric (SQD) | High (~1 kcal/mol) | Moderate | Near-chemical accuracy for larger systems |
| Traditional Quantum Methods | Very High | Limited | Gold standard for small systems |
| Force Fields | Variable | High | Fast simulation of large systems |
| Machine Learning Potentials | Moderate-High | Moderate | Balance of speed and accuracy |
Table 1: Comparison of Computational Methods for Non-Covalent Interactions
Perhaps more importantly, the research demonstrated that quantum computers could sample electronic configurations more efficiently than certain classical heuristics, particularly for larger active spaces and calculations further from equilibrium geometries. This suggests a path toward quantum advantage—where quantum computers outperform classical ones—for specific aspects of molecular simulations 3 .
The study also established several technical milestones: diagonalizing the largest subspace to date (249 million configurations) and successfully integrating active space selection with the quantum computing software stack, paving the way for more automated workflows in the future 3 .
Mapping intermolecular potential energy surfaces requires a sophisticated set of theoretical and computational tools. Here are the key components of the modern researcher's toolkit:
| Tool Category | Examples | Function |
|---|---|---|
| Quantum Methods | Coupled Cluster (CCSD(T)), Configuration Interaction | High-accuracy energy calculations for benchmark systems |
| Force Fields | GROMOS, CHARMM, AMBER, OPLS | Fast approximate simulations of large systems |
| Quantum Computing | Sample-based Quantum Diagonalization (SQD), Unitary Coupled Cluster | Leveraging quantum hardware for challenging electronic structure problems |
| Analysis Tools | Molecular Electrostatic Potential Analysis, Energy Decomposition | Interpreting and understanding the nature of interactions |
| Database Resources | Biofragment Database, S66, NENCI | Reference data for method development and validation |
Table 2: Essential Tools for Intermolecular PES Research
Each tool has its strengths and limitations. Force fields, for instance, are essential for simulating large systems like proteins in solution, but their accuracy depends heavily on careful parameterization against experimental or high-level theoretical data. As one evaluation study noted, "correlation coefficients between experimental values and simulation results range from 0.76 to 0.88" for different force fields, indicating significant but not perfect agreement with reality 4 .
The molecular electrostatic potential (V(r)) has emerged as a particularly valuable analysis tool, as it "has emerged as a widely used tool for extracting information of complex quantum chemical calculations to understand and characterize interaction sites of molecules" 5 . This approach helps researchers identify regions of molecules that are likely to participate in attractive or repulsive interactions.
The ability to automatically compute global potential energy surfaces is transforming molecular science. Recent developments suggest several exciting directions:
Systems like the Molecular Interaction Rules (MIR) workflow are making it possible to systematically generate and analyze thousands of interaction scenarios
Approaches like SQD that leverage the strengths of both quantum and classical computing
Models trained on quantum mechanical data that can achieve near-quantum accuracy at much lower computational cost
Comprehensive databases of non-covalent interactions that allow researchers to compare and improve their methods
| Method | Interaction Energy (kcal/mol) | Error vs. Gold Standard | Computational Cost |
|---|---|---|---|
| CCSD(T) (Gold Standard) | -5.02 | Reference | Very High |
| Quantum-Centric (SQD) | ~-4.5 to -5.5 | <1.0 kcal/mol | High |
| MP2 | -4.87 | ~0.15 kcal/mol | High |
| DFT with Dispersion Correction | Variable | 0.5-2.0 kcal/mol | Moderate |
| Classical Force Fields | Variable | 1.0-3.0 kcal/mol | Low |
Table 3: Performance of Selected Methods on Water Dimer PES
As these technologies mature, we're moving closer to a future where researchers can automatically generate accurate potential energy surfaces for complex molecular systems as routinely as we now determine molecular structures. This will dramatically accelerate progress in fields ranging from pharmaceutical development to materials design.
The automatic computation of global intermolecular potential energy surfaces represents more than just a technical achievement—it offers a fundamental shift in how we understand and design molecular systems. By mapping the complete landscape of possible interactions, rather than just the stable minima, researchers can predict not just what structures are possible, but how molecules transition between them.
This comprehensive understanding is particularly crucial for non-covalent interactions, where the subtle balance of multiple weak forces often determines function. As the recent quantum-centric computing study demonstrates, we're developing tools that can capture these subtleties with increasing accuracy and efficiency 3 .
Impact: The implications span across science and technology: designing drugs that more selectively bind to their targets, creating materials with precisely tuned properties, understanding the molecular machinery of life, and developing new catalysts for sustainable chemistry.
As we continue to refine these computational approaches, we're not just building better tools—we're developing new eyes to see the invisible forces that shape our molecular world.