The Computational and Theoretical Chemistry Institute (CTCI) at Pacific Northwest National Laboratory (PNNL) represents a strategic institutional platform to (i) consolidate PNNL’s strengths in theoretical / computational chemistry and data science, with emphasis in scalable simulations and emerging computing paradigms and (ii) elevate the Laboratory’s stature within the scientific community in the above science domains. Through the integrated development of theory, high-performance computing (HPC), quantum simulations, and artificial intelligence (AI), the CTCI will ensure that PNNL remains a leader in theoretical and computational chemistry domains, a vital contributor to Department of Energy’s (DOE's) scientific enterprise, and an innovation hub for computational chemical sciences.
The work of the CTCI is organized into the five following thrusts:
The frontier of chemical science demands theoretical frameworks that transcend traditional approximations. Many-body interactions, multiscale phenomena, and far-from-equilibrium systems challenge our fundamental understanding of chemical behavior. As the DOE pursues transformative advances in catalysis, quantum materials, and energy transduction, we must develop electronic structure methods capable of capturing emergent phenomena where quantum coherence, dissipation, and correlation effects interplay. This thrust builds on PNNL's world-renowned expertise in electronic structure theory, theoretical spectroscopy, and nonadiabatic quantum dynamics.
Chemical processes inherently span vast spatiotemporal scales—from femtosecond bond breaking to geological-timescale mineral transformations. The grand challenge of bridging atomistic insights to continuum behavior while maintaining chemical fidelity remains. This “scale-bridging problem” limits our ability to design catalysts, predict material degradation, and understand biogeochemical cycles. Success requires revolutionary approaches that couple quantum mechanics, molecular dynamics (MD), and field theories within unified frameworks. PNNL's unique position at the intersection of fundamental science and DOE’s applied missions makes us ideally suited to tackle this challenge.
Quantum computers promise exponential speedup for chemical problems, such as the exact simulation of catalyst active sites, protein folding, and strongly correlated materials. However, error-corrected quantum computers remain 10+ years away. The breakthrough insight lies in the development of hybrid quantum–classical algorithms that can achieve a quantum advantage on near-term hardware by strategically partitioning complex problems. PNNL’s leadership in quantum algorithms, coupled with domain expertise in chemistry, power systems, and climate modeling, positions us to be among the first to demonstrate a practical quantum advantage for real-world problems.
Machine learning (ML) is revolutionizing computational chemistry, yet current approaches suffer from critical limitations. ML interatomic potentials (MLIPs) fail catastrophically outside training domains, uncertainty quantification remains ad hoc, and scaling beyond 105 atoms challenges even state-of-the-art methods. Meanwhile, scientific machine learning (SciML) promises to bridge scales through operator learning, but combining multi-fidelity data and models remains an open mathematical challenge. As the DOE pushes towards autonomous laboratories and self-driving chemical discovery, robust artificial intelligence (AI) integration becomes mission critical.
As Moore's Law ends and AI training approaches 15% of U.S. electricity production, we need computing paradigms beyond silicon. Chemical computing—using molecular reactions as computational elements—offers transformative advantages: molecular-scale logic gates (sub-nanometer vs. 5 nm CMOS), inherent parallelism through simultaneous reactions, and a tunable energy-per-bit approaching thermodynamic limits. Most remarkably, chemical computers can operate in extreme environments (high radiation, underwater, high temperature) where electronics fail, while performing in situ computation within chemical processes themselves.
