An Informatics Paradigm for Predicting Organic Chemical Stability
Abstract
Despite numerous advances in physics-based simulations, machine learning, and automatedexperimentation, chemical stability, which is central to analyzing the life-performance andfeasibility of novel organic materials, remains beyond our current capability to predict and isultimately the Achilles heel of the in silico design paradigm. Recent computational design effortshave dramatically accelerated the prediction, optimization, and validation of organic chemical andmaterials properties, yielding libraries of conceptually interesting organic molecules and materialsthat nevertheless often fail to connect with real applications or still ultimately require costly trialand-error testing that sacrifices many of the gains of computational design. Although suchmaterials testing provides a workable solution to validating specific materials in real-worldenvironments, this process is enormously costly and provides information reentneed to establish the chemical basis for organic degradation in light of real-world stressors, suchthat degradation susceptibilities can be screened in advance alongside other functional attributesand generalized across the space of potential organic molecules and materials.The goal of this research program is to establish a predictive paradigm for organic materialsdegradation that is based on the characterization and generalization of stressor-specific degradativereaction networks. While degradation pathways are only generated retrospectively in theprevailing materials testing paradigm, we propose to reverse this and use reaction networks as apredictive knowledge base from which novel materials can be screened, rapidly characterized, andultimately inserted into applications. This will be accomplished by developing a comprehensivesuite of theoretical, experimental, and informatics tools to characterize, store, and predictdegradation reaction networks associated with combinations of stressors.Recognizing the need for tools of toquantitative characterization tools (e.g., multi-mode in situ analysis and physics-basedsimulations). By delivering the capability to predict which degradation reactions happen and when,we will also be able to determine how specific degradation pathways are connected with loss-offunctionand thus for the first time provide the means of predicting the life-performance of organicmaterials in real-use environments. Although the challenges associated with predicting organicmaterials stability are formidable, recent developments in computational reaction characterization,machine learning based modeling and generalization from large datasets, and high-throughputanalytical characterization make now an ideal time to undertake this ambitious multidisciplinaryresearch program to bring the stability of novel organic materials into routine predictability.The proposed work will usher in a new paradigm for degradation studies with dedicatedcomputational and experimental tools, curated datasets, and predictive methodologiesincorporated into user-friendly and web-enabled software. By delivering these assets, this projectwill establish a chemical basis for rationalizing and generalizing stability-structure relationshipsacross organic chemical space and thus provide transformational impact in the myriad areas wherelife-performance of materials is a critical bottleneck to materials insertion, as well as adjacent areassuch as energetic materials and combustion chemistry where complex reaction networks governfunction. These advances would thus curtail the timescale for organic chemical and materialsdevelopment and application insertion in areas crucial to the warfighter.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- May 05, 2021
- Source ID
- N000142112476
Entities
People
- Brett Savoie
Organizations
- Office of Naval Research
- Purdue University
- United States Navy