Download PDF Fall Bridge on the Materials Genome Initiative September 29, 2025 Volume 55 Issue 3 The Fall 2025 issue explores the Materials Genome Initiative’s progress and future outlook, showcasing advances in autonomous experimentation, sustainable polymers, next-generation batteries, and the broader role of AI in engineering. Frontiers in Polymer Materials, Sustainability, and AI/ML-Based Self-Driving Laboratories (SDLs) Monday, September 29, 2025 Author: Rigoberto Advincula Where AI meets chemistry: accelerating materials innovation. The ability to conduct empirical experiments or process optimization guided by the scientific method (hypothesis-driven) is the driving force behind advances in science and engineering. Experiments that test theories and simulations serve as a reality check, enabling the discovery of new phenomena or efficiencies that lead to widely adopted industrial technologies. As Max Planck is famously quoted as saying, “Experiment is the only means of knowledge at our disposal. Everything else is poetry, imagination.” A future has just arrived—an intelligent operating system for conducting experiments, embodied by an automated laboratory-operated robot and controlled by computers that autonomously set or reset experimental parameters—essentially, a 24/7 assistant that never sleeps. This assistant continuously analyzes vast amounts of data, identifies gaps in the experiment or process, sends instructions for recalculation or new simulation, and resets the experimental setup for the next phase. This represents the future of self-driving laboratories (SDLs), driven by artificial intelligence and machine learning (AI/ML) in the fields of chemistry and materials science (Abolhasani and Kumacheva 2023; Tom et al. 2024). They are poised to facilitate new scientific discoveries, optimize process engineering, and rapidly accelerate research and development (R&D) for manufacturing. In the context of materials, specifically polymer materials, this assistant could identify and discover new polymers with optimized properties for various applications. It could also design new synthesis routes, predict thermo-mechanical properties and degradation, and develop strategies for sustainability and recyclability. It may even predict project costs and environmental impacts, going beyond the current capabilities of life cycle analysis and technoeconomic analysis methods. Moreover, it can recalculate these in real time as new information or experimental pathways emerge. Will this be possible for today’s polymer science and engineering? Can we solve the plastic conundrum? This article explores the potential of AI/ML to drive innovation in polymer materials and examines the role of SDLs in addressing challenges in polymer science and engineering. Polymer Materials in Society Polymer materials play a crucial role in many aspects of society and daily life. Plastics are ubiquitous, and we accept them as an integral part of everyday living. By definition, polymer materials consist of large molecules, or macromolecules, and therefore naturally have high molecular weight (MW). They are composed of monomers, repeating structural units connected in linear, branched, or architected configurations. Their distribution, or polydispersity, is a key feature that determines their processability. From semi-crystallinity to optical and dielectric properties, polymers offer advantages that enable them to replace steel, glass, and ceramics in various applications, particularly when the proper weight-to-strength ratio is achieved. However, future research and innovation focus on understanding field effects in macromolecular synthesis, novel depolymerization or dynamic bonding rearrangements, and new structure–composition–property relationships in polymers. An emphasis on materials physics, architected microstructures, and advanced processing techniques, such as additive manufacturing (AM), requires new creative tools that can be enhanced with AI/ML tools (Advincula et al. 2025; Chen et al. 2025). Similar trends will emerge in more ML-driven analytical and characterization of advanced polymer materials. It is intriguing to consider whether many Nobel Prizes related to polymer chemistry and physics—such as those awarded to Staudinger, Ziegler and Natta, Flory, Merrifield, De Gennes, Heeger, MacDiarmid, Grubbs, Lehn, Stoddart, and others (Nobel Prize Organisation n.d.; Seymour et al. 1989)—might have been awarded earlier had AI/ML and SDLs been available to them. At the same time, the knowledge from their contributions serves as a foundation for future work using these new tools. Breakthroughs in polymers have led to advances in catalysis and manufacturing, lighter-weight vehicles with plastic parts, electrically conducting polymers and devices, liquid crystal displays, the total synthesis of biopolymer-based drugs, improvements in biomedical devices, and high-performance composites. Polymer materials play a crucial role in many aspects of society and daily life. Polymer Synthesis Tied to Properties It is helpful to categorize polymer synthesis and mechanisms into broad classes: step-growth and chain-growth reaction mechanisms (Lodge and Hiemenz 2020; Saldívar-Guerra and Vivaldo-Lima 2013). The use of a catalyst refers to agents, such as organometallic compounds or enzymes, that broadly accelerate and mediate the reaction process. The most common commodity plastics, such as polyethylene, polypropylene, polystyrene, polybutadiene, and ABS, are made through chain-addition reaction mechanisms. Popular polymers, such as polyesters, nylons, and polycarbonates, are produced through step-growth polymerization mechanisms; however, some can also be synthesized using ring-opening addition methods. They are distinguished by their MW, polydispersity, living or nonliving mechanisms, and the use of media (solutions or dispersions) and bulk polymerization methods. Polymers are not limited to linear structures, though; they can also be synthesized as grafts, blocks, hyperbranched structures, or even controlled microstructures. Copolymers are formed from two or more monomer compositions that can be connected in statistical, alternating, or block arrangements, and their structures can be predicted using the Mayo–Lewis equation. Overall, nanophase, mesophase, and microphase behaviors contribute to the observed macroscopic properties. Lastly, polymers can also be categorized as thermosets or elastomers, based on their degree of crosslinking and distinct thermo-mechanical properties (brittle but tough, or rubbery). Popular thermosets include epoxy, vinyl esters, phenolic resins (also known as phenol-formaldehyde), and polyurethane. Representative elastomers include natural rubber, silicone, thermoplastic polyurethane, and other synthetic rubbers. Polymers, Plastics, and Recycling Most polymer products are formulated with additives to optimize their properties for processing and specific applications. For polymers, the term “plastic” is more commonly used, and resins, epoxies, paints, adhesives, and films are also identified as standard formulations with a majority polymer composition. Formulation is a crucial process in the polymer conversion industry. The versatile properties, cost-effectiveness, and applicability of polymers across sectors—including packaging, medical supplies, coatings, military, automotive, aerospace, semiconductors, and building construction—underscore their importance and the performance of the final products. The development of smart and innovative functional polymers will open new opportunities in coatings, smart packaging, electronics, sensors, medical devices, and drug delivery systems. The term “recycling” is now primarily associated with plastic-related activities (Collias and Layman 2021). Yet, less than 10% of plastics are ever recycled, with the rest returning to the environment or being incinerated, often in their original state and potentially in a non-degradable form. Is it possible to convert our major classes of plastics into sustainable and even upcycled polymer materials? Typically, plastics are sourced from petroleum-based feedstocks. However, natural polymers and biopolymers constitute the majority of biomass (cellulose or non-cellulose) from plants, trees, and agriculture (Das et al. 2023). There are many types of biodegradable polymer materials, including polylactide, polyhydroxyalkanoates, and other lactic acid copolymers, as well as natural or synthetic polybutadienes, which can be classified as part of the biobased feedstock category. The term “bioplastics,” or sometimes biodegradable polymers, is often associated with these classes. These polymers also constitute living matter, including polysaccharides, polynucleotides, and polypeptides. Their self-assembly, or intelligent assembly, facilitated by enzymes and DNA, enables them to function correctly within living systems. Polymers are also referred to as soft matter (van Saarloos et al. 2024), a subset of condensed matter that focuses on the physics and phenomena of such systems, including biological ones. Thus, they are not only essential for life but also serve as critical materials for many industries. The development of smart and innovative functional polymers will open new opportunities in coatings, smart packaging, electronics, sensors, medical devices, and drug delivery systems. Smart polymers can be stimuli-responsive to various triggers, including pH, light, temperature, chemicals, and electric fields, leading to applications such as smart textiles and wearables, autonomous self-healing materials, and novel drug-releasing polymer excipients or implants. Additive manufacturing, better known as 3D printing, has advanced beyond rapid prototyping into limited production. Combining AM with stimuli-responsive polymers gives rise to a new concept known as “4D printing.” By enabling economic production and greater design freedom, these polymers can become part of specialty polymer or product lines in any primary polymer manufacturing industry. The New AI/ML-Driven Research and Optimization Tools in SDLs Is there a role for AI/ML and SDLs in the future of polymer materials? More interestingly, will the polymer scientist or engineer continue to be essential in driving polymer innovation and sustainability? Is the “scientist-in-the-loop” still critical in the laboratory of the future? The automated research laboratory will combine mechatronics, simulation, robotics, and data analysis to accelerate research and discovery. The combination of AI/ML with SDLs can automate the design, architecture, formulation, execution, and analysis of experiments, leading to more efficient polymer materials research and development (Beaucage et al. 2024) (Figure 1). For now, the answer should be “yes”: we still need the human-in-the-loop to utilize this tool. Simulation and Theory Lead the Way To start, simulation and theoretical predictions can be more directly linked with an automated laboratory (Gartner and Jayaraman 2019). Quantum chemical calculations, ranging from atomistic to coarse-grained methods, including molecular dynamics (MD), are crucial for predicting bond connectivities and reaction pathways. The use of density functional theory (DFT) is key to achieving optimal reactivity. Simulating pseudo-chiral and conformational properties is essential for understanding multi-phase behavior beyond MD. For practical correlation of the effects of MW or polydispersity, it is necessary to refine assumptions based on self-consistent field theories. In linking mesoscopic to macroscopic properties, finite element analysis and multi-physics simulations stand out as predictive tools for the desired processing methods and functional properties. Creating these digital twins ensures that algorithm development and experimental work are fundamentally grounded in reality. These simulation tools are essential for guiding the execution of real-time experiments and dynamically adjusting reaction parameters to optimize, for example, synthesis conditions. The generation of hypotheses begins with molecular bonding and macromolecular-driven paradigms, based on bond reactivity, reactivity ratios (for copolymers), and conformational and configurational considerations, including Kuhn and segment lengths (Rubinstein and Colby 2003). In this way, predictive modeling can be employed, where ML models analyze polymer structures and properties to predict new microstructures and their corresponding characteristics. For blends, formulations, chi-interactions, and solubility parameters are essential. A data-driven design will utilize new generative AI algorithms to uncover novel relationships between polymer characteristics and performance. Lastly, the goal could be the optimization of existing synthesis routes, reaction conditions, and catalysis to reduce the number of variables and enable a more straightforward scale-up of manufacturing processes. Specific examples of hypothesis-driven research leading the way in simulation and algorithm development include copolymerization (Kuenneth et al. 2021), catalytic olefin polymerization (Vittoria et al. 2022), and the design of polymer blend electrolytes (Wheatle et al. 2020). A key future direction is the application of agentic AI (AI capable of goal-driven actions and independent decision making) and multi-agent AI with graph reasoning for generating new research directions and paths to discovery (Ghafarollahi and Buehler 2024). Codes and LLMs Enable Faster Data Analytics Recently, efforts have been made to provide codes and molecular labeling of polymer materials that incorporate the polymer’s structure and, potentially, its function or properties. These methods for labeling polymer materials or coding will enable tracing based on structure, composition, sequencing, and function, which can significantly aid in accelerating simulation and data curation. The Simplified Molecular Input Line Entry System (SMILES) is an example. SMILES provides a specification in the form of line notation with short ASCII strings, representing entries in polymer databases that can be used as indexing identifiers (Lin et al. 2019). This refinement essentially serves as a more effective reference language. It can even be implemented in large language models (LLMs) with the development of more efficient Retrieval Augmented Generation (RAG) systems. For polymers, this involves indexing, retrieval, and generation. The LLM can use this “augmentation” to generate more precise and cited responses, rather than relying solely on general training data. First, the polymer data is processed and converted into a searchable format using numerical vectors, and then stored in a vector database. To retrieve information, a user submits a query, and the system searches the vector database to find the most relevant results. The retrieved data is then incorporated into the prompt sent to the LLM (Yu et al. 2024). An open AI platform such as ChatGPT could be particularly useful for polymer scientists and engineers. Such a platform could lead to more efficient data analytics for potential synthesis routes, characterization, and forensic analysis of failures. It could also allow for faster design and modification of new, functional, and biodegradable polymeric materials for advanced applications. Engineering the Science of Polymers from Statistics to Digital Twins In chemical engineering, ML is a subset that, when applied, enables feedback-loop methods for optimizing process or reaction yields. It allows computers to learn from data and refine predictions without explicit reprogramming. Process modeling based on grounded thermodynamic and kinetic principles includes mass transport and diffusion conditions. Often overlooked in continuous flow processing, such as in solution or melt, is treating these fluids as complex systems that can be modeled using statistical mechanics. In process optimization, parameters such as pressure, temperature, volume, flow rate, and dosing compositions must be controlled to achieve higher yields. By applying statistical or matrix process optimization methods, careful algorithm development can reduce the number of experiments and focus on more critical parameters beyond those addressed by principal component analysis. For example, a 60% reduction in the number of experiments while achieving a 90% reaction yield can be a game-changer for streamlining process development. Typically, experimental work relies heavily on expert intuition and even trial-and-error methods. This approach can be wasteful, however, and may be a poor economic driver for product launches. A data-driven approach to R&D experimentation, from start to finish, is required, grounded in predictive modeling and macromolecular information. Bayesian optimization platforms, combined with chemical descriptor databases, can bring reaction models closer to alignment with real-world chemistry and process engineering. Statistics Is Key to the Design of Experiments Statistical methods are fundamental to ML approaches (Bzdok et al. 2018). They provide tools to analyze data, build models, and evaluate their performance in experiments. Several key statistical concepts, including regression analysis, probability distributions, hypothesis testing, and clustering, can be more effectively applied in the design of experiments for polymer research. These methods facilitate the understanding of data, the identification of patterns, and the making of predictions. Beyond statistical methods, several ML-directed algorithms include regression analysis, decision trees, random forests, Bayesian optimization, and neural networks. By understanding the different types of ML algorithms (supervised or unsupervised learning) and their respective strengths, polymer scientists and engineers can make more informed decisions about which algorithm is best suited for a particular task and apply it in scientific experiments or process optimization. Unfortunately, statistics is not as widely emphasized in STEM education as a foundation for sound decision making in designing and planning experiments. By understanding the different types of ML algorithms (supervised or unsupervised learning) and their respective strengths, polymer scientists and engineers can make more informed decisions about which algorithm is best suited for a particular task and apply it in scientific experiments or process optimization. Unfortunately, statistics is not as widely emphasized in STEM education as a foundation for sound decision making in designing and planning experiments. Digital Twins and Neural Networks An optimized system, built on carefully simulated digital twins, could, in the future, adapt or reconfigure the simulation by learning from real-time experimental feedback. This would iteratively refine reaction conditions to minimize material waste (or the number of experiments) and enhance polymer properties in real time. It could also accelerate discovery cycles by incorporating more generative AI or deep learning (DL) methods. DL relies on neural networks, mathematical models inspired by the brain’s psychological functions. These artificial neural networks consist of layers of interconnected neurons that process and learn from data. The input layer receives raw data, which is then processed through hidden layers where patterns are identified before reaching the output layer and subsequently generates predictions or classifications. Convolutional neural networks are primarily used in image processing, detecting spatial hierarchies of patterns, and performing best with GPU support. Recurrent neural networks are well suited for handling sequential data, making them ideal for time-series analysis and language modeling. Deep neural networks with multiple hidden layers are used for highly complex, nonlinear problems. Given the complexity of polymer synthesis, characterization, and property prediction, DL will become increasingly integrated into polymer informatics, characterization, and experimental workflows to enhance efficiency and accuracy. Building an Autonomous-ML Research Using an SDL with Flow Chemistry For an SDL, other tasks such as the use of alternative reagents, storage, improved quality control, and data analytics can make experiments more seamless and faster. This operational system for an ML-driven experimental setup is best implemented with an automated, continuous-flow chemistry or reactor setup at the bench scale (Knox et al. 2025; Pittaway et al. 2025; Sumpter et al. 2023) (Figure 2). This is where most polymer chemists can work on different algorithms, test mechanistic hypotheses, demonstrate controlled kinetics, and enable plug flow (for solutions) and packed-bed (for heterogeneous catalyst supports) columns, all conducted under chemical engineering operation parameters. Building an SDL Requires a Team and Automation Technology Building a bench-scale setup requires skills in chemistry, chemical engineering, mechatronics, robotics, applied computing or programming, and device interfacing. By controlling flow rate (mass transport), reaction space engineering, leveraging microfluidics or microreactors, and regulating pressure, volume, and temperature per unit operation in time and space series, it is possible to leverage AI/ML-optimized protocols. By building monitoring stations or employing analytical and characterization methods under flow and in situ (e.g., online NMR, IR, Raman, UV-vis, ESR), data can be collected from multiple viewpoints in real time. The key to control is an edge server or computer that can both manage and process data gathered from monitoring stations and sensors. A feedback loop mechanism can integrate decision making and set or reset points. This edge server, when connected to high-performance computing platforms and potentially exascale supercomputers, enables the recreation of digital twins in real time. Cloud data storage is crucial in such circumstances for collecting and archiving the large amounts of data expected from this bench-scale chemistry operation. This approach is not only a vision for polymer chemistry but is also becoming increasingly advanced in the pharmaceutical and drug discovery industries. If fully implemented, it represents an auto-ML-driven discovery and research operation platform that will accelerate polymer science and engineering. By coupling robotics, programmable mechatronics, and additive manufacturing, this system offers a valuable platform for both academic and industrial research communities. We have been leading efforts in AI/ML-driven SDL development for polymers at Oak Ridge National Laboratory (ORNL 2025) and are part of the lead project with the INTERSECT initiative. Many other national laboratories and academic institutions are also pursuing similar efforts in materials research (Ferreira Da Silva et al. 2024). Challenges remain in integrating SDLs into materials and polymer research laboratories. These include lack of standards and integration of instruments with different application programming interfaces (APIs) and their ability to communicate with a standard operating system and edge server; real-time measurements and instrumentation that are closer to the chemistry of chemical intermediates or transition states for kinetic and thermodynamic control; hypothesis development to help researchers appreciate simulation methods and the hierarchy that relates them to real or multiphase environments, beyond vacuum or homogeneous conditions; and costs of scaling SDLs for manufacturing compared with laboratory bench-scale setups. The last of these may be the most immediate to overcome if SDLs are to be more widely utilized for materials and polymer research. Recently, some articles have emphasized user-automation infrastructures (Pelkie et al. 2025) and the democratization of SDLs through decentralization (Bayley et al. 2024). Another recent paper focused on the use of low-cost 3D printing for laboratory automation, integrated with ML and AI algorithms to enable flexible and affordable SDLs (Doloi et al. 2025). Replicating an Expert Polymer Scientist Is Still No Easy Task An AI-driven SDL will be required for seamless integration of AI, automation, and laboratory workflows in polymer science and engineering, extending beyond current human capabilities and addressing the immense combinatorial complexity in polymers. The use of both ML and DL will unlock new possibilities by predicting material properties, designing novel polymers, and optimizing synthesis conditions with high efficiency. It will also accelerate formulation development with additives and optimize processes for cost-effective manufacturing. Specifically, Reinforcement Learning (RL), a distinct type of ML in which models learn by interacting with an environment and receiving rewards for taking optimal actions, will be essential. RL algorithms can be complex and are designed to discover optimal strategies, making them particularly suited for decision making. They can be powerful tools for optimizing polymerization processes and autonomous experimental control, bringing SDLs closer to duplicating human heuristics and scientific expertise. It is time to rethink training and education for future polymer scientists and engineers. Many researchers, while intrigued by AI’s potential, are overwhelmed by its complexity due to a lack of training and experience. One immediate step is implementing more agentic AI in SDLs. This can enhance the pace of scientific discovery by automating repetitive, time-consuming tasks. It is not necessary to automate the entire research or discovery process; instead, embedding more agentic and proactive AI can lead to faster decision making and automation of tasks. Examples include scripting peak analysis of spectra in real time, real-time optimization using specific algorithms, and independent command-control protocols on instrumentation (Flores-Leonar et al. 2020). By embedding agentic AI protocols, researchers will have more opportunities to focus on creative ideation and experimental planning, potentially reducing the costs associated with automation. Agentic AI can also be applied in experiment design, data analysis, and even report generation (Gridach et al. 2025). The goal, however, is not to replace humans in the lab but to assist them in accomplishing tasks and connect them with expertise. Will SDLs displace jobs for polymer scientists and engineers? Perhaps, but learning to utilize AI/ML tools and maintaining a high level of skill and experience will remain indispensable. Ultimately, the “expert-in-the-loop” and human judgment will be difficult to replicate. Going Back to the Plastics Economy In our current linear plastic economy, we are burdened with the dual task of cleanup and transitioning to a more circular economy (DOE 2023). A key objective for more ML-driven R&D and the use of SDLs is to capitalize on opportunities for sustainable polymer synthesis, enabling the production of chemical intermediates and monomers through optimized bio-refineries and harnessing the potential of enzyme catalysis in polymer synthesis. The use of renewable resources, such as biomass, agro-waste, or even CO2 as part of the upstream raw materials has gained traction (Sun and Wang 2024). One possibility is to utilize AI/ML in designing pathways for polymer construction and deconstruction, with an emphasis on chemical upcycling (Figure 3). Mixed plastics can be converted to key oligomeric intermediates and incorporate CO2 as a key monomer, resulting in a polymer that may also be vitrimeric in nature and reformed with dynamic bonds. Controlling phase separation and bond reformation can yield new recyclable attributes and improved thermo-mechanical properties. AI will fundamentally transform how polymer research is conducted, shifting from empirical iteration to data-driven hypothesis generation and informed decision making. It remains essential, however, to develop new chemistry and processing methods that are designed from the outset (polymer genome) to achieve higher performance and upcycling potential. If pursued with the tools of AI/ML and SDL, this approach offers the possibility to mitigate some of the irreversible damage to the environment and food chain (e.g., microplastics). Reducing packaging waste and single-use plastics may be achieved by redesigning the life cycle of polymer materials. Looking ahead, it will be essential to strive for multiple uses of polymeric materials—from design and production to consumption, repair, reuse, and recycling. Encouragingly, numerous programs and projects are already underway in institutions and national laboratories across the United States, Europe, and Asia, with clear goals for both the upstream and downstream aspects of the plastics economy. Final Observations In conclusion, AI/ML-driven SDLs can significantly accelerate the discovery of new polymer materials and enhance process and formulation optimization. The convergence of AI and polymer science will unlock new possibilities, from predictive modeling to autonomous research workflows. As AI-driven methodologies mature, the next frontier will not only involve discovering novel macromolecules but also designing the algorithms that enable their discovery—generative AI. Moving forward, RL is expected to play a larger role in self-optimizing polymerization processes and unraveling complex composition–processing–property relationships. Rather than simply accelerating workflows, AI will fundamentally transform how polymer research is conducted, shifting from empirical iteration to data-driven hypothesis generation and informed decision making. SDLs will be essential tools in this transformation. With advances in robotic synthesis, in situ characterization, and self-learning models, these labs will accelerate research in real time, dynamically adjusting synthesis conditions to optimize properties with minimal human intervention. This shift will redefine the role of polymer scientists, moving from experiment control to precision-driven material design and intervention. By reducing experimental time, SDLs can enable the simultaneous and continuous execution of multiple experiments, facilitating the solution of complex problems. Optimizing experimental parameters with digital twins and AI/ML algorithms can help identify the optimal conditions for achieving desired material properties. And LLMs with RAGs can accelerate the identification of valuable training data and experimental protocols, enabling the exploration of vast chemical and macromolecular spaces to determine the most efficient synthesis routes and processing or recycling strategies. The time has come to deploy more AI/ML tools, agentic AI, and SDLs in polymer science and engineering. References Abolhasani M, Kumacheva E. 2023. The rise of self-driving labs in chemical and materials sciences. Nature Synthesis 2:483–92. Advincula R, Yuan Y, Wang H, Wang Q, Chen J. 2025. Bridging additive manufacturing and electronics printing in the age of AI. Nanomaterials 15(11):843. Bayley O, Savino E, Slattery A, Noel T. 2024. Autonomous chemistry: Navigating self-driving labs in chemical and material sciences. Matter 7:2382–98. Beaucage P, Sutherland D, Martin T. 2024. Automation and machine learning for accelerated polymer characterization and development: Past, potential, and a path forward. Macromolecules 57(18):8661–70. Bzdok D, Altman N, Krzywinski M. 2018. Statistics versus machine learning. Nature Methods 15:233–34. Chen J, Yuan Y, Ziabari AK, Xu X, Zhang H, Christakopoulos P, Bonnesen PV, Ivanov IN, Ganesh P, Wang C, Jaimes KP. 2024. AI for manufacturing and healthcare: A chemistry and engineering perspective. arXiv arXiv:2405.01520. Collias D, James M, Layman J. 2021. Circular Economy of Polymers: Topics in Recycling Technologies. Washington, DC: American Chemical Society; ACS Publications. Das A, Ringu T, Ghosh S, Pramanik N 2023. A comprehensive review on recent advances in preparation, physicochemical characterization, and bioengineering applications of biopolymers. Polymer Bulletin 80(7):7247–312. DOE [Department of Energy]. 2023. Strategy for Plastics Innovation. Online at https://www.energy.gov/sites/default/files/2023-01/DOE- strat-for-plastics-innova_1-19-23.pdf. Doloi S, Das M, Li Y, Cho ZH, Xiao X, Hanna JV, Osvaldo M, Tat LN. 2025. Democratizing self-driving labs: Advances in low-cost 3D printing for laboratory automation. Digital Discovery 4(7):1685–721. Ferreira Da Silva R, Moore R, Mintz B, Advincula R, Al Najjar L, and 27 others. 2024. Shaping the Future of Self-Driving Autonomous Laboratories Workshop. Technical Report, ORNL/TM-2024/3714. Online at https://www.osti.gov/servlets/purl/2481197. Flores-Leonar MM, Mejía-Mendoza LM, Aguilar-Granda A, Sanchez-Lengeling B, Tribukait H, Amador-Bedolla C, Aspuru-Guzik A. 2020. Materials acceleration platforms: On the way to autonomous experimentation. Current Opinion in Green and Sustainable Chemistry 25:100370. Gartner T, Jayaraman A. 2019. Modeling and simulations of polymers: a roadmap. Macromolecules 52(3):755–86. Ghafarollahi A, Buehler M. 2025. SciAgents: Automating scientific discovery through bioinspired multi-agent intelligent graph reasoning. Advanced Materials 37(22):2413523. Gridach M, Nanavati J, El Abidine K, Mendes L, Mack C. 2025. Agentic AI for scientific discovery: A survey of progress, challenges, and future directions. arXiv. arXiv:2503.08979. Knox ST, Wu KE, Islam N, O’Connell R, Pittaway PM, Chingono KE, Oyekan J, Panoutsos G, Chamberlain TW, Bourne RA, Warren NJ. 2025. Self-driving laboratory platform for many-objective self-optimisation of polymer nanoparticle synthesis with cloud-integrated machine learning and orthogonal online analytics. Polymer Chemistry 16(12):1355–64. Kuenneth C, Schertzer W, Ramprasad R. 2021. Copolymer informatics with multitask deep neural networks. Macromolecules 54(13):5957–61. Lin T, Coley C, Mochigase H, Beech H, Wang, W, Wand Z, Woods E, Craig S, Johnson J, Kalow J, Jensen K, Olsen B. 2019. BigSMILES: A structurally-based line notation for describing macromolecules. ACS Central Science 5(9):1523–31. Lodge T, Hiemenz P. 2020. Polymer Chemistry (3rd ed.). CRC Press. Nobel Prize Organisation. n.d. All Nobel Prizes. Online at https://www.nobelprize.org/prizes/lists/all-nobel-prizes. ORNL (Oak Ridge National Laboratory). 2025. Creating the laboratory of the future. Online at https://www.ornl.gov/file/labs-future/display. Pelkie B, Baird S, Aissi E, Aspuru-Takata K, Cao Y, and 22 others. 2025. Democratizing self-driving labs through user-developed automation infrastructure. Working Paper, Version 1. ChemRxiv 1–18. Pittaway PM, Knox ST, Cayre OJ, Kapur N, Golden L, Drillieres S, Warren NJ. 2025. Self-driving laboratory for emulsion polymerization. Chemical Engineering Journal 507:160700. Rubinstein M, Colby R. 2003. Polymer Physics. Oxford University Press. Online at https://doi.org/10.1093/oso/9780198520597.001.0001. Saldívar-Guerra E, Vivaldo-Lima E. 2013. Handbook of Polymer Synthesis, Characterization, and Processing. John Wiley & Sons. Seymour R, Mark H, Pauling L, Fisher C, Stahl A, Sperling L, Marvel C, Carraher C. 1989. Pioneers in Polymer Science. Springer Science & Business Media. Sumpter B, Hong K, Vasudevan R, Ivanov I, Advincula R. 2023. Autonomous continuous flow reactor synthesis for scalable atom-precision. Carbon Trends 10:100234. Sun C, Wang Y. 2024. Catalytic upcycling plastics with CO2: A novel strategy for carbon waste valorization. Chem Catalysis 4(3):100934. Tom G, Schmid SP, Baird SG, Cao Y, Darvish K, Hao H, Lo S, Pablo-García S, Rajaonson EM, Skreta M, Yoshikawa N. 2024. Self-driving laboratories for chemistry and materials science. Chemical Reviews 124(16):9633–732. van Saarloos W, Vitelli V, Zeravcic Z. 2024. Soft Matter: Concepts, Phenomena, and Applications. Online at https://softmatterbook.online. Vittoria A, Urciuoli G, Costanzo S, Tammaro D, Cannavacciuolo FD, Pasquino R, Cipullo R, Auriemma F, Grizzuti N, Maffettone PL, Busico V. 2022. Extending the high-throughput experimentation (HTE) approach to catalytic olefin polymerizations: From catalysts to materials. Macromolecules 55(12):5017–26. Wheatle B, Fuentes E, Lynd N, Ganesan V. 2020. Design of polymer blend electrolytes through a machine learning approach. Macromolecules 53(21):9449–59. Yu S, Ran N, Liu J. 2024. Large-language models: The game-changers for materials science research. Artificial Intelligence Chemistry 2(2):100076. About the Author:Rigoberto Advincula is Governor’s Chair of Advanced and Nanostructured Materials and Group Leader at Oak Ridge National Laboratory and the University of Tennessee, Knoxville.