Original source (on modern site)
Petrochemical plastics are lightweight, durable and inexpensive, enabling almost ubiquitous applications. However, less than 10% of petrochemical plastics can be recycled, and nearly 80% of used plastics end up in landfills or pollute the environment, resulting in global plastic pollution. One promising solution is to use natural components to develop sustainable, biodegradable plastic substitutes, which can attenuate the magnitude of plastic waste and prevent the release of microplastics. However, discovering biodegradable alternatives that meet specific property criteria, such as optical transparency, fire retardancy and mechanical resilience, presents substantial challenges. Current approaches rely on trial-and-error experiments and probe a broad range of parameters in a scattershot manner. Machine learning (ML) is a form of artificial intelligence (AI) that constructs a model to make predictions or recommendations across multiple degrees of freedom. Recently, AI/ML has benefitted the fields of organic/inorganic catalyst design, drug discovery, and quantum dot synthesis, where simulation tools or high-throughput analytical platforms can supply many high-quality data points. In contrast, substantial obstacles exist in constructing a high-accuracy prediction model for biodegradable plastic substitutes, as acquiring high-quality data points is time-consuming and labour-intensive. In a report titled "Machine intelligence-accelerated discovery of all-natural plastic substitutes," an integrated workflow that uses robotics and AI/ML predictions was realized to accelerate the discovery of all-natural plastic substitutes with programmable optical, thermal, and mechanical properties. The study demonstrated two-way design tasks by harnessing the model's predictive power, including (1) accurately predicting multiple characteristics of an all-natural nanocomposites from its composition and (2) automatically suggesting suitable biodegradable plastic alternatives with user-designated features. The report proposes a hybrid approach involving robot-assisted experiments, data science, and simulation tools. This approach offers an unconventional design platform to accelerate the invention of eco-friendly, biodegradable plastic substitutes from the GRAS database. The report states that the champion model accurately predicted the optical transmittances, fire resistances and stress-strain curves of multiple all-natural nanocomposites, which matched the experimental results well. By inputting all possible compositions within the feasible design space, the champion model produced a set of 3D heatmaps that visually represented the spatial distributions of all property labels, including thickness. The champion model, with its high prediction accuracy, was adopted to accelerate the discovery of high-strength structural materials using natural building blocks. A model expansion method was applied due to its excellent antimicrobial activity and biocompatibility properties to enrich the portfolio of all-natural plastic substitutes. The champion model was employed to automate the inverse design of all-natural plastic substitutes with programmable physicochemical characteristics, demonstrating the power of multi-property prediction. An unconventional design platform that utilized automated robots, machine intelligence, wet-lab experiments and simulation tools was developed to discover a library of all-natural nanocomposites as biodegradable plastic substitutes with programmable optical, fire-resistant and mechanical properties. This ML/robotics-integrated workflow stimulates the development of various functional materials with multi-property optimization, which can be applied to a wide range of nanoscience fields, including tactile sensors, stretchable conductors, electrochemical electrolyte optimization, and thermal insulative aerogels. Still, several ongoing challenges and limitations exist associated with the AI/ML-integrated workflow for the accelerated design of all-natural plastic substitutes. First, no available collaborative robotics systems can automate all-natural nanocomposites' entire preparation and characterization processes. Second, the quality of natural building blocks may vary from batch to batch. Third, data fusion with cost analyses and life cycle analyses into the champion model would be highly beneficial, allowing for identifying the optimal all-natural plastic substitutes that meet desired properties and providing the benefits of cost saving and environmental impact reduction. Last, the end-of-life processing of all-natural plastic substitutes, which could be converted into biofuels or other valuable chemicals, has not been considered.Property prediction
Integrated workflow