Data Science and Artificial Intelligence for smart sustainable plastic

点击翻译按钮获取中文摘要

High-density polyethylene (HDPE) is a semi-crystalline polymer, it is this combination of crystalline and amorphous regions that provide the polymer with its attractive properties that result in the polymer's widespread use in day-to-day life. The crystals within the polymer contribute towards the stiffness, barrier properties and toughness while the amorphous regions provide flexibility, ductility and contribute towards toughness. HDPE can be collected and mechanically recycled to produce a post-consumer resin (PCR), promoting a circular economy with a plastic that has a lower environmental footprint compared to new, virgin plastic. The sorted packaging is mechanically recycled through grinding, washing and then extruding to produce PCR pellets.One of the major issues with using PCR to replace virgin plastic in packaging is that it is a variable material. It may contain different grades of plastic differing in molecular weight or flexural strength, it can be contaminated with other materials and the recycling process itself can lead to degradation of the plastic. Many of these changes influence the crystallinity in the polymer. In order to be able to more effectively reuse PCR HDPE we need to understand how crystal size, degree of crystallinity, tie molecule concentration control the performance of the polymer. One application area where HDPE is widely used is packaging, particularly for containing home/personal care products and food, with approximately 270,000 tonnes used annually. HDPE is an excellent packaging material due to its low density, chemical inertness, and toughness. It is a thermoplastic that can be heated and then processed into bottles by extrusion blow-moulding. A practical outcome of the project is to understand how the crystallinity changes in recycled HDPE and how this subsequently impacts the in-user properties of the polymer. This understanding could be used in inform the selection of PCR for use in packaging, ultimately increasing the sustainability of plastics in packaging.This PhD studentship will run alongside our project funded by the UK Industrial Strategy Challenge Fund in Smart Sustainable Plastic Packaging, and is co-funded by Unilever. We will use existing literature data on virgin plastics, combined with new experimental data from various in-house chemical techniques (FTIR/Raman/DSC/XRD) performed on recycled plastics, to try and predictively link the chemical structure of the plastic to its properties. These properties ultimately determine whether a sample of recycled plastic will be desirable for use in manufacturing. Therefore, if an accurate predictive model could be established, we would be able to save time and money and not have to put a recycled PCR through the manufacturing process before we could determine its properties. We will do this using cutting edge data science techniques in the space of Topological Data Analysis. This should allow us to understand relationships between certain chemical techniques and certain chemical properties, and since peaks in the spectra correspond to certain physical phenomena, we may be able to establish new structure-property relationships. We also aim to understand how contamination with other plastics, physical weathering and oxidation in the extrusion process impact the physical properties of the plastic. This will help us to understand the viability of the current plastic recycling stream if challenges to shift to 100% recycled plastic in manufacturing are to be met.