Spectral cullet classification in the mid-infrared field for ceramic glass contaminants detection.

The presence of glass-like contaminants inside waste glass products, usually resulting from both industrial and differentiated urban waste collection, has greatly increased in recent years, due to the introduction to the market of a large amount of goods manufactured from ceramic glass. The presence of contaminants in the glass recycling streams reduces product quality and increases production costs. The detection of ceramic glass detection is an unresolved problem, as such material looks like normal glass and can only be detected by trained personnel. In this study an innovative approach to ceramic glass recognition, based on the spectral signature in the mid-infrared (MIR) field, was proposed and investigated. The study specifically addressed the spectral characterization of glass and ceramic glass fragments collected in a real recycling plant from two different production lines: coloured container glass and white container glass. To define suitable inspection strategies to separate the useful (glass) from the polluting (ceramic glass) materials at the recycling plants, fragments presenting different colour, thickness, size, shape and manufacturing were selected. Both dirty and clean cullet was considered. The analyses, carried out in the MIR spectral field (2280-4480 nm), show that ceramic glass and glass fragments can be recognized according to their different spectral signature. In particular, by selecting a specific wavelength ratio the two classes of materials can be rapidly recognized, suggesting the possibility of developing an integrated hardware and software sorting system for 'on-line' ceramic glass separation.