The value of recycled materials depends on the fraction purity. As an example, a 1000 bottle PET batch is useless if it contains 1 PVC bottle. In practice, there will be an economically motivated trade off between fraction purity, recovery, and capacity. In any case, the value in waste is relatively low so instrumentation cost must be kept down.

The first step in sorting is material classification. This involves probing the material physical properties for positive identification of particles to be recovered. This is a challenge under partly cumbersome industrial conditions and at high speed. However, optics offers a range of methods for material characterization that can be applied for fast measurements. In addition, or in combination, machine vision systems and data analysis methods can be used to extract more information for classification and sorting control.

SINTEF has helped develop low capacity classification systems for Tomra Systems reverse vending machines. In this case there is plenty of measurement time to comply with the very high classification requirements, but the challenge has been to implement laboratory techniques at lowest possible cost. Solutions include low cost filter wheels and DOE spectrometers for material classification, as well cameras and advanced image analysis for shape, bar code reading and anti-cheating systems, in combination with novel optical configurations.

For Titech Visionsorts waste sorting machines the challenge was meeting high capacity requirements; sorting 2-6 tons pr hour, at 10 000 measurements pr. sec, 24 hours a day, requires robust "working horse" technology. The key to the high performance and long lifetime of these machines lies in a relatively simple dedicated measurement system where every possible detail has been optimised to gain SNR.

Much of the technology SINTEF has developed for recycling has found use in other applications.

Some available techniques:

• NIR reflection and transmission measurements for sorting of plastics and other materials
• Specific colour measurement to separate paper from different printing processes
• Fluorescence measurements for identification of bacteria and contaminants
• Chemical analysis using Laser induced breakdown spectroscopy (LIBS) for e.g. Aluminum alloy composition determination
• Sorting of shredded Aluminum particles using shape, texture, and classification techniques.