How does NIR technology identify plastics?

Near-Infrared (NIR) spectroscopy has become the workhorse of modern plastic sorting – and for good reason. I’ve seen firsthand how this technology can pick out a PET bottle from a moving conveyor belt at lightning speed, even when it’s sandwiched between HDPE containers. But what’s really fascinating is how NIR doesn’t just “see” plastics; it essentially “fingerprints” them by reading their molecular vibrations. When that infrared light hits a plastic surface, different polymer bonds absorb specific wavelengths like a molecular barcode. A PET molecule will soak up completely different NIR frequencies than PVC, for instance.

The science behind the scan

Here’s where it gets technical (but stay with me). The magic happens because each plastic type has unique C-H, N-H, and O-H bond vibrations that resonate at particular infrared frequencies. NIR sensors typically operate in the 900-1700nm range, capturing these subtle differences. In our recycling facility tests, we found that PET shows strong absorption around 1,650nm (thanks to its ester groups), while polypropylene peaks at 1,210nm. The scanner’s computer then matches these spectral signatures against a pre-loaded library – sort of like Shazam for plastics.

Real-world challenges and solutions

Now, it’s not all smooth sailing. Black plastics used to be the nightmare of every recycling operator because carbon black pigments absorb nearly all NIR light. But newer systems combine NIR with Raman spectroscopy or even terahertz imaging to crack this problem. There’s also the issue of contaminated or multilayered materials – I remember one case where a food pouch with PET and aluminum layers fooled our basic NIR system until we upgraded to hyperspectral imaging. The takeaway? While NIR is incredibly powerful, the best sorting systems today use it as part of a sensor fusion approach.

What really blows my mind is how far this technology has come. Early NIR sorters in the 1990s could maybe distinguish three plastic types with 80% accuracy. Today’s AI-enhanced systems can identify over a dozen polymer families with 95-98% precision – all while processing several tons per hour. That’s not just tech progress; that’s a revolution in sustainable material recovery.