A New Chemometric Approach for Automatic Identification of Microplastics from Environmental Compartments Based on FT-IR Spectroscopy

One key step studying
interactions of microplastics with our ecological
system is to identify plastics within environmental samples. Aging
processes and surface contamination especially with biofilms impede
this characterization. A complex and time-consuming cleaning procedure
is a common solution for this problem. However, it implies an artificial
change of sample composition with a risk of losing important information
or even damaging microplastic particles. In the present work, we introduce
a new chemometric approach to identify heavily weathered and contaminated
microplastics without any cleaning. The main idea of this concept
is based on an automated curve fitting of most relevant vibrational
bands to calculate a highly characteristic fingerprint that contains
all vibrational band area ratios. This new data set will be used to
estimate the similarity of samples and reference standards for identification.
A total of 300 individual naturally weathered plastic particles were
measured with Fourier transformation infrared spectroscopy
in attenuated total reflection mode (FT-IR ATR) and identified successfully
with the new method. To that end, all samples were compared with a
selection of common reference plastics and bio polymers. As it turns
out, the accuracy of identification rises significantly from 76% by
means of conventional library searching algorithms to 96% by identifying
microplastics with our new method. Therefore, the new approach can
be a useful tool to compare and describe similarities of FT-IR spectra
of microplastics, which may improve further research studies on this
topic.