Microplastics and Nanoplastics Analysis

When using machine learning for particle analysis you may start with imaging the MNPs on a filter via SEM. The software ZEISS ZEN Intellesis automatically identifies particles with a deep learning model on a SEM image.

The user trains the classification model by mouse clicks on a few images, indicating different objects and assigning particle types manually. The resulting model distinguishes particle properties like diameter, orientation or chemical properties after correlation with Raman spectroscopy results, e.g. different polymers like PVC, PA, PS, and PE.

Time-consuming segmentation steps on hundreds of images are done by machine learning algorithms (including Deep Learning). ZEN Intellesis Object Classification, a Python-powered tool offers pixel classification with real multichannel feature extraction and segmentation. By pre-trained networks even complex multidimensional, multi-modal data can be analyzed, regardless of their origin.

An alternative to using two separate systems is to perform Raman-SEM correlations with a fully integrated solution. RISE (Raman Imaging and Scanning Electron Microscopy) is the combination of WITec Confocal Raman Imaging with a ZEISS FE-SEM all in one instrument.

Combining the chemical sensitivity of Raman with the SEM allows you to analyze nanoplastic particles more efficiently, accurately and reliably than before. The sample with its regions of interest stays in the vacuum chamber, is transferred from one objective, SEM or Raman, to the other. The workflow is streamlined and convenient to use, from SEM imaging to Raman spectroscopy, delivering complementary information of the specimen - a chemical fingerprint.