SEARCH

SEARCH BY CITATION

Keywords:

  • FTIR;
  • modeling;
  • polyolefins;
  • additives

Abstract

For plastics materials intended to be in contact with food, recent EU regulations, 72/2002/EC and 1935/2004/EC, enforce the assessment of the migration of 502 substances of the 932 positively listed substances. As mathematical modeling has been proposed to overcome such considerable effort in particular by providing maximum acceptable concentrations in the formulation, the compliance testing problem is efficiently reduced to the identification of substances and to their extent in initial materials. This work examines a fast identification and quantification procedure based on a semisupervised deconvolution procedure of FTIR spectra of polymer extracts in dichloromethane. The inversion procedure was implemented as a Tikhonov least-square problem and designed to work on large and open dictionary of substances by combining both spectra of reference additives and normalized responses of typical chemical functions. The sparsity of the overall solution was fulfilled with non-negativity constraints, while traces were detected by an iterative reweighting and stochastic resonance. The whole methodology was calibrated onto 21 typical additives of polyolefins and satisfactory tested on numerical examples and on extracts of processed films in high-density polyethylene including up to eight unknown compounds. Maps of possible confusions and biases were generated for all tested substances. The mass balance laws for molecules belonging to similar classes of additives were particularly highlighted. © 2010 Wiley Periodicals, Inc. J Appl Polym Sci, 2011