Full length article
Alterations of plastics spectra in MIR and the potential impacts on identification towards recycling

https://doi.org/10.1016/j.resconrec.2020.104980Get rights and content

Highlights

  • FTIR-ATR was studied to assess MIR HSI theoretical ability to sort plastics.

  • Limited resolution and SNR can strongly affect identification.

  • With >20w% carbon black loading, identification is harder but still possible.

  • Common fillers can be spotted and recognized on plastic spectra.

  • Some flame retardants are easily spottable and can disrupt identification.

Abstract

Plastic recycling is mainly limited by their sorting as their natures, forms and formulation are very numerous and most of them are strongly incompatible, leading to poor mechanical properties. Several industrial sorting technologies exist, and others are in development. However, each of them has drawbacks. Especially, NIR-HSI (Near-Infrared Hyperspectral Imagery) is limited by the use of carbon black, mainly as a pigment and UV agent in the case of thermoplastics. MIR-HSI (Mid-Infrared) could be a suitable and viable alternative to resolve this issue. Hence, this work, based on laboratory FTIR-ATR (Fourier-Transform Infrared Attenuated Total Reflection), focuses on possible sources of spectral alteration, which could impair identification of usual polymers using industrial MIR-HSI. It aims to help simple and rapid laboratory characterization and give tools to avoid misidentification or enable specific segregation during industrial sorting. First, acquisition parameters were degraded to simulate those imposed by industrial conditions: short acquisition time, diminished resolution and blank defaults. Then, impact on formulations of usual WEEE (Waste of Electric and Electrical Equipment) plastics were evaluated, with PE, PP, ABS and HIPS as matrices, and carbon black (at different concentrations), calcite, talc, titanium oxide and some flame retardants as additives. Several patterns found in homemade standard samples were recognized within a stock of about one hundred of real waste samples.

Introduction

Plastic recycling is viewed as one of the most interesting ways to deal with environmental problems linked to both their production and disposal (Wäger and Hischier, 2015; WRAP, 2008). Nevertheless, reaching interesting properties, as processability, aspect or mechanical performance, is very difficult to achieve as plastics natures, additives, fillers, forms and applications are incredibly diverse. Most of them are non-miscible and even incompatible (Maris et al., 2018), since their blends display functional properties inferior to each of them taken separately. An efficient and accurate sorting of end-of-life plastics is thus primordial for recycling to be economically pertinent (Vrancken et al., 2017). However, end-of-life management itself also entails ecological impacts. For instance, these plastics needs to be cleaned and thus need water and sometimes detergents. Also, collection and, transport can also bear a severe carbon footprint, especially if these actions are geographically scattered (WRAP, 2008).

Several types of separation technologies exist. They can be classified in two different groups: direct and indirect (Gundupalli et al., 2017). When separation is intrinsically due to property differences, it is direct. It is the case with sink-float tanks (Gent et al., 2009) where sorting is based on density differences. Other examples are froth flotation for surface tension (Wang et al., 2015), trommels for granulometry (Ashkiki et al., 2019), magnetic overhead belts and Eddy current for (para)magnetic properties (Krivtsova et al., 2009). Sorting is indirect when identification and separation are performed through two different steps (Gundupalli et al., 2017). Generally, these are optical technologies. Items to be sorted are moved on a conveyor belt where a sensor acquires their spectral signature after adequate excitation. Then, these items are physically sorted, usually through the use of compressed air nozzles.

The transposition to industrial dynamical conditions of several static technologies, used at laboratory scale, such as UV-Vis (Gundupalli et al., 2017), NIR (Near-Infrared) (Beigbeder et al., 2013; Huth-Fehre et al., 1995; Leitner et al., 2003), MIR (Mid-Infrared) (Bae et al., 2019; Becker et al., 2017; Rozenstein et al., 2017), Raman spectroscopies (Bae et al., 2019; Florestan et al., 1994; Roh et al., 2017), LIBS (Laser Induced Breakdown Spectroscopy) (Barbier et al., 2013; Grégoire et al., 2011), X-ray transmission or fluorescence (Hasan et al., 2011; Mesina et al., 2007) is currently studied. Laboratory analysis generally explores a single sample at a time, on a relatively small spot, can take from several seconds to hours, can be sometimes made with contact to the sample. However, industrial identification toward waste sorting is radically different. Objects are generally scattered along a high-speed conveyor belt and their identification must be made remotely within milliseconds to enable sorting at the end of the conveyor. This leads to strong technological differences, which, with the harsh acquisition conditions, often lead to difficult-to-analyze signals. Available data, spatial and spectral resolutions, signal-to-noise ratio (SNR) can be, at first glance, insufficient to enable identification. This requires signal processing and use of advanced classification algorithms (Roh et al., 2017; Spetale et al., 2016; Tachwali et al., 2007; van den Broek et al., 1998). However, a real physicochemical understanding of the differences at the source of classification is needed to avoid false identifications.

Additives and fillers can also have repercussions on sorting. As polyolefins are the main polymers with intrinsic densities below 1 g/cm3, they can be recovered thanks to tap water sink-float. Sadly, an important fraction is filled with calcite (calcium carbonate, CaCO3) or talc (Aluminum Silicate, Al2Si2O5(OH)4) and thus sinks (Maris et al., 2015). Peeters et al. (Peeters et al., 2014) highlighted density overlapping because of the use of flame retardants and styrenic blends, making density separation inefficient. In the frame of WEEE (Waste Electrical and Electronic Equipment) legislation and RoHS (Regulation of Hazardous Substances), numerous widely spread halogenated flame retardants (FR) are now forbidden because of their toxicity. Several other widely used FRs could also become forbidden in the future (Delva et al., 2018; Vilaplana et al., 2008). This imposes sorting materials according to these additives for recycled materials to stay below legal thresholds (Hennebert and Filella, 2018). X-ray transmission and fluorescence are among the most promising technologies towards this goal (Gallen et al., 2014; Kuang et al., 2018; Sharkey et al., 2018). Even in a dual energy configuration, which is well developed and perfectly suitable to metals (Mesina et al., 2007), transmission lacks selectivity between heavy elements. In the case of additive plastics, both concentration and thickness are unknowns in the Beer-Lambert law whereas concentration was assumed 100% in the case of metals. As it needs to detect way more wavelengths, fluorescence is technologically more challenging and thus more expensive. Solvent extraction of FRs or total dissolution could be very efficient (Grause et al., 2015; Vilaplana et al., 2009; Zhao et al., 2018), although transposition to industrial scale can be difficult, especially with solvent management. Pyrolysis, towards energy recovery and/or chemical recycling (Ma et al., 2016; Yang et al., 2013) could be seen as the only viable solution, even if rather controlled waste stocks (thus sorted) are needed not to disrupt the process and its efficiency too much. Even at laboratory scale, precise identification and quantification of flame retardants can be rather burdensome compared to simple infrared spectra, especially with the need of appropriate extraction (Guzzonato et al., 2016a; Otake et al., 2015; Schlummer et al., 2005; Vilaplana et al., 2008). However, Puype et al. (Puype et al., 2019) recently showed that Direct Analysis in Real Time - High Resolution Mass Spectrometry (DART-HRMS) was adequate to rapidly and accurately identify brominated FR at laboratory scale.

Another very common additive, carbon black, which is the most common way to color plastic in black or grey, strongly absorbs in NIR (Beigbeder et al., 2013; Huth-Fehre et al., 1995; Serranti et al., 2012), making NIR-HSI unable to sort dark colored plastics. Its absorption is easily explained by the almost infinite unsaturations conjugation within its graphitic structure (Kang et al., 2016), also explaining its deep black color as it absorbs in the visible range. It also absorbs beyond these boundaries in NIR but also in UV (Allen et al., 1998; Liu and Horrocks, 2002), making it an interesting additive to protect polymeric materials from photodegradation. Consequently, it also impacts Raman spectra more or less depending on the excitation wavelength (Bokobza et al., 2013; Yamaji et al., 2013). More generally, carbon black is used for coloration and UV protection from 0.5 to 2.0 wt. % (Turner, 2018), up to 20 wt. % for electrical conductivity where percolation is necessary (Probst et al., 2009; Zhou et al., 2006) and up to 50 wt. % for mechanical reinforcement (Kang et al., 2016; Li et al., 2019), especially in elastomers as in tires. As NIR-HSI, the most used technology to finely discriminate plastics according to their natures, is limited with dark plastics, several alternative sorting technologies are subject to extensive research (Grégoire et al., 2011; Huang et al., 2017; Küter et al., 2018; Langhals et al., 2014; Roh et al., 2017; Wang et al., 2015; Zhao et al., 2018; Zhao et al., 2018). MIR-HSI is one of them (Kassouf et al., 2014; Rozenstein et al., 2017; Signoret et al., 2019a, 2019b). It was chosen in the present study because this technology begins to be commercially available. Also, Fourier-Transform Infrared (FTIR), its laboratory equivalent as it works within the MIR range, was used for a long time for polymer analysis.

The present study aims to complete our previous works on spectral identification of polymers through the use of FTIR-ATR (Attenuated Total Reflection) with the scope of transposition to MIR-HSI. The first study described intrinsic signals of styrenics polymers and their blends (Signoret et al., 2019a). The second one focused on polyolefins and polymers with close spectra, namely PVC and POM (Signoret et al., 2019b). A third one was about spectral alterations due to ageing of LDPE, PPH, HIPS, ABS and PC in the scope of their identification (Signoret et al., 2020). Whereas these previous works focused on characteristic signals of polymers, the subject here is to anticipate how these patterns can evolve because of formulation: carbon black, common mineral fillers and flame retardants. The objective is to determine if loaded polymers could still be recognizable, even in degraded acquisition conditions. Also, characteristic patterns related to formulation can be useful to quickly identify additives, at least at laboratory scale, instead of using a heavier technique as ICP (Inductively Coupled Plasma) or X-ray fluorescence which can be more time and money consuming. However, accurate quantification needs another technique, namely thermogravimetric analysis (TGA).

Section snippets

Materials

Several polymers were used for formulated standard preparation: ABS reference was Terluran GP22 (Styrolution), HIPS was Polystyrol 485I (Repsol), ABS/PC was Bayblend T85 XF (Covestro), PE was HDPE Alcudia 4810 (Sabic), PP was PPC 83MF10 and PPH 505P (Sabic). Carbon black used in this study was Elftex 570 (Cabot). Several mineral fillers references were also involved: XP12-5630 kaolin (IMERYS), Omya BL calcite (Omya), Luzenac HAR T84 talc (IMERYS) and Aeroxide TiO2 P 25 titanium dioxide

Background/blank default: Atmospheric IR absorbent species - CO2 & H2O

CO2 and H2O are more and more visible as background/blank is too old compared to current acquisition and as intrinsic signals from the polymer are weak as seen on Fig. 1. As all analyzed surfaces were first cleaned with ethanol and left to dry, observed water signals are probably corresponding to atmospheric water. The polymer corresponding to this spectrum can still be identified as calcite loaded PVC. Calcite specific signals are described in Part 3.3. PVC is mainly identified through its 700

Conclusions

Nowadays, the industrial sorting of dark-colored End-of-Life plastics is an obstacle to their recycling, as one of the most spread sorting technologies, Near-Infrared Hyperspectral Imagery (NIR-HSI) is unable to perform this task. Mid-Infrared HSI (MIR-HSI) is one the potential answers to this issue.. Consequently, its laboratory equivalent, Fourier-Transform Infrared spectroscopy (FTIR), was here studied to evaluate the theoretical limits of its industrial equivalent. Possible dangers of

Funding

This work was supported by BPI France via the FUI 20 (Fonds unique interministériel) grant.

CRediT author statement

Charles Signoret (PhD student): Writing - original draft, Conceptualization, Methodology, Investigation and rewriting revision manuscript.

All the authors: Visualization, Investigation and validation.

Didier Perrin & Patrick Ienny: Supervision

Didier Perrin: Editing, Data curation, Reviewing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The authors would like to express their appreciation toward Benjamin Gallard, Robert Lorquet, Loïc Dumazert and Jean-Claude Roux from C2MA for technical support, respectively for polymer processing, spectroscopy experiments, thermal characterization and SEM/EDX analysis. Marc-Adrien Tronche, PhD student at ENSCM, is also thanked for his advice about carbon black managing. Suez and Pellenc ST are gratefully acknowledged for partnership in this work. Finally, the authors are very thankful to

References (70)

  • S.P. Gundupalli et al.

    A review on automated sorting of source-separated municipal solid waste for recycling

    Waste Manag

    (2017)
  • A. Guzzonato et al.

    Improving the accuracy of hand-held X-ray fluorescence spectrometers as a tool for monitoring brominated flame retardants in waste polymers

    Chemosphere

    (2016)
  • A.R. Hasan et al.

    Online sorting of recovered wood waste by automated XRF-technology: Part II. Sorting efficiencies

    Waste Manag.

    (2011)
  • P. Hennebert et al.

    WEEE plastic sorting for bromine essential to enforce EU regulation

    Waste Manag

    (2018)
  • M.Z. Hossain et al.

    Measurement of enthalpies of association between CO2 and polymers via in situ ATR-FTIR spectroscopy

    J. Supercrit. Fluids

    (2014)
  • T. Huth-Fehre et al.

    NIR - Remote sensing and artificial neural networks for rapid identification of Post Consumer Plastics

    J. Mol. Struct.

    (1995)
  • A. Kassouf et al.

    Rapid discrimination of plastic packaging materials using MIR spectroscopy coupled with independent components analysis (ICA)

    Waste Manag

    (2014)
  • J. Kuang et al.

    Brominated flame retardants in black plastic kitchen utensils: concentrations and human exposure implications

    Sci. Total Environ.

    (2018)
  • R. Leitner et al.

    Real-time classification of polymers with NIR spectral imaging and blob analysis

    Real-Time Imaging

    (2003)
  • M. Liu et al.

    Effect of Carbon Black on UV stability of LLDPE films under artificial weathering conditions

    Polym. Degrad. Stab.

    (2002)
  • X. Liu et al.

    Sodium persulfate-assisted mechanochemical degradation of tetrabromobisphenol A: efficacy, products and pathway

    Chemosphere

    (2016)
  • C. Ma et al.

    Chemical recycling of brominated flame retarded plastics from e-waste for clean fuels production: a review

    Renew. Sustain. Energy Rev.

    (2016)
  • E. Maris et al.

    Characterizing plastics originating from WEEE: A case study in France

    Miner. Eng.

    (2015)
  • J. Maris et al.

    Mechanical recycling: Compatibilization of mixed thermoplastic wastes

    Polym. Degrad. Stab.

    (2018)
  • M.B. Mesina et al.

    Automatic sorting of scrap metals with a combined electromagnetic and dual energy X-ray transmission sensor

    Int. J. Miner. Process.

    (2007)
  • J.R. Peeters et al.

    Closed loop recycling of plastics containing Flame Retardants

    Resour. Conserv. Recycl.

    (2014)
  • N. Probst et al.

    Compounding with conductive carbon black

    Plast. Addit. Compd.

    (2009)
  • F. Puype et al.

    Evaluation of Direct Analysis in Real Time – High Resolution Mass Spectrometry (DART-HRMS) for WEEE specific substance determination in polymeric samples

    Chemosphere

    (2019)
  • O. Rozenstein et al.

    Development of a new approach based on midwave infrared spectroscopy for post-consumer black plastic waste sorting in the recycling industry

    Waste Manag

    (2017)
  • M. Schlummer et al.

    Analysis of flame retardant additives in polymer fractions of waste of electric and electronic equipment (WEEE) by means of HPLC-UV/MS and GPC-HPLC-UV

    J. Chromatogr. A

    (2005)
  • S. Serranti et al.

    Classification of polyolefins from building and construction waste using NIR hyperspectral imaging system

    Resour. Conserv. Recycl.

    (2012)
  • M. Sharkey et al.

    Portable X-ray fluorescence for the detection of POP-BFRs in waste plastics

    Sci. Total Environ.

    (2018)
  • C. Signoret et al.

    MIR spectral characterization of plastic to enable discrimination in an industrial recycling context: I

    Specific case of styrenic polymers. Waste Manag.

    (2019)
  • C. Signoret et al.

    MIR spectral characterization of plastic to enable discrimination in an industrial recycling context: II. Specific case of polyolefins

    Waste Manag.

    (2019)
  • C. Signoret et al.

    MIR spectral characterization of plastic to enable discrimination in an industrial recycling context: III. Anticipating impacts of ageing on identification

    Waste Manag.

    (2020)
  • Cited by (26)

    View all citing articles on Scopus
    View full text