L’addition de traceurs dans les polymères : Une nouvelle voie ......M. Alain MILLE 69622 VILLEURBANNE Cedex M. Alain MILLE Université Claude Bernard Lyon 1 LIRIS - INFOMATHS Bâtiment

N° d’ordre 2010-ISAL-0095 Année 2010 Thèse

L’addition de traceurs dans les polymères : Une nouvelle voie spectroscopique par fluorescence X, pour l'identification rapide et le tri des matériaux plastiques Présentée devant L’institut national des sciences appliquées de Lyon Pour obtenir Le grade de docteur Formation doctorale Ecole doctorale matériaux de Lyon Spécialité : Matériaux polymères Par Feliks Bezati Ingénieur physico-chimiste Soutenance prévue le 16 novembre 2010 Jury

ANTONINNI Gérard Professeur Examinateur BELLON-MAUREL Veronique Professeur Rapporteur CASSAGNAU Philippe Professeur Examinateur COUFFIGNAL Bénédicte Docteur Invité DELOBEL René Professeur Rapporteur FROELICH Daniel Professeur Co-directeur de thèse LEGOUPIL Samuel Docteur Examinateur MASSARDIER-NAGEOTTE Valérie Maître de conférences Co-directeur de thèse MARIOGE Cathrine Docteur Invité Ingénierie des Matériaux polymères, UMR 5223 Laboratoire des matériaux macromoléculaire Laboratoire Conception Produit Innovation

2

INSA Direction de la Recherche - Ecoles Doctorales – Quadriennal 2007-2010

SIGLE ECOLE DOCTORALE NOM ET COORDONNEES DU RESPONSABLE

CHIMIE

CHIMIE DE LYON http://sakura.cpe.fr/ED206 M. Jean Marc LANCELIN

Insa : R. GOURDON

M. Jean Marc LANCELIN Université Claude Bernard Lyon 1 Bât CPE 43 bd du 11 novembre 1918 69622 VILLEURBANNE Cedex Tél : 04.72.43 13 95 Fax : [email protected]

E.E.A.

ELECTRONIQUE, ELECTROTECHNIQUE, AUTOMATIQUE http://www.insa-lyon.fr/eea M. Alain NICOLAS Insa : C. PLOSSU [email protected] Secrétariat : M. LABOUNE AM. 64.43 – Fax : 64.54

M. Alain NICOLAS Ecole Centrale de Lyon Bâtiment H9 36 avenue Guy de Collongue 69134 ECULLY Tél : 04.72.18 60 97 Fax : 04 78 43 37 17 [email protected] Secrétariat : M.C. HAVGOUDOUKIAN

E2M2

EVOLUTION, ECOSYSTEME, MICROBIOLOGIE, MODELISATION http://biomserv.univ-lyon1.fr/E2M2 M. Jean-Pierre FLANDROIS Insa : H. CHARLES

M. Jean-Pierre FLANDROIS CNRS UMR 5558 Université Claude Bernard Lyon 1 Bât G. Mendel 43 bd du 11 novembre 1918 69622 VILLEURBANNE Cédex Tél : 04.26 23 59 50 Fax 04 26 23 59 49 06 07 53 89 13 [email protected]

EDISS

INTERDISCIPLINAIRE SCIENCES-SANTE Sec : Safia Boudjema M. Didier REVEL Insa : M. LAGARDE

M. Didier REVEL Hôpital Cardiologique de Lyon Bâtiment Central 28 Avenue Doyen Lépine 69500 BRON Tél : 04.72.68 49 09 Fax :04 72 35 49 16 [email protected]

INFOMATHS

INFORMATIQUE ET MATHEMATIQUES http://infomaths.univ-lyon1.fr M. Alain MILLE

M. Alain MILLE Université Claude Bernard Lyon 1 LIRIS - INFOMATHS Bâtiment Nautibus 43 bd du 11 novembre 1918 69622 VILLEURBANNE Cedex Tél : 04.72. 44 82 94 Fax 04 72 43 13 10 [email protected] - [email protected]

Matériaux

MATERIAUX DE LYON M. Jean Marc PELLETIER Secrétariat : C. BERNAVON 83.85

M. Jean Marc PELLETIER INSA de Lyon MATEIS Bâtiment Blaise Pascal 7 avenue Jean Capelle 69621 VILLEURBANNE Cédex Tél : 04.72.43 83 18 Fax 04 72 43 85 28 [email protected]

MEGA

MECANIQUE, ENERGETIQUE, GENIE CIVIL, ACOUSTIQUE M. Jean Louis GUYADER Secrétariat : M. LABOUNE PM : 71.70 –Fax : 87.12

M. Jean Louis GUYADER INSA de Lyon Laboratoire de Vibrations et Acoustique Bâtiment Antoine de Saint Exupéry 25 bis avenue Jean Capelle 69621 VILLEURBANNE Cedex Tél :04.72.18.71.70 Fax : 04 72 43 72 37 [email protected]

ScSo

ScSo* M. OBADIA Lionel Insa : J.Y. TOUSSAINT

M. OBADIA Lionel Université Lyon 2 86 rue Pasteur 69365 LYON Cedex 07 Tél : 04.78.77.23.88 Fax : 04.37.28.04.48 [email protected]

*ScSo : Histoire, Geographie, Aménagement, Urbanisme, Archéologie, Science politique, Sociologie, Anthropologie

mailto:[email protected]

http://biomserv.univ-lyon1.fr/E2M2







3

Résumé

Ce travail de thèse a eu comme objectif principal de prouver la faisabilité technique de

détection par spectrométrie de fluorescence X de traceurs ajoutés dans une matrice polypropylène afin de développer un tri efficace et rentable.

Dans un premier temps nous avons choisi les traceurs convenant le mieux à notre problématique, en s’appuyant sur des critères tels que la toxicité et la radioactivité, l’intensité du signal de détection, la singularité du pic, la disponibilité des réserves et le prix. Suite à l’application de ces contraintes nous avons choisi les oxydes de terres rares comme traceurs.

Deuxièmement, après avoir sélectionné les traceurs, nous avons étudié leur dispersion dans la matrice PP ainsi que leur impact sur les propriétés du matériau tracé. Les résultats expérimentaux présentés ont montré que l’addition de 1000 ppm d’oxyde de terre rare, de taille micrométrique, dans une matrice polypropylène a un impact non significatif sur les propriétés mécaniques et physico-chimiques, ainsi que sur la photo-dégradation sous rayonnement UV. De plus, la dispersion est homogène sans formation d’agglomérats.

Pour finir nous avons validé leur choix par rapport à la détection par SFX. Des résultats expérimentaux et de modélisation ont montré que nous pouvons détecter les oxydes de terre rare dans une gamme de concentration [100-1000 ppm] pour un temps d’acquisition de 10 ms.

Mots-clés : traceur, dispersion de charge, propriétés de polyoléfines, vieillissement, détection, identification, spectrométrie de fluorescence X, tri, recyclage

Abstract

Rare earth oxides can be used as tracers for the identification of polymer materials in

order to have an economically efficient recycling and high speed automatic sorting of plastic wastes. This study focused on the detection of these particles by X-ray fluorescence spectrometry and their effect on PP matrix with respect to thermal and mechanical properties and to photo-degradation under UV irradiation exposure.

Addition of 1000 ppm of such particles, of micrometric size, has a minor effect on the mechanical and thermal properties of the traced materials as well as in the photo-degradation of the polymer after UV irradiation exposure. The SEM images together with the results obtained from image processing show a homogenous dispersion of tracers into PP matrix.

Regarding their detection by X-ray fluorescence spectrometry, experimental and modeling results have shown that the rare earth oxides studied could be detected in a range [100-1000 ppm] for 10 ms acquisition time.

Keywords: tracer, dispersion of fillers, properties of polyolefin, UV photo-degradation, detection, identification, X-ray fluorescence, sorting, recycling

4

Η Ιθάκη σ’ έδωσε τ’ ωραίο ταξείδι. Χωρίς αυτήν δεν θάβγαινες στον δρόμο. Άλλα δεν έχει να σε δώσει πια. Κι αν πτωχική την βρεις, η Ιθάκη δεν σε γέλασε. Έτσι σοφός που έγινες, με τόση πείρα, ήδη θα το κατάλαβες η Ιθάκες τι σημαίνουν.

Ithaque t’a donné le beau voyage. Sans elle, tu ne te serais pas mis en route. Elle n’a plus rien d’autre à te donner. Même si tu la trouves pauvre, Ithaque ne t’a pas trompé. Sage comme tu l’es devenu, avec tant d’expérience, tu dois avoir déjà compris ce que signifient les Ithaque.

Extrait d’Ithaque de Constantin Cavafy, 1911.

A mes parents

5

Remerciements

Ces travaux de thèse ont été réalisés au sein de l’UMR CNRS 5223 / Ingénierie des Matériaux Polymères, au Laboratoire des Matériaux Macromoléculaires de l’INSA de Lyon et au Laboratoire Conception Produit Innovation de l’Institut ARTS et METIERS ParisTech Chambéry.

Je tiens tout d’abord à remercier Pr. Jean-François Gérard, directeur du LMM et de

l’UMR, pour m’avoir accueillie dans son laboratoire. Je souhaite ensuite remercier mes directeurs de thèse le Dr. Valérie Massardier-Nageotte

et le Pr. Daniel Froelich de m’avoir proposé ce sujet de thèse si passionnant et pluridisciplinaire. Je tiens à remercier Valérie pour son encadrement tout au long de cette thèse, sa disponibilité, ses conseils pertinents, sa bienveillance et les décisions qu’elle a prises pour son bon déroulement. Je remercie Daniel pour toutes les discussions constructives qu’on a eues, que ce soit au sujet de la thèse où du monde en général et de m’avoir fait bénéficier de son expérience sur l’éco-conception et l’analyse de cycle de vie.

Je tiens également à exprimer ma reconnaissance au Dr. Samuel Legoupil, chef du

Laboratoire Imagerie, Tomographie et Traitement du CEA de Saclay, pour m’avoir accueilli dans son laboratoire durant 6 mois. J’exprime aussi ma gratitude au Dr. Dominique Chambellan pour toute son aide précieuse sur le montage du banc d’essai et pour la transmission de son expérience sur la spectrométrie de fluorescence X.

Je souhaite remercier également Bénédicte Couffignal, directrice scientifique de

ReCoRD et Catherine Marioge, de l’ADEME pour avoir cofinancé et facilité sa réalisation. Je remercie aussi tous les membres de ReCoRD, François Thery (EDF), Juliette Beaulieu et Sébastien Lepetit (Renault), Marie-Lise Sablayrolles (Suez Environnement), Stéphanie Navarro et Richard Biquillon (Veolia Environnement), Laurent Cimolino (Socotec), pour leurs remarques pertinentes leurs des réunions

Mes remerciements vont également au et Pr. … pour avoir accepté d’être les rapporteurs

de cette thèse et pour les remarques constructives qui en ont découlé. D’autre part, je remercie Pr. …, respectivement président du jury et examinateurs, pour leur participation et leur contribution pertinente à mon jury de thèse.

Je souhaite aussi remercier les personnes qui ont contribué à la qualité de ces travaux

grâce à leur aide précieuse. - Elisabeth Maris, pour ses conseils sur le tri et le recyclage des matériaux plastiques

tout au long de cette thèse, - Jean Balcaen, pour le temps qu’il a passé avec moi devant un ordinateur pour analyser

par Matlab les images MEB afin de caractériser la dispersion des traceurs. Merci Jean. - Pierre Alcouffe, pour son contribution à l’obtention des images MEB,

6

- Raphael Brunel, pour ses conseils sur la caractérisation mécanique, - Gilbert Martignago, pour sa disponibilité et l’assistance technique sur l’extrudeuse et

la presse à injecter. Je souhaite aussi remercier l’ensemble des permanents, doctorants et post-doctorants du

LMM, du LCPI et du LITT qui ont contribué à rendre agréable ces trois années. En particulier, je souhaite remercier les secrétaires du LMM, Elena, Malou et Isabelle ainsi que Sabine du LCPI, pour leur aide précieuse, leur disponibilité et leur gentillesse. Durant ces trois ans j’ai eu énormément besoin d’elles, merci mesdames.

Un grand merci également à tous mes ‘’co-bureaux ‘’, que ce soit au plateau du LCPI, Marion, Chrystel, Sophie, Charlotte, Carol et Yann, ou du côté du LMM, Benoît, Adil, Ghislain, Ali A., Marie et Amélie. Vous m’avez permis de travailler dans une bonne ambiance ces trois années et de répondre à mes nombreuses questions.

Je remercie aussi tous les thésards et post-doc du LMM pour tous les bons moments

passés ensemble : - les filles : Sandra, Antonella, Céline, Emilie D., Emilie G., Géraldine, Anne-Carine,

Caroline, Elise et Alexia. - les garçons : Nizar, Rodolphe, Fred, Pascal, Nico, Arnaud, Yves-Marie, Seb, Ludo,

Ali, Pierre, Sinbin, Nicolas J., Arthur, Grégoire, Stéphane, Maxime, Morgan et Yoann. Je tiens tout particulièrement à remercier Elisabeth. Tu étais là pour moi quand j’avais

besoin et tu m’as rendu ces deux dernières années de thèse si agréables à vivre. I’m so lucky to have met you.

Enfin, je souhaite remercier ma famille. Tout d’abord mon oncle qui m’a permis de

venir en France, mes grands-parents pour leur amour, mon frère pour son soutien et tous ses conseils en informatique, et un grand merci à mes parents dont les sacrifices m’ont permis de faire ce « beau voyage » et d’arriver où je suis aujourd’hui. Faleminderit mami dhe babi…

7

Table des matières Remerciements ....................................................................................... 5

Introduction Générale ........................................................................... 13

Introduction Chapitre I ......................................................................... 19

Chapitre I : Etude bibliographique et méthodologie de sélection des traceurs……………………………………………………………………………………..16

Addition of X-ray fluorescent tracers into polymers, new technology for automatic sorting of plastics: Proposal for selecting some relevant tracers ................................................................................................... 20

I.1. Introduction .................................................................................. 20

I.2. Sorting and separation technologies for plastics ........................... 22

I.3. Tracing: a new concept for plastic identification and automatic sorting ................................................................................................... 23

I.4. The selection of tracers ................................................................. 25

I.4.1.The detection system in XRF ................................................... 25

I.4.2.Selection of elements composing the tracers ............................. 26

I.4.2.1.High X-ray fluorescent yield .............................................. 26

I.4.2.2.Toxicity and radioactivity .................................................. 27

I.4.2.3.Availability and material behaviour ..................................... 27

I.4.2.4.Singularity of tracer signal ................................................ 28

I.4.2.5.The potential elements for XRF tracer application ................. 28

I.4.3.Selection of tracers – Rare earth oxides ................................... 29

I.4.3.1.The rare earth elements and their applications .................... 29

I.4.3.2.Availability of rare earth elements ...................................... 30

I.4.3.3.Toxicity and stability of rare earth oxides ............................ 31

I.4.3.4.Danger of rare earth elements during incineration or disposal 33

I.4.3.5.Intensity and distinguishability of tracer signal .................... 34

I.5. Conclusion .................................................................................... 35

References ............................................................................................ 36

Conclusion Chapitre I ............................................................................ 39

8

Chapitre II : Etude de la dispersion des traceurs et des propriétés du matériau PP après ajout de traceurs……………………………………………....40

Introduction Chapitre II ........................................................................ 44

Part A: Elaboration and characterization of traced polypropylene with rare earth oxides for automatic identification and sorting of end-of-life plastics .................................................................................................. 45

II.A.1. Introduction ............................................................................ 45

II.A.2. Tracers identification ............................................................... 46

II.A.3. Experimental Part .................................................................... 47

II.A.3.1.Materials ..................................................................... 47

II.A.3.2.Dispersion of tracers .................................................... 48

II.A.3.3.Characterisation of tracers’ dispersion ............................ 49

II.A.3.4.Thermal properties ....................................................... 49

II.A.3.5.Mechanical characterisation ........................................... 49

II.A.3.6.X-ray fluorescence device ............................................. 50

II.A.4. Results and Discussions .......................................................... 51

II.A.4.1.Detection of tracers and concentration effect ................... 51


II.A.4.3.Characterisation of the dispersion of tracers .................... 54


II.A.5. Conclusion and further research .............................................. 56

Part B: A study on the dispersion, elaboration, characterization and photo-degradation of traced polypropylene with rare earth oxides ....... 59

II.B.1. Introduction ............................................................................ 59

II.B.2. Experimental ........................................................................... 60

II.B.2.1.Materials and preparation of traced polypropylene composites …………………………………………………………………………………………………………….60

II.B.2.2.Microscopic observation ................................................ 61

II.B.2.3.UV irradiation procedure ............................................... 61

II.B.2.4.Infrared spectroscopy ................................................... 62

II.B.2.5.Differential scanning calorimetry (DSC) characterization ... 62

II.B.2.6.Thermal stability analysis .............................................. 62

II.B.2.7.Mechanical characterisation ........................................... 62

9

II.B.3. Results and Discussions .......................................................... 63

II.B.3.1.Characterization of the dispersion of tracers .................... 63

II.B.3.2.Stability under UV irradiation/condensation cycles ........... 67

II.B.3.3.Effect of tracers on thermal and crystallization behaviours 68

II.B.3.4.Effect of tracers on thermal stability ............................... 70

II.B.3.5.Effect of tracers on mechanical properties ....................... 72

II.B.4. Conclusion ............................................................................... 73

Conclusion Chapitre II ........................................................................... 76

Chapitre III : Etude de la détectabilité à haut vitesse par SFX des traceurs sélectionnés, dispersés en faible en concentration dans la matrice PP………………………………….……………………………………………....40

Introduction Chapitre III ...................................................................... 81

Part A: Addition of tracers into the polypropylene in view of automatic sorting of plastic wastes using X-ray fluorescence spectrometry .......... 82

III.A.1. Introduction ............................................................................ 82

III.A.2. The identification of tracers .................................................... 83

III.A.3. Experimental part .................................................................... 84

III.A.3.1.Materials ................................................................... 84

III.A.3.2.Dispersion of tracers ................................................... 85

III.A.3.3.X-ray fluorescence device ............................................ 85

III.A.4. Results and discussions ........................................................... 87

III.A.4.1.Detection of tracers and concentration effect ................. 87

III.A.4.2.The overlapping and mass absorption effect ................... 88

III.A.4.3.The effect of the acquisition time .................................. 89

III.A.5. Conclusion and further research .............................................. 90

References ............................................................................................ 92

Part B: Comparison of X-ray detectors for the optimisation of high speed detection of rare earth oxides, used as tracers into a polymer matrix ... 94

III.B.1. Introduction ............................................................................ 94

III.B.2. Theory ..................................................................................... 95

III.B.3. Methods and materials ............................................................ 96

III.B.3.1.Characteristics of detectors and tracers ......................... 96

III.B.3.2.Energy dispersive X-ray fluorescence device .................. 97

10

III.B.3.3.Software overview ...................................................... 98

III.B.3.4.Calculation of detection limit ........................................ 99

III.B.4. Results and Discussions ........................................................ 101

III.B.4.1.Comparison of the detectors with respect to experimental results…………………………………………………………………………………………………………………..101

III.B.4.2.Modeling results of the commercial detectors more adapted for the high speed detection of rare earth oxides .........................................103

III.B.4.2.1.Detectors specifications ........................................103

III.B.4.2.2.The detector and filter thickness effect ...................104

III.B.4.2.3.The X-ray generator voltage effect .........................107

III.B.4.2.4.The overlapping and mass absorption effect ............108

III.B.4.3.Comparison between experimental and modeling results 109

III.B.5. Conclusion ............................................................................. 110

References .......................................................................................... 112

Conclusion Chapitre III ....................................................................... 114

Conclusion Générale ............................................................................ 115

Perspectives ........................................................................................ 118

Annexes……………………………………………………………………………………120

11

Liste des abréviations ADEME Agence de l’Environnement et de la Maitrise de l’Energie RECORD Réseau de Recherche Coopératif sur les Déchets et l’Environnement PP Polypropylène/Polypropylene ABS Acrylonitrile-butadiene-styrene PET Polyethylène terephthalate PVC Polychlorure de vinyle PEhd PEhd Polyéthylène de haute densité PEbd PEbd Polyéthylène de basse densité PBDE Polybromodiphényléther VHU Véhicules hors d’Usage DEEE Déchets d’Equipement Electrique et Electronique RBA Résidus de Broyage Automobile SFX Spectrométrie de fluorescence X UV Ultra Violet MEB Microscopie Electronique à Balayage ED-XRF Spectrométrie à dispersion en énergie WD-XRF Spectrométrie à dispersion en longueurs d’onde ELV End-of-life vehicles WEEE Waste from Electronic and Electrical Equipment XRF X-ray fluorescence spectrometry SDD Silicon Drift Detector HPGe High Purity Germanium CdTe Cadmium Tellurium SEM Scanning Electron Microscope USGS United States Geological Survey

12

INTRODUCTION GENERALE

13

Introduction Générale Ce travail de thèse a été cofinancé par l’Agence de l’Environnement et de la Maitrise de

l’Energie (ADEME) et le Réseau de Recherche Coopératif sur les Déchets et l’Environnement (ReCoRD). Il a donné lieu à un programme ECOTECH de 4 ans financé par l’Agence Nationale de la Recherche, sous le nom de TRIPTIC (Traceurs Répartis pour l’Identification des Polymères et le Tri Industriel en Cadence).

Les performances considérables des plastiques, telles que leur faible poids, leur

durabilité, leur faible coût, leurs nombreuses propriétés (utilisation dans un vaste domaine de température, résistance à des nombreuses substances chimique et facilité de mise en œuvre) sont à l’origine d’une croissance de production annuelle de 10%, depuis 1950, et d’une demande mondiale en 2004 de 225 million de tonnes. Néanmoins, la gestion de leur fin de vie s’avère délicate et est devenue une priorité dans le monde politique, économique et environnemental pour tous les pays industrialisés.

En 2004, en Europe, 37% des plastiques étaient consommés dans l’emballage, 15 % pour des applications dans les secteurs de l’automobile et des équipements électriques et électroniques et le reste dans l’agriculture, le bâtiment, les loisirs, etc. Par contre, en ce qui concerne la génération de déchets, cette même année, 60% des déchets plastiques provenaient de l’emballage tandis que 10% étaient issus des véhicules hors d’usage (VHU) et des déchets d’équipements électriques et électroniques (DEEE).

En ce qui concerne le recyclage et la récupération de ces matières plastiques, actuellement, c’est dans le secteur de l’emballage que le recyclage est le plus développé (environ 25% sont recyclés) et cela grâce au tri automatisé par spectrométrie infrarouge. Cependant, dans le secteur automobile et des équipements électriques et électroniques, des contraintes techniques (couleur, granulométrie) empêchent d’utiliser efficacement ce type de techniques. Actuellement ces déchets sont essentiellement traités par des filières de tri post broyage et permettent de récupérer uniquement 6 à 8% des plastiques avec des niveaux de séparation conduisant à des grades dont les performances ne répondent pas à des applications à très haute valeur ajoutée.

D’un point de vue politique, afin de gérer la fin de vie de produits de consommation, la Commission Européenne a commencé à mettre en place deux directives, la 2000/53/CE et 2002/96/CE, traitant les matériaux de VHU et DEEE, respectivement. L’objectif pour les VHU est qu’à partir du 1er Janvier 2015, 95% des matériaux doivent être valorisés, dont 85% en recyclage, tandis que la valorisation pour dix catégories de DEEE doit attendre des taux compris entre 70 et 80%.

En effet, pour tenir les objectifs européens de recyclage des VHU, la filière va être obligée de recycler les plastiques. Aujourd’hui, environ 75% des matériaux sont récupérés (essentiellement métaux ferreux et non ferreux) tandis que le reste, ce qu’on appelle les résidus de broyage automobile (RBA) sont en général envoyés en centre d’enfouissement technique ou incinérés. Les RBA contiennent environ 50% de plastiques, et seule une fraction de polyoléfines non chargés de 6 à 8% est recyclée. A ce jour les plastiques ont une part croissante dans le poids du véhicule en vue de son allègement. Le problème de la fin de vie

14

des matières plastiques de VHU sera donc croissant si on n’anticipe pas le recyclage de ces matières.

En ce qui concerne les DEEE, afin de tenir les objectifs du recyclage fixés au niveau européen, le problème est non seulement de trier les plastiques entre eux mais aussi, au sein d’une même famille, d’extraire ceux qui sont ignifugés. En effet, il peut y avoir des problèmes de toxicité, s’il y a présence d’agents bromés et de perturbation du le processus de recyclage car les produits se dégradent si la température d’extrusion est trop élevée.

Outre la réglementation qui impose des objectifs de recyclage et d’utilisation de matières recyclés, le développement du recyclage contribue à la préservation des matières premières et à la réduction des émissions polluantes. La raréfaction des énergies fossiles, le développement de l’économie locale, les bénéfices environnementaux associés au recyclage jouent en sa faveur pour des applications à haute valeur ajoutée. Il est alors nécessaire de produire des matières secondaires de qualité pour proposer de nouvelles solutions à un marché dont la demande de matières plastiques recyclées est nettement plus forte que l’offre actuelle.

Pour les raisons citées ci-dessus, les acteurs du recyclage tentent de valoriser une matière la plus pure possible après des opérations de démontage, de broyage et de tri. Or, les technologies actuelles n’ont pas les capacités pour trier les matériaux polymères sombres, ni en fonction de leurs formulations, ni de leurs propriétés rhéologiques. Ceci, alors que plus de 60% des pièces plastiques utilisées dans l’automobile sont de couleur sombre voire noire et que ce chiffre est estimé à 40% dans le cas des déchets électriques et électroniques.

La nécessité d'améliorer les méthodes de tri afin de rendre le recyclage des matériaux plastiques viable devient urgente. Pour avoir un recyclage rentable et efficace, les matériaux plastiques doivent être triés automatiquement en fonction de la nature des formulations et à un niveau de pureté élevé.

Les technologies de tri existantes, utilisées pour les déchets plastiques, ne fournissent pas la flexibilité exigée pour un tri ultrarapide des plastiques par type et grade. La technique de traitement d’image haute résolution à partir de la transmission de rayons X, est limitée actuellement à la séparation du PVC et du PET tandis que les techniques de tri en temps réel basées sur la spectroscopie proche infrarouge ne peuvent pas identifier les différentes grades ou formulations du même type de plastique et trier les plastiques noirs présents dans de très nombreuses pièces pour l’industrie automobile.

Bien que des travaux de recherche aient été réalisés, les technologies industrielles n’utilisent pas un système traceur pour faciliter l’identification et le tri. Dans les années 90, sur un projet financé par la Communauté Economique Européenne, le Dr Ahmad (Université de Cranfield) a développé un nouveau concept d'identification de plastiques par introduction de traceurs fluorescents dans les polymères ce qui leur donne une signature en fluorescence UV. L’utilisation de traceurs doit permettre l’obtention de matériaux triés de haute pureté, une séparation des polymères par formulations, une identification positive et un tri rapide. Ces travaux ont reporté que la présence de colorants peut réduire le rendement de fluorescence et que le signal en présence de colorant noir était trop faible pour obtenir une identification du traceur. En plus de ces limitations, la spectroscopie UV est une technique de caractérisation surfacique ce qui impose une surface propre pour la détection des traceurs.

Après deux études confiées conjointement à l’IMP et au LCPI par ReCoRD sur l’ajout de traceurs dans le polypropylène, il a été conclu que la spectrométrie de fluorescence X

15

(SFX) peut offrir la possibilité de trier les matériaux polymères sombres. La SFX est une technique d’analyse élémentaire et non-destructive. Par rapport à la spectroscopie UV, le processus de détection n’est pas affecté par la présence des colorants noirs et une surface propre n’est pas nécessaire puisque elle permet une analyse en volume d’une épaisseur de 1 mm. Cependant, comme la SFX est une technique spectroscopique permettant une analyse élémentaire, cela signifie que le nombre de traceurs est limité aux éléments du Tableau Périodique de Mendeleïev.

L’étude sera principalement focalisée sur la valorisation de polymères fortement utilisés dans les industries automobiles, électriques et électroniques et c’est pour cette raison que nous avons décidé de travailler avec le polypropylène (PP), qui représente environ 30% des quantités totales des plastiques utilisés dans ces deux secteurs.

Les problématiques traitées dans le cadre de ce travail de thèse sont donc : - l’étude de la dispersion des traceurs et des propriétés du matériau après ajout de traceurs en faible concentration dans la matrice PP - la détectabilité par SFX et à grand vitesse de traceurs dispersés en faible concentration dans la matrice PP.

Le travail de thèse est présenté en trois chapitres. Dans le 1er, dans le cas de l’étude bibliographique, nous présentons les techniques de tri appliqués aujourd’hui et leurs limites pour la séparation des plastiques contenus dans les VHU et DEEE. Nous expliquons aussi pourquoi la SFX avec l’utilisation d’un système traceur peut être une technique très prometteuse et nous développons le protocole de sélection des traceurs basée sur des critères tels que le prix, la toxicité, l’écotoxicité, les réserves et la stabilité.

Une fois la sélection des traceurs effectuée, dans le Chapitre 2 nous présentons nos travaux de recherche sur la dispersion des traceurs dans la matrice PP et la caractérisation de cette matrice afin de savoir si le traceur en tant que charge pouvait avoir une influence sur les propriétés mécaniques, physico-chimiques ainsi que la photo-dégradation sous rayonnement UV. Suite à cette caractérisation nous pouvons ainsi estimer la gamme de concentration pour laquelle les propriétés de la matrice polymère ne sont pas affectées.

Pour terminer, le 3e Chapitre est consacré à l’étude de la détectabilité des traceurs par SFX. Nous présentons des tests en statique ainsi que des essais de modélisation afin d’optimiser la limite de détection, c’est-à-dire la quantité minimale de traceur qu’il faut ajouter pour avoir un matériau trié de haute pureté.

16

CHAPITRE I

Etude bibliographique

et méthodologie de sélection

des traceurs

« Rien ne se perd, rien ne se crée, tout se transforme »

Antoine-Laurent de Lavoisier

17

Addition of X-ray fluorescent tracers into polymers, new technology for automatic sorting of plastics: Proposal for selecting

some relevant tracers

F. Bezati, D. Froelich, V. Massardier, E. Maris

Soumis à “Resource, Recycling and Conservation”

18

Sommaire

Introduction Chapitre I ......................................................................... 19

I.1. Introduction .................................................................................. 20

I.2. Sorting and separation technologies for plastics ........................... 22

I.3. Tracing: a new concept for plastic identification and automatic sorting ................................................................................................... 23

I.4. The selection of tracers ................................................................. 25

I.4.1.The detection system in XRF ................................................... 25

I.4.2.Selection of elements composing the tracers ............................. 26

I.4.2.1.High X-ray fluorescent yield .............................................. 26

I.4.2.2.Toxicity and radioactivity .................................................. 27

I.4.2.3.Availability and material behaviour ..................................... 27

I.4.2.4.Singularity of tracer signal ................................................ 28

I.4.2.5.The potential elements for XRF tracer application ................. 28

I.4.3.Selection of tracers – Rare earth oxides ................................... 29

I.4.3.1.The rare earth elements and their applications .................... 29

I.4.3.2.Availability of rare earth elements ...................................... 30

I.4.3.3.Toxicity and stability of rare earth oxides ............................ 31

I.4.3.4.Danger of rare earth elements during incineration or disposal ……………………………………………………………………………………………………………….33

I.4.3.5.Intensity and distinguishability of tracer signal .................... 34

I.5. Conclusion .................................................................................... 35

References ............................................................................................ 36

Conclusion Chapitre I ............................................................................ 39

19

Introduction Chapitre I

Du fait de leurs performances techniques et économiques, les plastiques couvrent aujourd’hui une gamme très diversifiée d’utilisations pour lesquelles ils sont souvent devenus indispensables. Ceci est aussi le cas dans l’industrie automobile et électrique et électronique où leur part en Europe représente 15% de la totalité des plastiques produits. Les déchets provenant de ces deux secteurs sont souvent mis dans les centres d’enfouissement technique ou sont valorisés énergétiquement par des options faciles telles que l’incinération, qui ne semble ni économiquement viable, ni respectueuse pour l'environnement.

Afin de valoriser les déchets de ces deux secteurs il faut avoir un recyclage rentable et efficace donnant des matériaux de pureté suffisante pour leur valorisation. Toutefois, les technologies actuelles n’ont pas les capacités pour trier les matériaux plastiques sombres, la majorité des polymères utilisés dans ces industries.

Ce Chapitre présente dans un premier temps la situation européenne au niveau des déchets plastiques provenant des industries automobiles et électriques et électroniques en se basant sur 5 grands nations : la France, l’Allemagne, Angleterre, l’Espagne et l’Italie. Une fois le problème des déchets développé, nous expliquons pourquoi les méthodes de tri actuelles ne peuvent pas le résoudre en donnant leurs limites.

Les limites des techniques de tri actuelles ont conduit à l’émergence d’une nouvelle façon d’identifier des polymères, qui se base sur l’utilisation d’un système de traceurs fluorescents pour donner une signature en spectrométrie de fluorescence UV. Son point faible est aussi la difficulté d’identifier les traceurs en présence de colorants noirs.

Suite à des études, financées par ReCoRD et confiées conjointement à l’IMP et au LCPI, il a été conclu que la SFX peut être une solution pour l’identification de traceurs contenus dans une matrice de polymère noire. Nous développons ici les avantages de la SFX vis-à-vis de la fluorescence UV et nous présentons la procédure de la sélection des traceurs qui sont utilisés dans ces travaux.

CHAPITRE I : Etude bibliographique et méthodologie de sélection des traceurs

20

Addition of X-ray fluorescent tracers into polymers, new technology for automatic sorting of plastics:

Proposal for selecting some relevant tracers

I.1. Introduction

The substantial benefits of plastics in terms of low weight, durability, and low cost together with their properties to be used at a wide range of temperatures, be chemical and light resistant as well as to easily worked as a hot melt helps explain that since 1950, their production has increased an average of almost 10% every year on a global basis and that the annual worldwide demand has grown to 225 million tonnes in 2004 (Andrady and Neal, 2009; APME, 2006).

In the European context, according to PlasticsEurope (APME, 2006) the thermoplastics and thermosets demand by converters in Europe (EU25, Norway and Switzerland) was some 47 million tonnes in 2004. The major countries of plastic production are Germany, Italy, France, UK (United Kingdom) and Spain, which together account for around 70% of all European conversion.

Table 1 presents a breakdown of plastic demand and post-consumer waste in Germany, Italy, France, UK and Spain during the year 2004 (APME, 2005). Of the plastics consumed, 37% was used for packaging, 15% for durable consumer applications such as electronics goods and vehicles and the remainder for building/construction applications, domestic/household use, agriculture, etc. Post-consumer plastic waste generation across these five countries was 14.6 million tonnes in 2004, for a demand of 33.1 million tonnes. Table 1 confirms that the main source of plastic wastes has its origin in the field of packaging. However, it is clear that post-consumer waste from electronic and electrical equipment (WEEE) and end-of-life vehicles (ELV) are becoming significant sources by representing approximately 10% of plastic wastes.

Table 1 Plastic demand and post-consumer waste by sector in France, Spain, Italy, UK and Germany in 2004

Demand Waste ktonnes % Ktonnes %

Packaging 12250 37.0 8970 61.3 Electrical and electronics 2480 7.5 600 4.1 Automotive 2480 7.5 820 5.6 Other 15890 48.0 4240 29.0 Total 33100 14630

Fig. 1 summarizes the recovery and disposal of post-consumer plastic waste by sector in

France, Spain, Italy, UK and Germany in 2004 (APME, 2005). More than 50% of post-consumer plastic waste is sent to landfill for disposal. The automotive and electrical industries


21

are the worst performers with less than 10% of plastic wastes recycled from these sectors. Moreover, the energetic recovery of plastic wastes from these sectors is achieved through easy options such as incineration, which may cause the emission of harmful gases together with generation of toxic fly and bottom ash that contain lead and cadmium [Dodbiba and Fujita, 2004; Patel et al, 1998; Curlee and Das; 1991].

For giving a real policy of raw resource management based on the recycling of end-of-life products, the European Commission has set up two Directives, 2000/53/CE and 2002/96/CE, dealing with the materials of ELV and WEEE, respectively. The objective for the ELV is that from January 1st 2015, the reuse and recovery rate shall be increased to a minimum of 95% of average weight per vehicle and year, whereas the recovery quotas for 10 WEEE categories are fixed to a range of 70% to 80% for the end of 2006.

The recycling of polymer materials coming from ELV and WEEE is difficult because of the large number of polymer types, grades and blends available, besides, the presence of additives can produce significant changes to the mechanical and thermal properties of the polymers. Consequently, the separation and automatic sorting of polymers as part of a recycling scheme is of major importance. Moreover, the main plastics in car and electrical parts are blacks and, for today’s optical sorting technologies, it is still impossible to discriminate these kinds of materials.

Fig. 1. Recovery and disposal of post-consumer plastic waste by sector in France, Spain, Italy, UK and Germany

in 2004 (APME, 2005) Table 2 lists the main types of resins used in automotive industry and electronic and

electrical equipment and their consumption in Europe in 2004. PP together with ABS represents the largest volume of commodity plastics and could be excellent secondary sources of materials.

0

10

20

30

40

50

60

Packaging Electical Automotive Other Total

Perc

enta

ge o

f pla

stic

was

te r

euse

d (%

)

Mechanical recycling Feedstock recycling Energy recovery


22

Table 2 Plastics consumption in automotive industry and electronic and electrical equipment by resin type in Europe in 2004 (APME, 1999; APME, 2001; APME, 2006; Maudet-Charbuillet, 2009)

Abbreviation Name Automotive Electrical ktonnes % Ktonnes %

PP Polypropylene 1516 43 635 18 ABS Acrylonitrile-butadiene-styrene 247 7 1163 33 PE Polyethylene 282 8 35 1 PS Polystyrene - - 670 19 PVC Polyvinyl chloride 106 3 141 4 PA Polyamide 423 12 106 3 PC Polycarbonate 106 3 141 4 Other - 846 24 635 18

In this paper, we will review the current technologies for plastics sorting and point out

their limits, in order to propose a new technology for automatic sorting of plastic wastes through the use of a tracer system, mainly orientated for the ELV and WEEE.

I.2. Sorting and separation technologies for plastics For an economically efficient recycling of polymer materials, waste plastics need to be sorted cheaply and automatically into individual types and grades due to the various characteristics that each of the different resin types hold. Bruno (Bruno) has separated the automated sorting of plastic wastes into two categories: macrosorting and microsorting. The macrosorting section deals with the sorting of whole bottles or containers whereas the microsorting section follows the sorting of plastics after it has been chopped into pieces. A comparison of the applications of macrosorting and microsorting and their limits are given in Table 3.

As it can be seen in Table 3, for the macrosorting systems, optical sorting is limited for colour separation of plastics only; the near Infra-red is unsuitable for dark objects whereas middle Infra-red can identify them but cannot provide a high-speed identification. The X-ray technology, transmission or fluorescence, is limited to the separation of PVC from PET and the laser induced breakdown spectroscopy is unsuitable for high speed automatic sorting.

Regarding the microsorting techniques, density separation and froth flotation are slow processes which require at least one separation step for each material, and do not provide polymers of high purity. The triboelectric/electrostatic separation could be used for plastics of significantly different dielectric constant, but this technique requires dry and clean plastic surfaces. Moreover, all the sorting techniques mentioned above could not identify different grades of the same polymer.


23

Table 3 Comparison of the applications of macrosorting and microsorting technologies and their limits

Sorting technology

Application Limits

Mac

roso

rtin

g

Optical (Pascoe, 2003) Sorting of polymers by colour, removing coloured impurities.

Does not identify the polymer, limited for only colour separation.

Near Infra-red (Alam et al, 1994; Scott et al, 1995)

Bottle sorting. Unsuitable for dark objects.

Mid Infra-red (Pascoe, 2003) Technology which can identify dark plastics.

Cannot be used for high-speed identification and requires relatively smooth, clean surface.

X-ray (Dinger, 1992; Kenny and Bruner, 1994)

Proven and established technology for the identification of PVC.

Cannot identify the polymer families since they are composed of the same elements. Only used for the separation of PVC from PET.

Laser Induced Breakdown Spectroscopy (Gondal et al, 2007; Cuesta, 2009)

Identification of PE, PP, ABS. Unsuitable for high speed automatic sorting, time of analysis: 1-5 s.

Mic

roso

rtin

g

Density separation (Bruno; Atland et al, 1995)

A low cost separation technique used for any material mixture of different densities.

Similar densities for some plastics (PE vs. PP, PVC vs. PET), which leads to a low purity of sorted fractions. Slow process, requires at least one separation step for each material.

Froth Flotation (Shent et al, 1999; Fraunholcz, 2004)

Used for any kinds of polymers. The material is treated with a surfactant for changing the wettability.

Selectivity may be difficult to achieve. Requirement of a surfactant to modify the plastics. Separation of only one component at a time.

Triboelectric/Electrostatic Separation (Hearn and Ballard, 2005; Iuga et al, 2005)

Used for plastics of significantly different dielectric constant.

Requires a dry and clean plastic surface.

I.3. Tracing: a new concept for plastic identification and automatic sorting

As shown previously, the existing technologies of sorting do not provide the versatility

or flexibility needed for separating the dark plastics into monopolymeric fractions which are the essential prerequisite materials for any efficient recycling process. In the end of 90’s, Simmons et al (Simmons, 1998) and Ahmad (Ahmad, 2000) proposed a new concept of identification of plastics by marking them with binary combination fluorescent tracers detectable by UV (ultraviolet) spectroscopy. The use of a tracer system could provide a high purity of the sorted materials, a separation by polymer grade as well as polymer type, a separation by additive system, a high speed positive identification and a high speed sorting. This project, founded by the European Economic Community, has focused on the sorting of rigid plastics packaging from household waste for demonstrating the concept. They concluded that the speed and purity of sorting were limited by the mechanical singulation inadequacy of the conveyor system at high speed and that the pigments found in plastics reduced the


24

fluorescence yield, whereas in the case of black pigments, the reduction was too drastic to permit identification. Aside from these limitations, UV spectroscopy is a surface detection method and this may imply a “clean” surface for tracer identification. The use of "tags" for plastic identification by UV/Vis spectroscopy has also been studied by Corbett et al (Corbett et al, 1194). They showed that the addition of phosphor luminescent "tags" to classes of polymers is viable. However, the detection can be disturbed by contaminants, which may be luminescent, and the stability of organic "tags" during the reprocessing of polymers.

The companies specialized in magnetic sorting, as Eriez (Mankosa and Luttrell, 2005), has also proposed a magnetic sorting process of polymers in which a magnetic substance is dispersed. The main advantage of the magnetic detection is the lack of sensitivity with respect to the additives contained in polymers. However, the magnetic tracer system provides only binary separation and requires amounts of tracers which are of the order of the percent and these quantities could engender homogenisation problems and affect the mechanical properties of the polymer.

Table 4 summarizes the advantages and drawbacks of some spectroscopic techniques which can be used for detecting a tracer system together with the magnetic detection as reported by Froelich et al (2007). Table 4 Comparison of the detection techniques which can be used for detecting a tracer system (Froelich et al, 2007)

Detection technique Advantages Drawbacks UV High speed identification. Tracer

concentration : 1-10 ppm High quantities of tracer for dark plastics. Fluorescence of the polymer matrix. Surface detection.

Near Infra-red High speed identification. Unsuitable for the identification of tracers in dark plastics.

Mid Infra-red Can identify the tracers in dark plastics. Unsuitable for high-speed identification.

XRF (X-ray fluorescence)

Can identify the tracers in dark plastics. High speed identification. Volume detection.

The number of tracers limited by the Mendeleyev’s Periodic Table.

Magnetic detection Lack of sensitivity with the additives contained in plastics. Proven and established technology.

Only binary discrimination. Requires an elevated quantity of tracers.

From the comparison between the detection techniques of Table 4, it comes out that the

XRF spectrometry is compromising for detecting a tracer system dispersed into dark polymer materials coming from ELV and WEEE. XRF spectrometry is a volume, non-destructive elemental analysis. Compared to Ultra Violet fluorescence the detection process is not affected by black pigments, and a “clean” surface is not required due to a volume detection of around 1mm depth . However, as XRF is a spectroscopic method enabling elemental analysis of material, the number of tracers is limited to the elements of Mendeleyev’s Periodic Table.


25

I.4. The selection of tracers

I.4.1. The detection system in XRF

For having an efficient sorting, the requirements for the detection system are that it should be able to reliably identify the tracers, singly or in any combination, at very high speeds and analyse the data generated. To achieve the above requirements, the identification system is based on four compounds, as shown in Fig. 2:

- Illumination source: an X-ray generator. - Filter system: a copper filter, for reducing the noise of measurements. - XRF detection: silicon, high purity germanium or cadmium tellurium detectors, proceeding to 200000 – 1000000 counts/second. - Data processing electronics: processed the output from the detectors for identifying the tracers. The excitation of the tracers is achieved through the use of an X-ray generator and the

detection by X-ray fluorescence spectrometry. Each tracer emits a unique radiation in XRF, which depends on the atomic number of the element. The detection system, coupled with a data processing system, detects the emitted radiation and identifies the signature of the tracer, and thus the nature of the polymer matrix. The tracer concentration must be in the range of [100-1000 ppm] in order not to affect the properties of the polymer matrix and for having an economically efficient sorting of plastics. To achieve an automatic sorting of plastics by grade and type, the tracers might be used in a matrix, such that each combination corresponds to a specific type of polymer. For example, in Fig. 2, by using only 3 tracers, it is possible to identify 7 (23-1) variations of different plastics.

Fig. 2. X-ray fluorescent tracer detection system

The detection of tracers will be achieved by EDXRF (energy dispersive X-ray

fluorescence) (Harvilla, 1997), a two-step process that begins with the removal of an inner shell electron of an atom, while the resulting vacancy is filled by an outer shell electron. The second step is the transition from the outer shell electron orbital to an inner shell electron

A

B

A

B

X-ray generator

Filter Plastic material with A and B tracer

A

B

C

XRF detection

Polymer 1Polymer 2

Polymer 3

Polymer 4

Polymer 5

Polymer 6Polymer 7

Identification Type of plastics

A - -- B -

- - C

A B -

A - C

- B CA B C


26

orbital, accompanied by the emission of an X-ray photon. The emitted fluorescent photon is characteristic of the element and is equal to the difference in energy between the two electron energy levels. Since the energy difference is always the same for given energy levels, the element can be identified by measuring the energy of the emitted photon. The emission process is similar to other fluorescent measurement techniques, but it is restricted to the X-ray region of the electromagnetic spectrum that ranges from 4 to over 80 keV. The photon energies detected are designed as K, L, or M X-rays, depending on the energy level being filled. For example, CeKα1 represents, for the cerium element, the transition corresponding to the passage of L to K level and 1, the relative intensity of the transition in the series (1, more intense than 2). There are as many possible X-ray lines as there are inner shell electrons. However, the most analytically useful and most intense X-ray lines are the K shell electrons; hence the identification of tracers by EDXRF will be carried on by detecting their K energy lines.

I.4.2. Selection of elements composing the tracers

The selection of tracers adapted for the XRF spectrometry detection system was achieved by a two steps process. As XRF is an elemental analysis, the first step was to chose the more adapted elements from the Mendeleyev’s Periodic Table and then for these elements to find the chemical formula which will obey to a number of key criteria such as to be compatible with the potential applications as well as the objective of identification and sorting.

The elements composing the tracers were selected by imposing several criteria to elements which will not be adapted for such an application. These basic criteria are summarized as follows:

- High X-ray fluorescent yield: for an intense signal and distinguished peaks - Toxicity and Radioactivity: elimination of toxic and radioactive elements - Availability of elements and material behaviour: elimination of elements having poor reserves and of elements which cannot be added as solids in the polymer matrix - Singularity of tracer signal: elimination of elements contained in polymer additives, such as “natural tracers”.

I.4.2.1. High X-ray fluorescent yield In order to operate at energies in which the detectors have high efficiency, the

identification of tracers will be carried on by detecting their K energy lines. Therefore, elements emitting at energies lower than 10 keV will be eliminate, since these energies are more easily absorbed by air, plastic dust coverings, thin metal foils, which may be present in the construction of an ‘‘industrial” sorting environment. Consequently, by applying these criteria, the elements having an atomic number lower than 30 (ZnKα1 = 9.67 keV) could not be chosen as tracers.


27

I.4.2.2. Toxicity and radioactivity

The elements which can be selected as tracers must have very low toxicity and radioactivity for meeting the regulations associated with their processing and their use. Toxic elements such as Gallium (Ga), Arsenic (As), Selenium (Se), Cadmium (Cd), Tin (Sn), Antimony (Sb), Mercury (Hg), Thallium (Tl), Lead (Pb) and Bismuth (Bi) will be rejected for a potential XRF tracer application (Lawyers, 2007).

Furthermore, radioactive elements such as Curium (Cm), Americium (Am), Plutonium (Pu), Uranium (U), Thorium (Th), Radium (Ra), Polonium (Po) and Actinium (Ac) or synthetic ones such as Americium (Am), Berkelium (Bk), Bohrium (Bh), Curium (Cm), Californium (Cf), Darmstadtium (Ds), Dubnium (Db), Einsteinium (Es), Fermium (Fm), Hassium (Hs), Lawrencium (Lr), Mendelevium (Md), Meitnerium (Mt), Neptunium (Np), Nobelium (No), Rutherfordium (Rf), Seaborgium (Sg) and Technetium (Tc) will also be eliminated.

I.4.2.3. Availability and material behaviour

Elements having poor reserves (Fig. 3), and which are expensive such as Ruthenium (Ru), Rhodium (Rh), Palladium (Pd), Silver (Ag), Indium (In), Tellurium (Te), Osmium (Os), Iridium (Ir), Platinum (Pt), Gold (Au), Protactinium (Pa), Rhenium (Re) and Astatine (At) will be rejected as potential tracers for XRF application.

With respect to the material behaviour the elements of the noble gases group such as Helium (He), Neon (Ne), Argon (Ar), Krypton (Kr), Xenon (Xe) and Radon (Rn) will be eliminated since their standard state is gas.

Fig. 3. Abundance in weight percentage of elements in Earth's upper continental crust as a function of atomic number. The rare earth elements are labelled in red bold whereas the 11 rarest "metals" are labelled in green

italic (Pannetier, 1985).

Po

U

Bi

Pb

TlHg

AuPt

Ir

Os

Re

W

Ta

Hf

Lu

YbEr

Tm

HoTb

SmGdDy

Eu

Pr

NdCe

LaCs

Xe

Te

ISb

Sn

In

CdAg

Pd

Rh

Ru

Mo

Nb

Zr

Y

Rb Sr

KrBr

Se

AsGe

Ga

ZnCu

Co Ni

Fe

Mn

CrV

Ti

Sc

CaK

A

Cl

SP

SiAl

NaMg

Ne

F

OC

N

BeB

Li

He

H Ba

1.E-11

1.E-10

1.E-09

1.E-08

1.E-07

1.E-06

1.E-05

1.E-04

1.E-03

1.E-02

1.E-01

1.E+00

0 10 20 30 40 50 60 70 80 90 100

Atomic Number, Z

Abu

ndan

ce o

f ele

men

ts

http://fr.wikipedia.org/wiki/Curium


28

I.4.2.4. Singularity of tracer signal In order to avoid confusion caused by the emissions from the elements composing the

additives incorporated in the polymer matrix, it is necessary to identify these elements, by considering them as “natural tracers”, for ensuring an adequate and reliable discrimination between tracers. Table 5 summarizes the main additives found in plastic materials.

The elements composing the additives will be eliminated as “natural tracers” and can probably be used for identifying the properties that they add in the polymer matrix. These elements are the following one: Fluor (F), Magnesium (Mg), Aluminium (Al), Phosphor (P), Chlorine (Cl), Calcium (Ca), Titanium (Ti), Chrome (Cr),Manganese (Mn), Iron (Fe), Cobalt (Co), Nickel (Ni), Zinc (Zn), Strontium (Sr), Brome (Br), Cadmium (Cd), Tin (Sn), Antimony (Sb), Barium (Ba), Mercury (Hg) and Lead (Pb). Table 5 The main additives used in plastics according to Murphy's ‘Additives for Plastics Handbook’ [Murphy, 1996]

Additives Specifications of additives Lubricants Ca, Zn, Li, Ba, Al, Pb stearates, organic compounds Plasticizers Chlorinated, phosphor and amino organic polymers Release agents Lubricants, alcohols, metallic stearates (Ca, Zn, Li, Ba, Al, Pb, Mg, Cd, Na) Fillers and reinforcements

Glass fibers, carbon fibers, talc, mica, wollastonite, barite, calcium carbonate, Pb, Mn, Ba, Ti, Fe, Co, Ni oxides

Colorants (75% inorganic compounds): TiO2 (white), Pb (coloured or white), iron oxides (yellow, red, black-brown), Cr (yellow), Mo (orange), Cd (red), Co (blue), Zn (white), aluminium (green), carbon black and organic pigments

Anti-foaming agents Insoluble oils, silicones, alcohols, stearates and glycols Stabilizers Salts of Ba, Ca, Sr, Cd, Zn, ZnO Antioxidants S-organic, P- organic, carbon black, organic compounds Flame retardants Chlorinated paraffin, Sb2O3, P/Br-copolymers

I.4.2.5. The potential elements for XRF tracer application

By applying key criteria for selecting the potential elements for XRF tracer application,

such as high X-ray fluorescent yield, toxicity and radioactivity, availability, material behaviour together with singularity, the list of potential elements can be summarized as follows:

Rare earth elements such as Yttrium (Y), Lanthanium (La), Cerium (Ce), Praseodymium (Pr), Neodymium (Nd), Samarium (Sm), Europium (Eu), Gadolinium (Gd), Terbium (Tb), Dysprosium (Dy), Holmium (Ho), Erbium (Er), Thulium (Tm), Ytterbium (Yb), and some solids such as Germanium (Ge), Zirconium (Zr), Niobium (Nb), Molybdenum (Mo), Iodine (I), Lutetium (Lu), Hafnium (Hf), Tantalum (Ta) and Tungsten (W).


29

I.4.3. Selection of tracers – Rare earth oxides During the selection process of the most adapted elements for the X-ray fluorescence

tracing of plastic material, it was reported that rare earth elements are potential candidates. For these elements, it is convenient to use as tracers their most stable chemical form, the oxide (Sastry, 1966). In order to confirm their selection as potential tracers, we have first overviewed their applications for identifying if some of them are used as additives in plastics. Second, we have collected data on their availability and their total production as well as their stability and toxicity. Third, we have considered their end-of life as tracers by studying what it will be the danger of such particles in the plastic wastes during the incineration or disposal. Finally, in addition to above criteria, the X-ray fluorescence signatures of the emissions from each tracer need to have a high yield and be spectrally distinguishable from each other when used in binary combinations and detected simultaneously.

I.4.3.1. The rare earth elements and their applications

In the contrary to what they name indicate, rare earth elements are not so rare elements.

Their content in the earth's upper crust is estimated to 0.02 % and they have more important reserves than the copper and the lead (Fig. 3). The three main minerals from which rare earth oxides are extracted are the monazite, bastnaesite and xenotime (Leveque, 2005; Gupta and Bose, 1989).

Since their discovery, rare earth elements have found many applications in various fields of materials science, thanks to their specific electronic structure, by offering high specificity and unit value. For example, cerium oxide plays an important role in the domain of multifunctional catalysts by reducing the emission level of pollutants such as nitrogen oxide and monocarboxylic oxide (Nunan et al, 1992; Diwell, 1991). Cerium oxide together with erbium oxide is also known to be one of the best polishing agents for glass (Niinistö, 1987). Thanks to their small, lightweight and high-strength, rare earth element magnets, such as alloys containing Nd, Sm, Gd, Dy, or Pr, have allowed miniaturization of numerous electrical and electronic components used in appliances, computers, automobiles, and military gear. Some of rare earth elements, such as Y, La, Ce, Eu, Gd and Tb, have found applications to the new energy-efficient fluorescent lamps and batteries containing La and Ce are gradually replacing Ni-Cd batteries in computer as well as communications applications and could probably replace lead-acid batteries in automobiles (USGC, 2005; Falconnet, 1993).

The use of rare earth elements in automotive pollution control catalysts, permanent magnets, and rechargeable batteries are expected to continue to increase as future demand for conventional and hybrid automobiles, computers, electronics and portable equipment grows, which means that markets are expected to require greater amounts of higher purity mixed and separated products for meeting the demand (Hedrick, J.B., 2007; Maestro and Huguenin, 1995).


30

I.4.3.2. Availability of rare earth elements

According to the 2010 edition of the U.S. Geological Survey (USGS) Mineral Commodity Summaries (USGS, 2010 a,b), the world's rare earth reserves are estimated to 100 M tonnes. As shown in Table 6, 36.4% of the world reserves are located in China, 13.1 % in United States, 5.5% in Australia and 3.1% in India; the remain reserves can be found in places such as Canada, Brazil, Malaysia, Africa, etc.

Because of its relative abundance of rare earth containing minerals, China has been very active in 2009 and has developed a major rare earth element industry, by eclipsing all other countries in production of both ore and refined products, by producing almost the totality (97 wt%) of rare earth minerals. It can also been seen in Table 6 that countries with important mine ores reserves such as United States and Australia, have not produced in 2009; for the reason that the prices proposed by the Chinese market are very low, due to much lower labour and regulatory costs in China (USGS, 2005). This fact shows also the importance of integrating other countries than China in the rare earth mine production, mostly due to the total dependence on a unique international supplier. Table 6 World reserves of rare earth ores (USGS, 2010 a,b)

Country Mine production in 2009 Reserves tonnes wt% ktonnes wt%

China 129000 97 36000 36.4 Commonwealth of Independent States Not available Not available 19000 19.2 United States - - 13000 13.1 Australia - - 5400 5.5 India 2755 2 3100 3.1 Brazil 665 0.5 48 0.05 Others 580 0.5 22500 22.7 World total (rounded) 133000 100000

As can be seen in Fig. 3 and Table 7, there are differences in the abundances of individual rare earth oxides. First, the even atomic numbers rare earth elements (58Ce, 60Nd, 62Sm, …) have greater abundances than the odd ones (57La, 59Pr, 63Eu, …). Second, the lighter rare earth elements are more incompatible and therefore more strongly concentrated in the continental crust than the heavier ones (USGC, 2005).

If it is considered that the objective is to trace 1000 ktonnes of plastics per year in Europe for WEEE and ELV, at a tracer concentration of about 100 ppm, this means that the quantity of each rare earth oxide used as tracer in binary combination, will be approximately 400 tonnes.

)/(400)_(3

)(/)(1,0)(4)(3)_(1000 yeartracerstonnestracersnombre

plasticskgtracergityrepeatabiltracersplasticsktonnes=

⊗⊗⊗


31

As can be seen in Fig. 3, rare earths are relatively abundant in the Earth’s crust, but discovered minable concentrations are less common than for most other ores. The world resources of rare earths are located primarily in American and Chinese bastnaesite and monazite. Table 7 shows the percentage of each rare earth oxides in these two ores. In order to estimate the reserves and the production of each rare earth oxide, it was considered that in the reserves and the mine production of 2009 given by U.S. Geological Survey the distribution of rare earth is similar to the one of the Earth’s crust. By comparing the data of Earth’s crust distribution with the rare earth content in American and Chinese bastnaesite and monazite, considering that the distribution of rare earths is similar to the Earth’s crust seems to be a representative first order approximation. For the yttrium oxide, the values were obtained according to the U.S. Geological Survey data.

The data of Table 7 show that the 400 tonnes of rare earth oxides does not seem to be a problem with respect to reserves and production for the light rare earth elements such as Y, La, Ce, Pr, Nd and Sm and the even atomic numbers rare earth elements, such as Gd, Dy, Er and Yb, even if for these last ones the reserves seem to be more limited compared to light rare earth elements. The availability as well as the production of rare earth elements are very important parameters which have to be taken in consideration for the selection of tracers. Table 7 Percentage of rare earth oxides in American and Chinese bastnaesite and monazite (Hedrick, J.B., 2007), their estimated production in 2009, their reserves and their prices in 2010 given by AMPERE for a purity of 99.9%

Tracers Bastnaesite Monazite Distribution in

Earth’s crust (%) Production in 2009 (tonnes)

Reserves (ktonnes)

Prices for 1 kg (€) USA China USA China

Y2O3 0.1 trace 3.2 3.2 - 8900 540 12 La2O3 33.2 23.0 17.5 17.5 15.3 18900 15300 14 CeO2 49.1 46.0 43.7 43.7 39.1 48500 39100 6 Pr2O3 4.3 6.2 17.5 17.5 4.7 5800 4700 31 Nd2O3 12.0 18.5 17.5 17.5 20.3 25200 20400 28 Sm2O3 0.8 0.8 4.9 4.9 5.5 6800 5500 50 Eu2O3 0.1 0.2 0.8 0.8 0.2 200 170 560 Gd2O3 0.2 0.7 6.6 6.6 5.1 6300 5100 85 Tb2O3 trace 0.1 0.3 0.3 0.8 1000 770 800 Dy2O3 trace 0.1 0.9 0.9 3.1 3800 3100 340 Ho2O3 trace trace 0.1 0.1 1.0 1200 1000 500 Er2O3 trace trace trace trace 1.7 2100 1700 30 Tm2O3 trace trace trace trace 0.2 200 170 300 Yb2O3 trace trace 0.2 0.2 2.5 3200 2500 250

I.4.3.3. Toxicity and stability of rare earth oxides

The tracers must have very low toxicity and need to meet the regulations associated

with their processing and their use. Table 8 presents the hazard classification of some rare earth oxides. With respect to Annexe I of European Directive 67/548/EC, the majority of rare


32

earth oxides are classified as irritant and their risk code is R36, which means that they may be irritating to eyes. Regarding the Median lethal dose (LD50), the results show that these substances have a very low toxicity. Moreover, a study done by Hirano and Suzuki (Hirano and Suzuki, 1996) on the exposure, metabolism, and toxicity of rare earths and related compounds showed that these elements are not highly toxic, and their cytotoxicity to macrophages is comparable to that of Cd or silica in vitro. The rare earth oxides, used as tracers are intending for plastics used in automotive and electronic industry, thus their property for being irritating to eyes will not be a problem for such applications.

The substances used as tracers must not chemically react with the host materials and cause changes to any of their thermal, mechanical or physical properties beyond acceptable limits. The characteristics of some rare earth oxides are given in Table 9. Rare earth oxides are inert particles and thus cannot chemically react with the host polymer material. For most of them, they crystallise to a cubic close packaging arrangement with some exceptions such as lanthanum and neodymium oxides which crystallise to a hexagonal one. They can be provided to a micrometric size in order not to act as probable nucleator agents by modifying the thermal properties of the polymer matrix (Xiaomin et al, 1996; Ye et al, 1996; Liu et al, 1993). Thermal stability must enable the tracers to be processed without degradation under normal extrusion or injection moulding conditions appropriate for the host plastics. As it can be seen in Table 9, the melting point of rare earth oxides is higher than 2000 °C, while the temperature of plastic processing is less than 250 °C, thus it can be concluded that these substances will stay stables.

Table 8 Hazard classification of some rare earth oxides

Rare earth oxides

CAS number

1Hazard Symbols Risk code

4LD50 (mg/kg)

Y2O3 1314-36-9 2Xi 3R36 > 5 000 La2O3 1312-81-8 Xi R36 > 5 000 CeO2 1306-38-3 Xi R36 > 5 000 Pr2O3 12037-29-5 Xi R36 > 5 000 Nd2O3 1313-97-9 Xi R36 > 5 000 Gd2O3 12064-62-9 Xi R36 > 5 000 Dy2O3 1308-87-8 Xi R36 > 5 000 Er2O3 12061-16-4 Xi R36 > 5 000 Yb2O3 1314-37-0 Xi R36 > 5 000 1Classification with respect to Annexe I of European Directive 67/548/EC 2Irritant, 3Irritating to eyes 4Median lethal dose of a toxic substance (LD50): dose required to kill half the members of a tested population.

Regarding the effect of tracers in the thermal and mechanical properties as well as in their stability under weathering; for products such as PP of automotive parts, which is likely to have long useful live under normal outdoor environment, the authors have already reported in previous work (Bezati et al, 2010) that the addition of rare earth oxides of micrometric particle size at 0.1 wt% content into PP matrix has a minor effect on the mechanical and


33

thermal properties of the traced materials as well as in the photo-degradation of the polymer after UV irradiation exposure. Consequently, the tracer concentration should be included within a range [0.01–0.1 wt%], in order to avoid their effect on the properties of the polymer matrix and being economically viable.

Table 9 Characteristics of some rare earth oxides

Tracers Mean particle size (µm)

Density (g/cm3)

Crystalline form of the oxide

Melting point (°C)

Y2 O3 1 - 10 2989 Cubic 2415 La2 O3 1 - 10 5872 Hexagonal 2250 CeO2 1 - 10 6110 Cubic 2600 Nd2 O3 1 - 10 6453 Hexagonal 2272 Gd2 O3 1 - 10 7600 Cubic 2340 Dy2 O3 1 - 10 8161 Cubic 2340 Er2 O3 1 - 10 8660 Cubic 2355 Yb2 O3 1 - 10 9200 Cubic 2346

I.4.3.4. Danger of rare earth elements during incineration or disposal

Even if the best scenario for the traced plastics is to consider that hundred percent will

be recovered, although, one possibility could be that part of the post-consumer traced plastic wastes are sent to landfill for disposal or to incineration for energy recovery. Therefore it is important to analyse the life cycle of the traced plastics and obtain more knowledge in the thermal decomposition of rare earth oxides and their potential risk to soils.

According to Hussein, (Hussein, 1996) the oxide compound of rare earth elements is the thermal decomposition product of rare earth metals and carbonates as well as the most stable molecule.

Diatloff et al as well as Wahid et al (Diatloff et al, 1995; Wahid et al, 2000) have reported the positive effects of rare earth elements on agricultural production. In particularly, La, Ce, Pr and Nd can promote the root growth of coconut, corn and mungbean at a low rate of application. On the contrary, high application of rare earth elements may lead to their scattering and bioaccumulation in the environment and cause environmental pollution (Haley, 1965; Sabbioni et al 1982).

Zhang et al. (Zhang et al, 2001) have studied the rare earth element content in various waste ashes and the potential risk to Japanese soils. The results showed that Y, La, Ce, Pr, Nd, Dy, Yb, Ho, Er, Tm and Lu in the waste ash samples were normally distributed, whereas Sc, Sm, Eu, Gd and Tb were not. Ce, with an average of 26 mg/kg, was the most abundant of the whole rare earth elements, followed by La, with an average of 14 mg/kg, whereas the least abundant was Tm with an average of 0.2 mg/kg. They also found that the content in waste ashes followed the sequence of Ce>La=Y>Nd>Sm>Pr>Gd>Dy>Eu>Tb>Er>Yb>Ho>Lu>Tm.

Based on the studies described above, the selection of rare earth oxide as tracers will probably not affect the life cycle of the plastic material during the incineration and disposal.


34

I.4.3.5. Intensity and distinguishability of tracer signal

In order to have visible and well distinguishable peaks from the background sample, the

intensity of tracers has to be maximalist and their fluorescent transmissions should lie in the energy range of [15–60 keV], where the detectors performed the best. The fluorescent peak should preferably be narrow, for reducing the background noise, and, furthermore, they should differ from other selected tracers by at least 0.5 keV, (resolution of commercial detectors) for ensuring adequate and reliable discrimination between tracers.

Table 9 gives the energies and relative intensities of the Kα2, Kα1, Kβ1, and Kβ2 lines expected for the elements composing the tracers, the detectors and the anode of X-ray generator. Intensities are normalized to 100% for the Kα1 energetic line.

Table 10 Kα1, Kβ1, Kα2 and Kβ2 energetic lines and their intensities for the elements composing the tracers, the anode of X-ray generator and the detectors

Lines Kα1 Kα2 Kβ1 Kβ2

Energy (keV)

Intensity (%)

Energy (keV)

Intensity (%)

Energy (keV)

Intensity (%)

Energy (keV)

Intensity (%)

Tracers

Y 14.96 100 14.88 50 16.7 22 17 3.1 La 33.44 100 33.03 52 37.8 28 38.7 6 Ce 34.72 100 34.28 52 39.2 29 40.2 6 Pr 36.03 100 35.55 52 40.7 29 41.8 6 Nd 37.36 100 36.85 52 42.2 29 43.3 6 Gd 43.00 100 42.31 53 48.7 30 50.0 7 Dy 46.00 100 45.21 53 52.1 31 53.5 7 Er 49.13 100 48.22 53 55.6 32 57.2 7 Yb 52.39 100 51.35 54 59.3 32 61.0 7

Anode W 59.32 100 57.98 54 67.2 33 69.1 8

Detectors

Si 1.74 100 1.74 100 1.84 100 - - Ge 9.89 100 9.86 50 11.0 19 11.1 0.6 Cd 23.17 100 22.98 51 26.1 26 26.6 5 Te 27.47 100 27.2 51 31.0 27 31.7 6

The most intense energy lines are the Kα1 and Kα2 followed by Kβ1 which is

approximately 3 times less intense than Kα1. The energetic resolution supplied by the commercial detectors do not allow the individual identification of the couples "Kα1 and Kα2",

hence for identifying the tracers, the entire counting rate between the two energetic lines Kα1 and Kα2 should be measured.

The idea of codification is based on the simultaneous presence of several tracers in the polymer matrix. For this reason, the mass absorption, the enhancement effects as well as the overlapping interferences caused by the other elements composing the tracers must be minimal. Table 9 shows that Kβ1 energy lines of Nd, Gd and Er are close to the Kα1 energy lines of Gd, Er and Yb respectively, thus by the simultaneous presence of these tracers, the


35

raw counting rate of Gd, Er and Yb will probably be overvalued by the additional counts of Kβ1 energy lines of Nd, Gd and Er respectively.

The probable overlapping interferences of some tracers together with the small extent of absorption and enhancement effects caused by the simultaneous presence of several tracers, limit their use to a binary matrix combination of 3 to 4 tracers. I.5. Conclusion

In this paper we have shown that the existing sorting techniques cannot discriminate the

polymers in different families and grades and cannot identify the black plastics, which are the main polymer materials used in automotive and electronic industry. Based on this research work, it appears that X-ray fluorescence, used with a tracer system, could be a solution for the high speed identification and automatic sorting of black plastics into different families and grades.

The addition of X-ray fluorescent tracers to plastics provides positive and specific versatility, high purity of sorted fractions as well as efficient and effective sorting. The major advantage of XRF compared to other optical techniques, such as UV spectroscopy, is its ability to detect the tracers even if the polymer matrix is black. However, for XRF the number of tracers is limited to the elements of Mendeleyev’s Periodic Table.

This study has concluded that the tracers the more adapted for the XRF detection process are some rare earth oxides. The selected micrometric particles are non toxic, non radioactive, with important reserves, and in concentration levels below 1000 ppm do not affect the properties of the polymer matrix.

The results of this work together with recent publications of the authors (Bezati et al, 2010 a, b) has lead to a four-years project, mainly funded by the French Agency of Research, with the objective of demonstrating the technical feasibility and commercial viability of sorting of plastics containing X-ray fluorescent tracer.


36

References Ahmad, S.R., 2000. Marking of products with fluorescent tracers in binary combinations for automatic identification and sorting. Assem. Autom. 20, 58–64. Alam, M.K., Stanton, S.L., Hebner, G.A., 1994. Near-infrared spectroscopy and neural networks for resin identification. Spectroscopy 9, 30–40. Altland, B.L., Cox, D., Enick, R.M., Beckman, E.J., 1995. Optimization of the High-Pressure, Near-Critical Liquid-Based Microsortation of Recyclable Post-Consumer Plastics. Resour. Conserv. Recycl. 15, 203-217. Andrady, A.L., Neal, M.A., 2009. Applications and societal benefits of plastics. Phil. Trans. R. Soc. B 364, 1977-1984. APME, 2006. An analysis of plastics production, demand and recovery in Europe. Association of Plastic Manufactures in Europe, Brussels, Belgium. http://www.apme.org. APME, 2005. Plastic waste in European key countries. Association of Plastic Manufactures in Europe, Brussels, Belgium. http://www.apme.org. APME, 2001. Plastic: a material of innovation for the electrical and electronic industry. Association of Plastic Manufactures in Europe, Brussels, Belgium. http://www.apme.org. APME, 1999. Plastic: a material of choice for the automotive industry. Association of Plastic Manufactures in Europe, Brussels, Belgium. http://www.apme.org. Bezati, F., Froelich, D., Massardier, V., Maris, E., 2010a. Addition of tracers into the polypropylene in view of automatic sorting of plastic wastes using X-ray fluorescence spectrometry. Waste Manag. 30, 591–596. Bezati, F., Massardier, V., Froelich, D., Maris, E., Balcaen, J., 2010b. Elaboration and characterization of traced polypropylene with rare earth oxides for automatic identification and sorting of end-of-life plastics. Waste Biomass Valor. Published online June 2010. Bruno, E.A., Automated Sorting of Plastics for recycling. http://www.p2pays.org. Corbett, E.C., Frey, J.G., Grose, R.I., Hendra, P.J., Taylorbrown, T., 1994. An investigation into the applicability of luminescent tagging to polymer recovery. Plast. Rub. Compos. Pro. 21, 5-11. Cuesta, J.M.L, Mascaro, J.F., Lorecki, B., 2009. Nouvelles technologies de tri et identification des matières plastiques. Journées MIEC, La Seyne sur Mer, France. Curlee, T.R., Das, S., 1991. Plastics Wastes (Management Control Recycling and Disposal). Noyes Data Corporation, New Jersey. Diatloff, E., Smith, F.W., Asher, C.J., 1995. Rare earth elements and plant growth: II. Responses of corn and mungbean to low concentrations of lanthanum in dilute, continuously flowing nutrient solutions. J. Plant. Nutr. 18, 1977-1989. Dinger, P., 1992. Automatic sorting for mixed plastics. BioCycle 33, 80-82. Kenny, G.R., Bruner, R.S., 1994. Experience and advances in automated separation of plastics for recycling. J. Vinyl Addit. Technol. 16, 181–186. Diwell, A.F., Rajaram, R.R., Shaw, H.A., Truex, T.J., 1991. The role of ceria in three-way catalysts. Stud. Surf. Sci. Catal. 71, 139-152. Dodbiba, G., Fujita, T., 2004. Progress in separating plastic materials for recycling. Phys. Sep. Sci. Eng. 13, 165–182.

http://www.apme.org/




http://www.p2pays.org/


37

Falconnet, P., 1993. The rare earth industry: a world of rapid change. J. Alloys. Compd. 192, 114-117. Fraunholcz , N., 2004. Separation of waste plastics by froth flotation-a review. Part I. Miner. Eng. 17, 261–268 Froelich, D., Maris, E., Massardier, V., 2007. Etat de l’art technico-economique sur les procédés et techniques d’incorporation de traceurs dans des matériaux polymères, en vue du tri automatisé des déchets plastiques des produits hors d’usage. Report N° 05-0907/1A, RECORD, French Industry-University Cooperative Research Network on Waste. www.record-net.org. Gondal, M.A., Siddiqui, M.N., 2007. Identification of different kinds of plastics using laser-induced breakdown spectroscopy for waste management. J. Environ. Sci. Heal. A 42, 1989-1997. Gupta, C.K., Bose, D.K., 1989. Process technology - rare and refractory metals. Bull..Mater..Sci. 12, 381-405. Haley, T.J., 1965. Pharmacology and toxicology of the rare earth elements. J. Pharm. Sci. 54, 663-670. Sabbioni, E., Pietra, R., Gaglione, P., 1982. Long-term occupational risk of rare-earth pneumoconiosis. A case report as investigated by neutron activation analysis. Sci. Total. Environ. 26, 19-32. Havrilla, G.J., 1997. Handbook of Instrumental Techniques for Analytical Chemistry. F. A. Settle, Prentice Hall, Upper Saddle River 459-479. Hearn, G.L., Ballard, J.R., 2005. The use of electrostatic techniques for the identification and sorting of waste packaging materials. Resour. Conserv. Recycl. 44, 91–98. Hedrick, J.B., 2007. Minerals Yearbook, Rare Earths. United State Geological Survey. http://minerals.usgs.gov/ Hirano, S., Suzuki, K.T., 1996. Exposure, Metabolism, and Toxicity of Rare Earths and Related Compounds. Environ Health Perspect. 104, 85–95. Hussein, G.A.M, 1996. Rare earth metal oxides: formation, characterization and catalytic activity. Themoanalytical and applied pyrolysis review. J. Anal. Appl. Pyrol. 37, 111-149. Iuga, A., Calin, L., Neamtu, V., Mihalcioiu, A., Dascalescu, L., 2005. Tribocharging of plastics granulates in a fluidized bed device. J. of Electrostatics 63, 937-942. Lauwerys, R.R., Haufroid, V., Huet, P., Lison, D., 2007. Toxicologie industrielle et intoxications professionnelles. Masson, Issy-les-Moulineaux. Leveque, A., Maestro, P., 2005. Terres Rares. Techniques de l'Ingénieur. Liu, J., Tang, G., Qu, G., Zhou, H., Guo, Q., 1993. Crystallization of rare earth oxide-filled polypropylene. J. Appl. Polym. Sci. 47, 2111–2116. Mankosa, M.J., Luttrell, G.H., 2005. Plastic material having enhanced magnetic susceptibility, method of making and method of separating. N° WO/2004/012920. Maudet-Charbuillet C., 2009. Proposition d’outils et démarches pour l’intégration de filières de recyclage de matières plastiques dans la supply chain automobile. Institut ParisTech Chambéry, France. Murphy, J., 1996. The additives for plastics handbook. Elsevier Advanced technology, Oxford.

http://www.record-net.orgl/

http://minerals.usgs.gov/


38

Niinistö L., 1987. Industrial applications of the rare earths, an overview. Inorg. Chim. Acta. 140, 339-343. Maestro, P., Huguenin, D., 1995. Industrial applications of rare earths: which way for the end of the century. J. Alloys. Compd. 225, 520-528. Nunan, J.G., Robota, H.J., Cohn, M.J., Bradley, S.A., 1992. Physicochemical properties of Ce-containing three-way catalysts and the effect of Ce on catalyst activity. J.Catal. 133, 309-324. Pannetier, R., 1985. Momento du vade mecum nucleaire. SCF du Bastet, Paris. Patel, M.K., Jochem, E., Radgen, P., Worrell, E., 1998. Plastics streams in Germany-an analysis of production, consumption and waste generation. Resour. Conserv. Recycl. 24, 191–215. Pascoe, R.D., 2003. Sorting of plastics using physical separation techniques. Recycling and reuse of waste materials. Proceedings of the international symposium, 173–188. Sastry, R.L.N., Yoganarasimhan, S.R., Mehrotra, P.N., Rao, C.N.R., 1966. Preparation, characterization and thermal decomposition of praseodymium, terbium and neodymium carbonates. J. Inorg. Nucl. Chem. 28, 1165-1177. Scott, D. M., and Waterland R. L., 1995. Identification of Plastic Waste Using Spectroscopy and Neural Networks. Polym. Eng. Sci. 35, 1011-1015. Shent, H., Pugh, R.J., Forssberg E., 1999, A review of plastics waste recycling and the flotation of plastics. Resour. Conserv. Recycl. 25, 85–109. Simmons, B.A., Overton, B.W., Viriot, M., Ahmad, S.R., Squires, D.K., Lambert, C., 1998. Fluorescent tracers enable automatic identification and sorting of waste plastics. Br. Plast. Rubber 4-8. USGC, 2010a. Rare Earths. United State Geological Survey. http://minerals.usgs.gov/. USGC, 2010b. Yttrium. United State Geological Survey. http://minerals.usgs.gov/. USGC, 2005. Rare Earth Elements-Critical Resources for High Technology. United State Geological Survey. http://minerals.usgs.gov/. Wahid, P.A., Valiathan, M.S., Kamalam, N.V., Eapen, J.T., Vijayalakshmi, S., Prabhu, R.K., 2000. Effect of rare earth elements on growth and nutrition of coconut palm and root competition for these elements between the palm and Calotropis gigantea. J. Plant. Nutr. 23, 329-338. Xiaomin, Z., Jingshu, L., Zhihui, Y., Jinghua, Y., 1996. Rheological properties and crystallization behavior of yttrium oxide filled low ethylene content polypropylene copolymer. J. Appl. Polym. Sci. 62, 313–318. Ye, C., Liu, J., Mo, Z., Tang, G., Jing, X., 1996. Crystal structure of polypropylene filled with rare earth oxides. J. Appl. Polym. Sci. 60, 1877–1881. Zhang, F., Yamasaki, S., Kimura, K., 2001. Rare earth element content in various waste ashes and the potential risk to Japanese soils. Environ.Int. 27, 393-398.





39

Conclusion Chapitre I Ce Chapitre, inscrit dans le cadre de la recherché bibliographique, avait comme objectifs

de montrer les limites des techniques de tri actuelles et de proposer une nouvelle méthode pour l’identification des matériaux plastiques sombres, la majorité des polymères utilisés dans les industries automobiles et électriques et électroniques.

En se basant sur ce travail et des études faites par l’IMP et le LCPI il apparaît que la SFX, utilisée avec un système traceur, peut être une solution viable pour l’identification rapide et le tri automatique des plastiques noirs en les discriminant entre différents types et grades.

L’ajout de ce type de traceurs discriminants présente tout son intérêt pour permettre la reconnaissance en ligne des plastiques, tels que les sombres, dont les moyens technologiques actuels ne permettent pas de les identifier efficacement. Ainsi, l’utilisation de traceurs spécifiques pourra permettre d’optimiser la récupération des matériaux plastiques au sein du procédé de tri des déchets.

L’avantage principal de la SFX comparée à d’autres techniques spectroscopiques, comme la fluorescence UV, est sa capacité à détecter les traceurs même si la matrice polymère est noire. Cependant, pour ce type de méthode d’analyse, le nombre de traceurs est limité aux éléments du Tableau Périodique de Mendeleïev.

Suite à certain nombre de critères bien définis, tels que la toxicité, l’écotoxicité, la radioactivité, la stabilité, la facilité de mise en œuvre, les réserves disponibles ainsi que la rentabilité économique, nous avons conclu que les traceurs les plus adaptés pour la SFX sont quelques oxydes de terre rare.

Une fois les traceurs sélectionnés, les oxydes de terre rare, nous allons vérifier leur impact sur les propriétés de la matrice polymère, les résultats seront présentés dans le Chapitre 2, et leur détection via la modélisation et des résultats expérimentaux, feront l’objet du Chapitre 3.

40

CHAPITRE II

Etude de la dispersion des

traceurs et des propriétés du

matériau PP après ajout de

traceurs

« Ce qui me scandalise, ce n'est pas qu'il y ait des riches et des pauvres : c'est le gaspillage »

Mère Teresa

41

Elaboration and Characterization of Traced Polypropylene with Rare Earth Oxides for Automatic Identification and Sorting of

End-of-Life Plastics

F. Bezati, V. Massardier, D. Froelich, E. Maris, J. Balcaen

Waste and Biomass Valorisation 1 (2010) 357–365

A study on the dispersion, elaboration, characterization and photo-degradation of

traced polypropylene with rare earth oxides

F. Bezati, V. Massardier, J. Balcaen, D. Froelich

Soumis à “Polymer Degradation and Stability ”

42

Sommaire

Part A: Elaboration and characterization of traced polypropylene with rare earth oxides for automatic identification and sorting of end-of-life plastics ……………………………………………………………..45

II.A.1. Introduction ............................................................................ 45

II.A.2. Tracers identification ............................................................... 46

II.A.3. Experimental Part .................................................................... 47

II.A.3.1.Materials ..................................................................... 47

II.A.3.2.Dispersion of tracers .................................................... 48

II.A.3.3.Characterisation of tracers’ dispersion ............................ 49



II.A.3.6.X-ray fluorescence device ............................................. 50

II.A.4. Results and Discussions .......................................................... 51

II.A.4.1.Detection of tracers and concentration effect ................... 51


II.A.4.3.Characterisation of the dispersion of tracers .................... 54


II.A.5. Conclusion and further research .............................................. 56

References

Part B: A study on the dispersion, elaboration, characterization and photo-degradation of traced polypropylene with rare earth oxides ....... 59

II.B.1. Introduction ............................................................................ 59

II.B.2. Experimental ........................................................................... 60

II.B.2.1.Materials and preparation of traced polypropylene composites .............................................................................. 60

II.B.2.2.Microscopic observation ................................................ 61

II.B.2.3.UV irradiation procedure ............................................... 61

II.B.2.4.Infrared spectroscopy ................................................... 62

43

II.B.2.5.Differential scanning calorimetry (DSC) characterization ... 62

II.B.2.6.Thermal stability analysis .............................................. 62

II.B.2.7.Mechanical characterisation ........................................... 62

II.B.3. Results and Discussions .......................................................... 63

II.B.3.1.Characterization of the dispersion of tracers .................... 63

II.B.3.2.Stability under UV irradiation/condensation cycles ........... 67

II.B.3.3.Effect of tracers on thermal and crystallization behaviours 68

II.B.3.4.Effect of tracers on thermal stability ............................... 70

II.B.3.5.Effect of tracers on mechanical properties ....................... 72

II.B.4. Conclusion ............................................................................... 73

References

44

Introduction Chapitre II

Après avoir sélectionné les traceurs, la prochaine étape dans le cheminement de la thèse était leur ajout dans la matrice PP et caractérisation de leur effet dans celle-ci. Afin de mieux comprendre leur dispersion et leur impact sur les propriétés mécaniques, physico-chimiques ainsi que la photo-dégradation sous rayonnement UV nous avons décidé de travailler avec un grade de PP non chargé et contenant très peu de stabilisants.

Ce Chapitre sera composé de deux articles. Le premier a été publié chez « Waste and Biomass Valorisation » tandis que le deuxième a été soumis chez « Polymer Degradation and Stability ».

Dans le 1er article nous présentons dans un premier temps la détectabilité des traceurs dans une matrice PP et ensuite, nous nous intéressons à la dispersion des traceurs (en l’observant par MEB), et à leur influence sur les propriétés mécaniques et physico-chimiques.

Le 2e article traite uniquement de l’impact des oxydes de terres rares sur la matrice polypropylène. En sachant que l’homogénéité de la dispersion peut jouer un rôle très important sur la détection des traceurs ainsi que sur les propriétés mécaniques nous avons poussé notre analyse en traitant les images MEB par Matlab afin de donner une réponse quantitative. Dans un deuxième temps, nous avons étudié la photo-dégradation sous rayonnement UV des matériaux tracés avec les oxydes de terres rares pour avoir une meilleure compréhension de leurs effets sur le vieillissement et la stabilité thermique.

Suite aux conclusions de ces deux articles et à la caractérisation des matériaux tracés, nous pourrons estimer la gamme de concentration pour laquelle les propriétés de la matrice polymère ne sont pas affectées.

CHAPITRE II : Etude de la dispersion des traceurs et des propriétés du matériau PP après ajout de traceurs

45

Part A: Elaboration and characterization of traced polypropylene with rare earth oxides for automatic

identification and sorting of end-of-life plastics

II.A.1. Introduction

Plastics have an impact on the environment, society and economic dimensions. The consumption of plastics materials reached an overall of 53.5 million tons in 2004 according to the Association of Plastics Manufactures in Western Europe. Of the plastics consumed, the breakdown by sector indicates that 37% of the plastic were used for packaging, 20% for construction, and 7.5% for the electrical/electronics industry with a further 7.5% being used by the automotive industry (Pascoe, 2003; APME, 2007).

Conventional plastics manufactured from these petroleum polymers generally degrade very slowly in the environment and the recycling process is very difficult since they are incompatible with each other. Currently, the energetic recovery of plastic wastes is achieved through easy options as incineration, which may cause the emission of harmful gases together with generation of toxic fly and bottom ash that contain lead and cadmium (Dodbiba and Fujita, 2004; Patel et al, 1998; Curlee and Das, 1991). Moreover, energy recovery is a consumptive recycling process, turning the recycled material into energy rather than usable material, and thus does not conform to the reuse-ration required by the European legislations and directives (Zoboli, 1998).

The automotive and electrical industries are currently the worst performers due to the complexity of the waste materials that these sectors produce. In 2004, a study showed that in France, 82% of plastic wastes coming from ELV (End-of-life Vehicles) and WEEE (Waste of Electric/Electronic Equipments) were sent to landfill (Linder and Herold, 2005). To resolve this problem, the European Commission has recently introduced very strict regulations (European Parliament, 2000, 2002). For example, in the case of automotive sector, a ratio of 5 % of ELV residues landfill deposal is required in the framework of developing solutions in order to reduce the impact of end of life products.

Therefore, it is necessary to improve technologies such as the sorting of polymer materials so as to make their recycling profitable. For an economically efficient recycling of polymer materials, waste plastics need to be sorted cheaply and automatically into individual types and grades (Froelich, 2007)

Techniques based on optical spectroscopy, such as infrared reflection/absorption (Alam et al, 1994; Moore, 1999; Florestan et al, 1994) have reached their limits. They are unsuitable for dark objects such as those containing carbon black that absorb and scatter at NIR frequencies, and they can neither identify different grades of the same polymers nor be used if the surface of the plastic wastes is wet, painted or dirty.

The technique of high resolution imaging using X-rays is limited to the separation of PVC from PET (Kenny and Bruner, 1994) and the elimination of PVC and brominated aromatic compounds contained in the electronic wastes and substituted combustibles.


46

In 1992, Ahmad et al (Ahmad, 1992, 2004) developed a new concept of identification of plastics by incorporating fluorescent tracers into the polymers and giving them a fluorescent signature in Ultra Violet spectroscopy. This research concluded that the speed and purity of sorting were limited by the mechanical singulation inadequacy of the conveyor system at high speed and that the pigments found in plastics reduced the fluorescence yield, whereas in the case of black pigments, the reduction was too drastic to permit identification. Aside from these limitations, Ultra Violet spectroscopy is a surface detection method and this implies a “clean” surface for tracer identification.

The challenge of overcoming the inefficiency of existing sorting technologies associated with the need for identifying black plastic wastes coming from the ELV and WEEE, has led to the development of new methods of identification of polymeric materials.

The principal objectives of this work are to prove the technical feasibility of detection of tracers (rare earth oxides) added to polypropylene through the use of XRF (X-ray fluorescence) spectrometry and to study their effect as fillers, on the thermal and mechanical properties of traced materials. In previous work (Bezati et al, 2010), the authors have already studied some parameters which can influence the detection of tracers.

XRF spectrometry is a volume, non-destructive elemental analysis. Compared to Ultra Violet fluorescence the detection process is not affected by black pigments, and a “clean” surface is not required due to a volume detection of around 1mm depth . Nevertheless, as XRF is a spectroscopic method enabling elemental analysis of material, therefore the number of tracers is limited to the elements of Mendeleyev’s Periodic Table.

The choice of rare earth oxides as tracers was achieved by eliminating from the Mendeleyev’s Periodic Table, the toxic and radioactive elements, in addition to elements contained in polymer additives, low atomic number elements; which do not give an intense signal and elements which do not have important reserves.

The choice of polypropylene, one of the world’s major plastics, was justified by its large use in the fields of automotive and electric/electronic industries. Supposing that another polymer matrix was studied, the detection results would have been the same as that of the PP matrix, since polymers are generally composed of elements, such as hydrogen, carbon, nitrogen and sulphur, which have a low atomic number and hence will not hinder tracer identification. Regarding the thermal and mechanical properties, the tracer concentration should be included within a range [0.01-0.1 wt%], in order to avoid their effect on the properties of the polymer matrix.

II.A.2. Tracers identification

It turns out that the effectiveness of sorting and in particular the speed of identification of plastic wastes can be improved by the use of a tracer system giving a unique signature to each type and grade of polymer. In the detection system shown in Figure 1, the excitation of the tracers is carried out through the use of an X-ray source and their detection is achieved by X-ray fluorescence spectrometry. This work of identification technology was first proposed by SR Ahmad et al, 2000, and Simmons, 1998, who used fluorescent tracers emitting in Ultra Violet spectrometry.


47

The identification system is based on the dispersion of very small quantities of one or more substances -tracers- emitting X-ray fluorescence into the plastics requiring identification. The concentration of the tracers must be within a range of [0.01-0.1 wt%], in order not to influence the properties of the polymer matrix.

Fig.1. Principle of tracers' identification by binary code

Due to the great diversity of the nature of plastics, including the types of polymers, as well

as the grades, the additives and colorants, it is not practical to have a tracer for each potential variant. To overcome this constraint, tracers can be used in a matrix, such that each combination corresponds to a type, grade or additive. For example, theoretically, by using only 4 tracers, it is possible to identify 15 (24 - 1) variations of different plastics.

The identification is carried out by exciting the plastics using an X-ray source. This radiation is absorbed by the tracers, which are excited to higher atomic energy levels. Each tracer emits a unique radiation of X-ray fluorescence, which depends on the atomic number of the element. The detection system, coupled with a data processing system, detects the emitted radiation and identifies the signature of the tracer, and thus the nature of the polymer matrix.

II.A.3. Experimental Part

II.A.3.1. Materials

ISPLEN PP 050 G1E is a medium melt flow rate polypropylene homopolymer particularly formulated and adapted for injection moulding and extrusion applications. It is intended for applications that require good impact resistance balanced with high stiffness. The original pellets have a melt mass-flow rate of 5.8 g/ 10min (2.16kg at 230°C) and a density of 0.905 g/cm3.

The rare earth oxides used as tracers are described in Table 1. The rare earth oxides were chosen as tracers because they satisfy the required specifications (Hedrick, 1988; Maestro and Huguenin, 1995; Morteani, 1991). They are compatible, not abrasive and stable in the environment of evolution. They are also non toxic (Lauwerys, 2007; Peltier, 1986) during their application and use. Regarding the detection method, they have a high atomic number, facilitating their detection in X-ray fluorescence.

The prices given by AMPERE for a purity of 99.9% for one tonne of rare earth oxides of Y2O3, CeO2, Nd2O3, Gd2O3 and Er2O3, and for 10 kg of Dy2O3 and Yb2O3 are shown in Table

A

B

A

B

X-ray generator

Filter Plastic waste with A and B tracer

Electronic Network

A

B

C

Tracers detection

A - -

- B -

- - C

A B -

A - C

- B CA B C

Polymer 1

Polymer 2

Polymer 3

Polymer 4

Polymer 5

Polymer 6Polymer 7


A

B

A

B

A

B

A

B

X-ray generator

Filter Plastic waste with A and B tracer

Electronic Network

A

B

C

Tracers detection

A - -

- B -

- - C

A B -

A - C

- B CA B C

A - -

- B -

- - C

A B -

A - C

- B CA B C

Polymer 1

Polymer 2

Polymer 3

Polymer 4

Polymer 5

Polymer 6Polymer 7

Polymer 1

Polymer 2

Polymer 3

Polymer 4

Polymer 5

Polymer 6Polymer 7



48

1. As it can be seen in the table, for 1 g of tracer per 1 kg of plastic, the prices vary from 0.006€, in the case of cerium oxide to 0.34€, in the case of dysprosium oxide. Prices of rare earth oxides will also depend on the “Supply and Demand” economic model and may decrease in the case of Dy2O3 and Yb2O3, so as to equalize the quantity demanded by consumers, and the quantity supplied by producers, resulting in an economic equilibrium of price and quantity.

Table 1 Materials specifications

Materials Chemical Name Supplier Mean particle

diameter (µm) Density (g/cm3)

Toxicity (LD 50) (mg/kg)

Prices in € for 1 g

PP Polypropylene REPSOL 2000 0.905 - -

Y2O3 Yttrium Oxide RHODIA 2.250 5 > 5000 0.012

CeO2 Cerium Oxide RHODIA 2.250 7.3 > 5000 0.006

Nd2O3 Neodymium

Oxide AMPERE 3.500 7.24 > 1000 0.022

Gd2O3 Gadolinium

Oxide RHODIA 1.850 7.4 > 1000 0.075

Dy2O3 Dysprosium

Oxide RHODIA 2.250 7.8 > 1000 0.340

Er2O3 Erbium Oxide AMPERE 8.200 8.6 > 1000 0.022

Yb2O3 Ytterbium

Oxide AMPERE 1.000 9.2 > 1000 0.250

II.A.3.2. Dispersion of tracers

All traced polypropylene samples were prepared under identical mixing and moulding conditions. Several samples of polypropylene containing the rare earth oxides in different concentrations were prepared for analysis (Table 2), using a twin screw extruder CLEXTRAL BC21 machine with a screw of L/D = 90 and D = 25mm. The extrusion temperature and screw speed were 205°C and 120 rpm respectively and the residence time 15 min. The screw and temperature profiles used in this study are supplied in Figure 2.

Fig. 2. Screw and temperature profiles for the extrusion processing

The pellets obtained from extrusion were then injected in a Battenfeld Unilog B2/350 Plus

injection moulding machine. A general purpose screw was used in the barrel, which was kept at 220 °C. Probes of diameter 6 cm and thick 2 mm were injected, for the X-ray fluorescence analysis.


49

Table 2 Characteristics of traced PP samples Code PP/tracer_wt% Tracer Masse Composition of

each tracer (wt%) Volume Composition of each

tracer (vol%) PP_reference PP-reference 0 0 PP/Nd2O3_0.1 Nd2O3 0.1 0.140 PP/Nd2O3_1 Nd2O3 1 1.402 PP/Gd2O3_0.1 Gd2O3 0.1 0.119 PP/Gd2O3_1 Gd2O3 1 1.191 PP/all_0.025 Y2O3, CeO2, Nd2O3,

Gd2O3, Dy2O3, Er2O3, Yb2O3

0.025 0.029 PP/all_0.1 0.1 0.113 PP/all_0.145 0.145 0.163

II.A.3.3. Characterisation of tracers’ dispersion

The dispersion of tracers in the PP matrix was investigated by scanning electron microscopy (SEM) using a Hitachi S3500N model. The specimens were prepared by injection moulding with a press Battenfeld Unilog B2/350 Plus and fractured in liquid nitrogen and sputter-coated with gold before being examined with the microscope at an accelerating voltage of 30 kV.

II.A.3.4. Thermal properties

Thermal properties of blends were measured by differential scanning calorimetry using DSC Q10 of TA Instruments. The samples were cut from probes and placed in aluminium pans and then heated in an argon atmosphere from room temperature to 210°C under a controlled heating rate of 10°C/min. Cooling of the samples was performed from 210 to 0°C with a cooling rate of 10°C/min after holding the samples at 210°C for 5 min. The sample mass was typically between 5 to 10 mg. The degree of crystallisation, χC, was calculated considering a melting enthalpy of 209 J/g for a 100% crystalline polypropylene.

Thermal stability of specimens was characterized by using a thermogravimetric analyzer (Q500 Thermogravimetric Analyzer, TA Instruments). Measurements were conducted under an argon flow rate of 90mL/min at heating rate of 20°C/min. The scanning temperature ranged from 25 to 500°C. Similar to the DSC characterizations, small samples of 10 to 15 mg were cut from probes for measurement. The temperature of maximum decomposition rate (Tmax) was calculated for each traced PP sample.

II.A.3.5. Mechanical characterisation

Tensile test specimens were prepared according to ISO 527-1:1993. Tensile mechanical tests were made with an Instron machine model 4469 at ambient temperature. Young’s


50

modulus was measured at 1mm/min with a strain of between 0.05 and 0.3% of strain. Yield stress and elongation at break were measured at cross-head 250 mm/min speed and evaluated from stress-strain curves on an average of at least six specimens.

II.A.3.6. X-ray fluorescence device

The X-ray fluorescence device, developed by NRT, is a test system which allows taking static measurements of the samples spectrum. A schematic of a typical EDXRF (energy dispersive X-ray fluorescence) spectrometer is shown in Figure 3. The X-ray radiation from the X-ray source passes through a hole onto the sample and then travels to the detector.

Fig. 3. Schematic of the EDXRF device

The signal from the detector is then processed electronically and sent to the computer,

which also controls the X-ray source. The XRF spectra can then be analyzed and used for the separation of the samples containing the tracers.

This EDXRF technique (Harvilla, 1997) is a two-step process that begins with the removal of an inner shell electron of an atom. The resulting vacancy is filled by an outer shell electron. The second step is the transition from the outer shell electron orbital to an inner shell electron orbital. The transition is accompanied by the emission of X-ray photon. The fluorescent photon is characteristic of the element and is equal to the difference in energy between the two electron energy levels. Because the energy difference is always the same for given energy levels, the element can be identified by measuring the energy of the emitted photon. In turn, the intensity of the emitted photons determines the concentration of the element. Therefore, the measure of the photon energy provides the identification of the element and the intensity of the photon emission provides a measure of the amount of the element.

The emission process is similar to other fluorescent measurement techniques, but it is restricted to the X-ray region of the electromagnetic spectrum that ranges from 4 to over 80 keV. The photon energies detected are designed as K, L, or M X-rays, depending on the energy level being filled. For example, GdKβ1 represents, for the gadolinium element, the transition corresponding to the passage of M to K level and 1, the relative intensity of the transition in the series (1, more intense than 2). There are as many possible X-ray lines as there are inner shell electrons. However, the most analytically useful and most intense X-ray


51

lines are the K shell electrons. Although, the multitude of emitted X-ray lines could result in complex spectra, the relative low intensities of the lines below the L level allow obtaining a clear spectrum with minimal interference.

The volume of the analysed sample was 1mm3. The distance between the sample and the source was 10 cm and the angle 45°. The detection time was one minute and the detector could make 2000 counts per second. The X-ray source used was a tungsten tube.

II.A.4. Results and Discussions

II.A.4.1. Detection of tracers and concentration effect

The objective of these tests was to prove the detection of the 7 selected tracers in the PP

matrix. Table 3 shows the expected energy of the element composing the tracers. The detection of tracers is separated in two domains. The first one, between 7-20 keV (Lα1, Lβ1 of ytterbium and Kα1, Kβ1 of yttrium) and the second between 34-60keV (Kα1, Kβ1 of cerium to ytterbium). The tested samples contained all the tracers in three different concentrations (PP/all_0.025, PP/all_0.1 and PP/all_0.145). The exposure time for the four samples was 1minute. The obtained spectrum is shown in Figure 4.

Table 3 X-ray fluorescent energy values of tracers’ elements [19]

Energy lines (keV)

Y2O3 CeO2 Nd2O3 Gd2O3 Dy2O3 Er2O3 Yb2O3 Source Y Ce Nd Gd Dy Er Yb W

Kα1 14.96 34.71 37.36 42.99 45.99 49.12 52.38 59.31 Kβ1 16.74 39.26 42.27 48.72 52.18 55.69 59.35 67.23 Lα1 1.92 4.84 5.23 6.06 6.49 6.95 7.41 8.40 Lβ1 1.99 5.26 5.72 6.71 7.25 7.81 8.40 9.67

As shown in the spectrum, five of the tracers (Y2O3, CeO2, Nd2O3, Gd2O3 and Dy2O3), at

concentration levels of 0.1 and 0.145 wt%, are clearly visible and distinguishable from the “background” sample by their Kα1 energy line.

For the same five tracers at 0.025 wt%, the fluorescent peak is not sufficiently different from the background in order to make a positive identification.

Concerning the detection of Er2O3 and Yb2O3, their Kα1 energy lines (49.12 and 52.38 keV respectively) are emitted in the same domain of energy (49-53 keV) as the tungsten source (WKα1: 59.31 keV), which is rescheduled at about 8 to 9 keV.

Regarding the detection of the Lα1 and Lβ1 energy lines of CeO2, Nd2O3, Gd2O3, Dy2O3, Er2O3 and Yb2O3, a large fluorescent peak appears in the range [5-9 keV] corresponding to these energies. Consequently, it is difficult to distinguish the Lα1 and Lβ1 energy lines of these elements.


52

Fig. 4. Comparison of XRF spectra of PP/reference, PP/all_0.025, PP/all_0.1 and PP/all_0.145

Moreover, the detection of the Lα1 and Lβ1 energy lines present three additional

difficulties:

- They have significantly less fluorescent yield than the K energy lines - They are much closer in energy to other elements which might be present (example

cerium versus titanium) and thus require a combination of longer exposure times and more sensitive detectors.

- The lower energy of L X-rays are more easily absorbed by air, plastic dust coverings, thin metal foils, that are present in the construction of an "industrial" sorting environment.

II.A.4.2. Thermal properties

The minerals used as fillers in polypropylene are principally talc and calcium carbonate. However, little attention has been paid to rare earth compounds used as fillers in polymers. As rare earth minerals are abundant in China, studies investigating PP composites containing rare earth compounds were published by Chinese scientists.

Liu and al. (Liu et al, 1993; Ye et al, 1996) have studied the effect of a large number of rare earth oxides on isothermal crystallization and melting behaviour with differential scanning calorimetry. They found that a series of rare earth oxides in a fine powder form may act as a nucleator and influence the growth rate of the spherulite and the mechanical properties of polypropylene.

Other studies of Xiaomin et al, 1996 have shown that the addition of Y2O3 has some effect on the viscosity of the system. Crystallization characteristics have indicated that the filler acts as a nucleating agent, increasing crystallization degree of the investigated polymer, and changing β crystal form content.

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70

Energy (keV)

Inte

nsity

Kα Y

Kβ Y

Kα CeKα Nd

Kα GdKα Dy

Kα ErKα Yb

PP/all_0.025

PP/Reference

PP/all_0.145

PP/all_0.1

Lα, Lβ-tarcers


53

Along this work, the thermal properties of PP composites containing rare earth oxides

were studied. The values of melting and crystallization enthalpies and temperatures, the

maximum decomposition rate temperature and the crystallization degree of PP containing

Gd2O3, Nd2O3 and all the tracers are listed in Table 4. Figure 5 includes the DSC data and the

way in which enthalpies and temperatures where calculated.

Table 411

Thermal properties and crystallization degree of traced PP samples obtained by DSC and ATG

Sample ∆Hf (J/gPP ) Tf (°C) ΔHc (J/gPP) Tc (°C) χc (%) T max (°C)

PP_reference 92 ± 3 166.9 ± 0.5 96 ± 3 104.9 ± 0.5 44 ± 1 471.7 ± 0.5

PP/Nd2O3_0.1 86 ± 3 167.6 ± 0.5 90 ± 3 105.6 ± 0.5 41 ± 1 473.6 ± 0.5

PP/Nd2O3_1 92 ± 3 165.9 ± 0.5 97 ± 3 107.5 ± 0.5 44 ± 1 473.9 ± 0.5

PP/Gd2O3_0.1 90 ± 3 166.1 ± 0.5 95 ± 3 108.5 ± 0.5 43 ± 1 475.0 ± 0.5

PP/Gd2O3_1 90 ± 3 167.2 ± 0.5 97 ± 3 108.4 ± 0.5 43 ± 1 476.2 ± 0.5

PP/all_0.1 91 ± 3 166.2 ± 0.5 95 ± 3 106.5 ± 0.5 43 ± 1 475.2 ± 0.5

PP/all_0.145 89 ± 3 165.7 ± 0.5 94 ± 3 108.8 ± 0.5 43 ± 1 475.8 ± 0.5

Fig.5. Melting and crystallization peaks of traced PP samples after the 2

nd heating run

The addition of tracers within the concentration range of [0.1-1 wt%] to the PP matrix

does not seem to influence the melting and crystallization enthalpies and temperatures as well

as the crystallization degree. Even if an increase of the crystallization degree at 1 wt% was

expected, the size of neodymium and gadolinium oxide particles is not small enough so as to

act as a nucleator agent. Regarding the thermal stability, the addition of rare earth oxides

Hea

t F

low

(W

/g)

En

do

ther

mic

PP_reference

PP/Nd2O3_0.1

PP/Nd2O3_1

PP/Gd2O3_0.1

PP/Gd2O3_1

PP/all_0.1

PP/all_0.145

Temperature (C°)

Integration of ΔHfIntegration of ΔHc

Hea

t F

low

(W

/g)

En

do

ther

mic

PP_reference

PP/Nd2O3_0.1

PP/Nd2O3_1

PP/Gd2O3_0.1

PP/Gd2O3_1

PP/all_0.1

PP/all_0.145

Temperature (C°)

Integration of ΔHfIntegration of ΔHc


54

improves slightly the degradation level of the PP matrix as the Tmax increases with tracer content.

The fact that tracers at 0.1% do not influence the thermal properties of the PP matrix means that it is not necessary to take further precautions for plastic processing of polypropylene, since the melting and crystallization temperature and enthalpies remain unchanged.

II.A.4.3. Characterisation of the dispersion of tracers

An important parameter of our study is the dispersion of tracers in the PP matrix. The homogenous dispersion of tracers will allow a good detection by X-ray fluorescence spectrometry for a given surface. It is also noted that the better they are dispersed in the PP matrix, the lower will be the modifications on the mechanical properties.

The morphology of the compounds up to a concentration of 1% of gadolinium oxide and neodymium oxide is shown in Figure 6. The morphology consists of finely dispersed particles in the PP matrix. In both samples, for a scanning surface of 0.15 mm², no aggregates were observed. This is probably because they are broken up to primary particles during the extrusion process. The SEM images were analysed by MATLAB by applying image processing. The obtained particle size was approximately 1µm for both Gd2O3 and Nd2O3.

Fig.6. A – SEM images of fractured specimens of gadolinium oxide dispersed in the PP matrix at 1%

B - SEM images of fractured specimens of neodymium oxide dispersed in the PP matrix at 1%

Table 5 Image processing results obtained with MATLAB

Sample dth_cubic (µm)

dth_hexagonal (µm)

dcalculated (µm)

Mean particle size (µm)

PP/Nd2O3_1 19.8 14.2 17.3 1.06 PP/Gd2O3_1 17.7 12.7 14.5 0.86

A B


55

The dispersion of rare earth oxides particles was estimated by comparing the distance (dcalculated) of the first ten neighbouring spheres calculated for the SEM data and the theoretical distance for a cubic (dth_cubic) or a hexagonal (dth_hexagonal) close packing arrangement (Table V). In the case of neodymium oxide the arrangement is more near to the cubic close packing arrangement whereas the gadolinium oxide is arranged into a hexagonal close packing. Both SEM images and data obtained by image processing show a homogenous dispersion of gadolinium and neodymium oxide into the PP matrix.

II.A.4.4. Mechanical characterisation

The inorganic fillers can change the characteristics of a polymer in two ways: Primarily, the properties of particles themselves such as size, shape and modulus can have a profound effect, especially on mechanical properties. Secondarily, the particles may cause a change in the micromorphology of the polymer, which may then give rise to differences in the observed bulk properties. In order to study the influence of rare earth oxides on the mechanical properties of the PP matrix, the tensile properties of traced PP are shown in Table 6.

Table 6 Tensile mechanical properties of traced PP samples at room temperature

Sample Young’s Modulus E (MPa) Yield Stress σy (MPa) Elongation at break εb (%) PP_reference 665 ± 28.8 50.5 ± 0.5 99.7 ± 5.4 PP/Nd2O3_0.1 670 ± 17.4 48.8 ± 0.4 78.6 ± 6.6 PP/Nd2O3_1 685 ± 11.6 49.0 ± 0.4 62.8 ± 7.7 PP/Gd2O3_0.1 655 ± 22.5 49.1 ± 0.5 72.7 ± 6.1 PP/Gd2O3_1 689 ± 28.4 48.9 ± 0.4 67.7 ± 16.1 PP/all_0.1 686 ± 16.5 49.4 ± 0.2 83.6 ± 9.0 PP/all_0.145 697 ± 24.6 49.8 ± 0.1 64.1 ± 15.9

The strain-stress curves’ results show that the Young’s modulus of the system increases

slightly from 10 to 30 MPa, with particle content. This increase is not significantly important compared for example to a polypropylene containing 5% of calcium carbonate particles, where the difference with an unfilled PP is around 200MPa. As for the yield stress, the tensile mechanical tests show that it slightly decreases upon the addition of rare earth oxides particles (Thio et al, 2002; Zuiderduin et al, 2003).

The obtained results also reveal that the elongation at break drops upon increasing the filler fraction. The decrease in the elongation at break with rigid fillers arises from the fact that the actual elongation experienced by the polymer matrix is much greater than the measured elongation of the specimen. Although the specimen is part filler and part matrix, all the elongation comes from the matrix if the filler is rigid. In the case of rare earth oxide-filled PP, there is a good adhesion between the fillers and the matrix, and thus the fracture path tends to go from particle to particle, resulting in a decrease in deformation (Leong, 2004).

At 0.1 wt% tracer concentration level, the elongation at break slightly decreases from 10 to 15% with a standard deviation of 6 to 9%. In this case the observed drop value is closed to


56

the standard deviation of measurements and hence it can be considered that the influence of rare earth oxides is negligible.

II.A.5. Conclusion and further research This research work demonstrated the technical feasibility of the detection of tracers

dispersed in the PP matrix through the use of X-ray fluorescence spectrometry. The tracer system and detection technique used can provide specific and positive identification versatility, efficient sorting, high speed identification, high purity of the sorted fraction and solutions in dark plastics detection.

We proved that by using a test system device which allows taking static measurements of samples using energy dispersive X-ray fluorescence technology, it was possible to detect 5 of the 7 tracers tested at the concentration level of 0.1wt% for 1 minute acquisition time with a 2000 counts/s detector.

A major parameter in the development of an industrial device is the acquisition time which must be around 10 ms. In this work, for the detection of tracers, a 2000 counts/s detector was used. Currently, commercialized CdTe (Cadmium-Tellurium) detectors are performing under 200000 counts/s. By using these types of detectors, the acquisition time can be reduced from 1 min to 600 ms. Moreover, in the test system proposed, no filter was used in order to reduce the noise of the signal and the measurements were done without optimising the device. To achieve the objective of 10 ms acquisition time, further work is under progress with the objective of developing a pilot plant devoted to the optimization of the concentration level of tracers and the acquisition time.

Concerning the thermal and mechanical properties of the traced PP, rare earth oxides do not seem to have a major role at 0.1wt%. The SEM images show a homogenous dispersion of tracers in the PP matrix. More investigations are under progress in order to study the influence of rare earth oxides on the photo-degradation of the PP matrix.

Acknowledgement

The authors would like to thank the French Industry-University Cooperative Research Network on Waste (RECORD – www.record-net.org ), the French Environment and Energy Management Agency (ADEME – www.ademe.fr ), RENAULT, SITA and VEOLIA ENVIRONNEMENT for their contribution to the funding of this work and for providing industrial orientations and scientific supervision to the research. We would also like to acknowledge the National Recovery Technologies Inc. for helping us to carry out the tests in their X-ray fluorescence device.


http://www.ademe.fr/


57

References Ahmad, S.R.: A new technology for automatic identification and sorting of plastics for recycling. Environ. Technol. 25(10), 1143-1149 (2004) Ahmad, S.R.: Marking of products with fluorescent tracers in binary combinations for automatic identification and sorting. Assem Autom. 20(1), 58-64. (2000) Ahmad, S.R.: Partners wanted for polymer research project. Mater. Recycl. Weekly (UK), 14 (1992) Alam, M.K., Stanton, S.L., Hebner, G.A.: Near-infrared spectroscopy and neural networks for resin identification. Spectroscopy. 9(2), 30-40 (1994) APME. Annual Report 2007- Safe guarding the planet by reaching out. (2007) Bezati, F., Froelich, D., Massardier, V., Maris, E.: Addition of tracers into the polypropylene in view of automatic sorting of plastic wastes using X-ray fluorescence spectrometry. Waste Manage. 30, 591-596 (2010) Curlee, T.R., Das, S.: Plastics Wastes (Management, Control, Recycling and Disposal), Noyes Data Corporation, New Jersey, (1991) Dodbiba, G. Fujita, T.: Progress in separating plastic materials for recycling. Phys. Sep. Sci. Eng. 13(3-4), 165-182. (2004) European Parliament: Directive 2002/96/EC of the European Parliament and of the Council of 27 January 2003 on waste electrical and electronic equipment. (2003) European Parliament: Directive 2000/53/EC of the European Parliament and of the Council of 18 September 2000 on end-of-life vehicles. (2000) Florestan, J., Lachambre, A., Mermilliod, N., Boulou, J.C., Marfisi, C.: Recycling of plastics: Automatic identification of polymers by spectroscopic methods. Resour. Conserv. Recycl. 10(1-2), 67-74 (1994) Froelich, D., Maris, E., Haoues, N., Chemineau, L., Renard, H., Abraham, F., : State of the art of plastic sorting and recycling: Feedback to vehicle design. Minerals. Eng. 20, 902-912 (2007) Havrilla, G.J.: Handbook of Instrumental Techniques for Analytical Chemistry. 459-479 (1997) Hedrick, J.B.: Availability of rare earths. Am. Ceram. Soc. Bull. 67(5), 858-861(1988) Kenny, G.R., Bruner, R.S.: Experience and advances in automated separation of plastics for recycling. J. Vinyl. Add. Tech. 16(3), 181-186 (1994) Lauwerys, R.R., Haufroid, V., Huet, P., Lison D. : Toxicologie industrielle et intoxications professionnelles. Elsevier, Masson. Issy-les-Moulineaux (2007) Leong, Y.W., Ishak, Z.A.M., Ariffin, A.: Mechanical and thermal properties of talc and calcium carbonate filled polypropylene hybrid composites. J. Appl. Polym. Sci. 91(5), 3327-3336 (2004) Linder, C., Herold, M.: Plastic Waste in European Key Countries. APME. (2005) Liu, J., Tang, G., Qu, G., Zhou, H., Guo, Q.: Crystallization of rare earth oxide-filled polypropylene. J. Appl. Polym. Sci. 47(12), 2111-2116 (1993) Maestro, P., Huguenin, D.: Industrial applications of rare earths: which way for the end of the century. J. Alloys. Compd. 225(1-2), 520-528 (1995)


58

Moore, S.: Infrared scanner affords easy plastic identification. Modern Plastics (NY). 76, 32-33 (1999) Morteani, G.: The rare earths: their minerals, production and technical use. Eur. J. Mineral. 3(4), 641-650 (1991) Pannetier, R.: Momento du vade mecum nucléaire. SCF du Bastet, Paris, (1985) Pascoe, R.D.: Sorting of plastics using physical separation techniques. Recycling and Reuse of Waste Materials. Proceedings of the International Symposium. 173-188. (2003) Patel, M.K. Jochem, E., Radgen, P., Worrell, E.: Plastics streams in Germany - An analysis of production, consumption and waste generation. Resour. Conserv. Recycl. 24(3-4), 191-215. (1998) Peltier A. Exposition aux poussières de terres rares. 122 (1986) Simmons, B.A., Overton, B.W., Viriot, M., Ahmad, S.R., Squires, D.K., Lambert, C.: Fluorescent tracers enable automatic identification and sorting of waste plastics. Br. Plast. Rubber. 8, 4-6 (1998) Thio, Y.S., Argon, A.S., Cohen, R.E., Weinberg, M.: Toughening of isotactic polypropylene with CaCO3 particles. Polymer. 43(13), 3661-3674 (2002) Xiaomin, Z., Jingshu, L., Zhihui, Y., Jinghua Y.: Rheological properties and crystallization behavior of yittrium oxide filled low ethylene content polypropylene copolymer. J. Appl. Polym. Sci. 62(2), 313-318 (1996) Ye, C., Liu, J., Mo, Z., Tang, G., Jing, X.: Crystal structure of polypropylene filled with rare earth oxides. J. Appl. Polym. Sci. 60(11), 1877-1881 (1996) Zoboli, R.: Implications of environmental regulation on industrial innovation. Report carried out under the IPTS-DGIII framework project "Impact of the EU Regulation on Innovation of European Industry," Luxembourg, Office for Official Publications of the European Communities, (1998) Zuiderduin, W.C.J., Westzaan, C., Huétink, J., Gaymans, R.J.: Toughening of polypropylene with calcium carbonate particles. Polymer. 44(1), 261-275 (2003)


59

Part B: A study on the dispersion, elaboration, characterization and photo-degradation of traced

polypropylene with rare earth oxides

II.B.1. Introduction Tracers consist of systems formed by one or by several substances dispersed into a

material, to add a selective property to it, with the aim of improving the identification. In recent works (Bezati, 2010a, 2010b), an X-ray fluorescence (XRF) detection system based on the use of tracers, for the identification of polymer materials, in order to have an economically efficient automatic sorting of plastic wastes was reported by the authors. The addition of such selected tracers in polymer materials, during the compounding process, offer many benefits by increasing the efficiency and purity of sorting and providing a solution for the identification of dark plastics.

Tracers must not chemically react with the host material, and cause changes to any of their physical, chemical or mechanical properties beyond acceptable limits. For products which are likely to have long useful lives under normal outdoor environment, such as automotive parts, the tracer particles need to be highly stable under weathering and at the processing temperature of plastics (Ahmad, 2004).

The automotive and electrical industries are currently the worst performers, concerning the recycling of plastics, partly due to the complexity of the waste material that these sectors produce (Linder and Herold, 2005). For this study, polypropylene (PP), one of the world’s major plastics, abundantly present in polymers to be recycled from the End of Life Vehicles (ELV) and the Waste of Electric and Electronic Equipments (WEEE) was chosen.

XRF is a volume, non-destructive spectroscopic method enabling elemental analysis of materials; hence the choice of tracers is directly relied on the elements of Mendeleyev’s Periodic Table. Once from Mendeleyev’s Periodic Table are eliminated the toxic and radioactive elements, in addition to elements presented in polymer additives, low atomic number elements; which do not have an intense signal in XRF and elements which do not have important reserves the selection leads to the choice of rare earth elements. For these last ones we have chosen the most stable chemical molecules, the oxides (Sastry, 1966). The rare earth oxides (Y2O3, CeO2, Nd2O3, Gd2O3, Dy2O3, Er2O3 and Yb2O3) used for this study are neither abrasive, nor toxics during their application and use and also stable in the environment of evolution (Lauwerys et al, 2007; Morteani, 1991).

The minerals used as fillers in polypropylene are principally talc and calcium carbonate. However, little attention has been paid to rare earth compounds used as fillers in polymers. As rare earth minerals are abundant in China, some studies investigating polypropylene composites containing rare earth oxides were published by Chinese scientific committee. Liu and al, 1996, 1993, have studied the effect of a large number of rare earth oxides on isothermal crystallization and melting behaviour with differential scanning calorimetry. They found that a series of rare earth oxides in a fine powder form may act as a nucleator agent and


60

influence the growth rate of the spherulite and the mechanical properties of polypropylene. Other studies of Xiaomin et al, 1996 have shown that the addition of Y2O3 has some effect on the viscosity of the system. Crystallization characteristics have indicated that the filler acts as a nucleating agent, increasing crystallization rate of the investigated polymer, and changing β crystal form content.

Regarding the effect of rare earth oxides in photo-degradation, cerium oxide (CeO2) nanoparticles are used in wood coating technologies as UV absorbers (Fauchadour, 2005), as they show a UV cut-off threshold at around 370 nm, similar to that of titanium oxide. The use of fine particles of cerium oxide has also been mentioned as an effective inorganic sunscreen for personal care products for replacing titanium and zinc oxides (Sato, 2004). Tessier et al, 2008 have studied Y/Ce substitution of ceria in order to shift efficiently the absorption edge towards the UV/Vis transition (400 nm). They have concluded that the compositions proposed are interesting for UV absorbers applications for the wood industry. Moreover, the use of Yttrium oxide (Y2O3) was mentioned in literature for its application as a protector of aluminium and silver mirror coatings, due to its capacity to absorb widely in the near-UV spectrum (300 nm) (Bezuidenhout and Pretorius, 1986; Atanassov et al, 1993).

As mentioned above, previous works (Bezati, 2010a, 2010b) reported on the possibility of detecting rare earth oxides through the use of XRF for 0.1 wt% tracer content. The main objectives of this study are first, to quantitatively characterize the dispersion of tracers in the PP matrix which is of prime importance for their detection and the mechanical properties of the traced materials, second, to study the effect of tracers, as fillers, in the PP matrix with respect to the thermal and mechanical properties and to the photo-degradation under UV irradiation exposure, and last to estimate the tracer concentration for which the properties of the polymer matrix will not be affected.

II.B.2. Experimental

II.B.2.1. Materials and preparation of traced polypropylene composites

The investigation described in this paper was conducted with a commercial grade of

isotactic polypropylene homopolymer (ISPLEN PP 050 G1E) manufactured by Repsol. The original pellets have a melt mass-flow rate of 5.8 g/ 10min (2.16kg at 230°C) and a density of 0.905 g/cm3. ISPLEN PP 050 G1E is intended for applications that require good impact resistance and is believed to contain low quantities of UV stabilizers and a small amount of heat stabilizer in order to minimize degradation during processing.

The characteristics of the rare earth oxides used as tracers are given in Table 1. The tracer content in the composites is 0.1 and 1%, respectively. All traced materials were prepared under identical mixing and moulding conditions. The injection moulded specimens of 2 mm thickness were produced by using a Battenfeld Unilog B2/350 Plus injection molding machine. The barrel temperature was fixed at 220 °C for all zones. Before being injected the pellets were extruded twice by using a twin screw extruder CLEXTRAL BC21 machine


61

(screw of L/D = 90 and D = 25mm) for dispersing homogenously the tracers into PP matrix. The extrusion temperature, the screw speed and the residence time were 205°C, 120 rpm and 15 min, respectively. The screw and temperature profiles used in this study are supplied in Figure 1. Table 1 Materials specifications

Materials Chemical Name Supplier Particles' size (µm) PP Polypropylene REPSOL 2000 Y2O3 Yttrium Oxide RHODIA 2.250 CeO2 Cerium Oxide RHODIA 2.250 Nd2O3 Neodymium Oxide AMPERE 3.500 Gd2O3 Gadolinium Oxide RHODIA 1.850 Dy2O3 Dysprosium Oxide RHODIA 2.250 Er2O3 Erbium Oxide AMPERE 8.200 Yb2O3 Ytterbium Oxide AMPERE 1.000

Fig. 1. Screw and temperature profiles for the extrusion processing.

For PP containing 0, 0.1 and 1 wt% of tracers (T), the materials will be designed as PP,

PP_T_0.1 and PP_T_1, respectively. For the samples subjected to UV irradiation treatment, they will be referred to as UV-PP, UV-PP_T_0.1 and UV-PP_T_1, respectively.

All samples were kept for at least 1 month at room temperature before undergoing UV exposure in order to ensure that they had reached a near-equilibrium state, following ageing due to post-moulding effects such as secondary crystallization.

II.B.2.2. Microscopic observation

The tracers and their dispersion in PP matrix were observed by scanning electron microscopy (SEM) using a Hitachi S3500N model. The specimens prepared by injection moulding were fractured in liquid nitrogen and sputter-coated with gold before being examined with the microscope at an accelerating voltage of 30 kV. The SEM images were analysed by MATLAB by applying image processing.

II.B.2.3. UV irradiation procedure

UV irradiation treatment for the injection moulded specimens were carried out using a QUV accelerated weathering machine (QUV Solar Eye), with a light intensity of 0.68 W/m². The source of UV radiation was fluorescent tubes UVA-340, which have an output matching reasonably close to the solar radiation in the UV range (Labomat). The surface of the samples


62

was subjected to a cycle of 4 hours of UV irradiation at 60 °C and 4 hours of condensation at 50 °C. The total exposure time of specimens was 2160 hours (3 months).

II.B.2.4. Infrared spectroscopy

Samples used for Fourier transform infrared spectroscopy (FTIR), were collected from the surface to a depth of 0.5 mm and powdered. Approximately, 1-2 mg of the material powder was mixed and cold pressed with KBr to produce discs. FTIR measurements were obtained in transmission through the discs with a Nicolet Magna-IR 550 spectrometer. The equipment operates with a resolution of 2 cm-1 at a transmission mode using 16 scans.

II.B.2.5. Differential scanning calorimetry (DSC) characterization

Differential scanning calorimetry characterization (DSC Q10, TA Instruments) was

performed to investigate the crystallization and the melting behaviours of the traced material. Thin slices of samples about 0.5 mm thick and 5-10 mg weight were cut from the surfaces of the specimens for the measurement. Thermograms were recorded in three consecutive runs: (1) a first heating, from room temperature to 210 °C, followed by (2) cooling, from 210 to 25 °C after holding the sample at 210°C for 5 min, and finally (3) a second heating, from 25 to 210 °C. All experiments were performed at heating/cooling rate of 10 °C/min under argon atmosphere to avoid thermal degradation. The degree of crystallisation, χC, was calculated considering a melting enthalpy of 209 J/g for a 100% crystalline polypropylene (Brandrup and Immergut, 1989).

II.B.2.6. Thermal stability analysis

A TA Instruments Q500 Thermogravimetric Analyzer was used to characterize the thermal stability of specimens before and after UV irradiation. Measurements were conducted under an argon flow rate of 90mL/min at a heating rate of 20°C/min. The scanning temperature ranged from 25 to 500 °C. Similar to the DSC characterizations, thin slices of samples about 0.5 mm thick and 10-15 mg weight were cut from specimens for the measurement. The temperature at 5% weight loss (T_5%) and the temperature of maximum decomposition rate (Tmax) were calculated for each traced PP sample.

II.B.2.7. Mechanical characterisation

Mechanical testing was carried out on specimens, prepared according to ISO 527-1:1993, using an Instron 4469 uniaxial tensile testing machine. Yield stress and elongation at break were measured at ambient temperature for a cross-head speed of 250 mm/min and the values reported represent an average of six measurements.


63

II.B.3. Results and Discussions

II.B.3.1. Characterization of the dispersion of tracers

The homogenous dispersion of the filler particles in the PP matrix is of prime importance since it may strongly affect their detection and mechanical properties (Gendron and Binet, 1998; Liang, 2007). Therefore, a quantitative characterization of the dispersion status of the tracers is important for understanding their distribution in the PP matrix. Image analysis may be a powerful tool for describing quantitatively the dispersion of tracers, but this technique is subtle and delicate to use. The quality of results is highly dependent on user knowledge in this technique. In particular, the setting parameters are able to create artefacts or erase a part of the information and by this way bias the results. Consequently, the processed images should show a high contrast between the objects (rare earth oxides) and their environment (PP matrix). Moreover the image processing must be subject to an easy programmable criterion uniformly applicable to all images in question.

In our case, the objective is to extract the rare earth oxides particles from a polymer matrix through the use of scanning electron microscope. The use of back scattered electrons simplifies the identification of tracers since the atomic number of rare earth oxides is higher than the elements composing the polymer matrix. This difference in the atomic weight contrast makes them appear like white dots on a black background.

In order to have a global view of the distribution of tracers, the samples have to be fractured so as to visualize the particles in material’s depth. This consideration causes two main problems. First, the slice of polymer material at ambient temperature makes them melt in front of the blade and second, as rare earth oxides are rigid particles, they break in multiple parts under blade pressure by modifying their geometrical properties.

To avoid these problems so as to conserve charge and matrix geometry, the cryofracturation method was chosen for sample preparation. However, this method creates a relief in the SEM images which may interfere with the detection of rare earth oxides. Likewise, the fracture will propagate preferentially from charge to charge, as non specific interactions between charge and matrix are known, by probably inducing an over evaluation of charge population.

The SEM image of PP matrix containing 1% of gadolinium oxide is given in Fig. 2-A. Even if the image shows a good contrast between charge and matrix, the relief creates illumination variation which makes the charge extraction from the matrix complicated.

For this reason, it is necessary to obtain an image in which each pixel attributed to a charge will get the value “1” and each pixel attributed to the matrix will get the value “0”. One way to obtain this binarisation is by clipping. Threshold value is made by minimising environmental residue while maximising charge visibility. The choice of this value is a very important parameter and is taken by comparing the original SEM image to the proceed one.

Fig. 2-B shows the thresholding image for a filter of 37% of its total dynamic. In the dark part of the image it can be observed a loss of “white dots”, compared to the original SEM image, which may bias the distribution results.


64

One of the charge properties not used in previous thresholding process is their small size compared to the relief. This property can be used by applying a classic image processing named “unsharping”, consisting in blurring a copy of the image, mostly with a convolution by a Gaussian function, and subtract it to the original image with the aim of only saving the fine details. As it can be seen in Fig. 2-C, the relief is mitigated while the charge remains as white as before the processing.

After the “unsharping” process the image of Fig. 2-C was proceeded to a clipping too and the obtained image is shown in Fig. 2-D. A point to point comparison between the original image and the processed one shows that quality has been improved. Thus the charge surface on the image passed from 0.26% for the simple clipping to 0.38% for the unsharpped thresholding by increasing to 36% the charge gain.

Fig.2. SEM image of the cryofractured PP_Gd2O3_1 sample (A).

Clipping of Fig. 2-A for a filter power of 37% of its total dynamic (B). Unsharping of Fig. 2-A with a Gaussian of a 38.4 pixels equivalent σ (C).

Clipping of Fig. 2-C for a filter power of 20% of its total dynamic (D).

Once the image has been binarised, the coordinates of the constitutive points of the charge can be extracted and multiple parameters like gravity centre position of each particles and their size can be obtained. The gravity centre position can be used in order to characterize the dispersion of the charge in the matrix. The mean distance from a particle to all the others can be affected by boundary effects and the minimal distance to the first neighbours is extremely sensitive to object fragmentation. With regard to minimise boundary and local effects the mean distance to the ten first neighbouring particles was chosen as a representative value.

C

A B

D


65

Fig. 3 presents the curve of spatial dispersion of the first ten neighbouring particles distance for the PP_Gd2O3_1 sample. The noisy spectrum (raw curve) corresponds to the number of counted particles. This number is a consequence of the sampling and the scale chosen for the image. The image sampling should be chosen between 1/10 and 1/100 of the inter-particle distance in order to have a reasonable sampling noise on each measure. Thus, as the image have a characteristic scale of 1000x1000 pixel, the number of particles will be about 100 to 1000 by image (in our case, 1421 objects in a 2560 x 1920 pixels image), and as they are here represented on a scale of 1000 step, most of them are visible individually.

For comparing the results given by image processing, Fig. 3 shows the mean distances to the ten proximal neighbouring particles in a cubic and hexagonal close packing arrangement of the same surface and the same number of particles as in the case of the experimental data. As it can be seen in Fig. 4 the hexagonal network is more isotropic at low scale than the square one and this is the reason why the mean distance appears shorter.

Fig.3. Curve of spatial dispersion of charges, normalised to the most probable inter-particle distance for the

PP_Gd2O3_1 sample.

Fig.4. Cubic and hexagonal close packing arrangement representation.

Raw curve

Distance (µm)

Nor

mal

ized

pro

babi

lity

of e

xist

ence

Smoothed version of the raw curve

Inter-grain distance for a hexagonal net

Mean distance to the 10 proximal neighbors

Inter-grain distance for a cubic net

10)224(2dth_cubique

_δ+=

10)323(2d aleth_hexagon

_δ+=

Cubic close packing arrangement

Hexagonal close packing arrangement

δ δ


66

Fig. 5-A, B gives the results of particle’s diameter dispersion in number and in surface, respectively. As it can be seen, the distribution of particles’ diameter in number follows a decreasing exponential curve, which can be explained by the fact that during the extrusion process particles are subject to brittle in smallest pieces, whereas the distribution of particles’ diameter in surface is represented by a linearly decreasing curve showing the importance of each population.

Fig.5. Number average distribution of particles’ diameter for the PP_Gd2O3_1 sample (A)

Surface average distribution of particles’ diameter for the PP_Gd2O3_1 sample (B). The results obtained for the distance of the first ten neighbouring spheres (dcalculated)

calculated for the SEM data and the theoretical distance for a cubic (dth_cubic) and hexagonal (dth_hexagonal) close packing arrangement, together with average diameter in number (DNumber) and average diameter in surface (DSurface) are given in Table 2.

Table 2 Image processing results obtained by MATLAB

Samples dth_cubic (µm) dth_hexagonal (µm) dcalculated (µm) DNumber (µm) DSurface (µm) PP_Y2O3_1 17.99 12.91 14.78 ± 0.44 0.47 ± 0.01 1.20 ± 0.02 PP_CeO2_1 21.37 15.33 17.71 ± 0.53 0.47 ± 0.01 1.46 ± 0.02 PP_Nd2O3_1 19.79 14.20 17.33 ± 0.52 0.75 ± 0.02 2.12 ± 0.03 PP_Gd2O3_0.1 40.66 29.16 30.61 ± 3.06 0.39 ± 0.04 1.14 ± 0.06 PP_Gd2O3_1 17.73 12.71 14.48 ± 0.43 0.42 ± 0.01 1.12 ± 0.02 PP_Dy2O3_1 19.40 13.92 12.95 ± 0.39 0.40 ± 0.01 1.30 ± 0.02 PP_Er2O3_1 24.35 17.46 19.66 ± 0.59 0.50 ± 0.02 1.38 ± 0.02 PP_Yb2O3_1 18.56 13.31 15.90 ± 0.48 0.49 ± 0.01 1.26 ± 0.02

For characterizing the rare earth oxides dispersion in the PP matrix, the mean distance of

the first ten neighbouring spheres calculated for the SEM data (dcalculated) was compared with the theoretical distance for a cubic (dth_cubic) or hexagonal (dth_hexagonal) close packing arrangement. As dcalculated is ranged between the cubic and hexagonal theoretical values this implies that the dispersion isotropy is quite good and homogenous. Moreover the SEM image of Fig. 2-A leads to the same conclusion.

Particles’ diameter (µm)

Rel

ativ

e im

port

ance

nor

mal

ized

in n

umbe

r

DNumber

A


Rel

ativ

e im

port

ance

nor

mal

ized

in n

umbe

r

DNumber

A

Rel

ativ

e im

port

ance

nor

mal

ized

in n

umbe

r

DSurface

B


Rel

ativ

e im

port

ance

nor

mal

ized

in n

umbe

r

DSurface

B



67

II.B.3.2. Stability under UV irradiation/condensation cycles

The main products of polypropylene degradation, carbonyls, are easily observed in the

wavelength range [1700-1800 cm-1] by giving a fairly broad peak as shown in Fig. 6 (Rabello and White, 1995). In addition, for a more quantitative characterization of the PP molecular degradation, the carbonyl index, corresponding to the areas under the carbonyl peak divided by the area of a reference peak was calculated, Fig. 7. For the reference, several peaks were found in literature for PP (Rabello and White, 1995; Blinov et al, 1989; Livanova and Zaikov, 1992; Katbab and Moushirabadi, 1991), as 840, 1166, 1455 and 2720 cm-1. The peak at 1455 cm-1 is overlapped with several others (Morales and White, 1988), whereas 840 and 1166 cm-1

are sensitive to PP crystallinity (Karcan et al, 1993). Consequently, the peak at 2720 cm-1, associated with CH3 stretching and CH bending, was chosen as the most appropriate (Rabello and White, 1995; Zhao, 2006).

Fig.6. FTIR spectrum of unfilled PP and samples containing the tracers at 0.1 and 1 wt% after exposure to UV irradiation/condensation cycles for 3 months (A). The spectrum of unexposed unfilled PP is also shown (B).

Fig.7. Effect of tracers and their content on the carbonyl index.

Tra

nsm

issi

on %

3000 2500 2000 1500 1000

Wavenumber (cm-1)

UV-PP

UV-PP_Y2O3_0.1

PP

UV-PP_CeO2_0.1

UV-PP_Nd2O3_0.1

UV-PP_Gd2O3_0.1

UV-PP_Dy2O3_0.1

UV-PP_Er2O3_0.1

UV-PP_Yb2O3_0.1

Tra

nsm

issi

on %

3000 2500 2000 1500 1000

Wavenumber (cm-1)

UV-PP

UV-PP_Y2O3_0.1

PP

UV-PP_CeO2_0.1

UV-PP_Nd2O3_0.1

UV-PP_Gd2O3_0.1

UV-PP_Dy2O3_0.1

UV-PP_Er2O3_0.1

UV-PP_Yb2O3_0.1

T

rans

mis

sion

%

3000 2500 2000 1500 1000

Wavenumber (cm-1)

UV-PP

UV-PP_Y2O3_1

PP

UV-PP_CeO2_1

UV-PP_Nd2O3_1

UV-PP_Gd2O3_1

UV-PP_Dy2O3_1

UV-PP_Er2O3_1

UV-PP_Yb2O3_1

Tra

nsm

issi

on %

3000 2500 2000 1500 1000

Wavenumber (cm-1)

UV-PP

UV-PP_Y2O3_1

PP

UV-PP_CeO2_1

UV-PP_Nd2O3_1

UV-PP_Gd2O3_1

UV-PP_Dy2O3_1

UV-PP_Er2O3_1

UV-PP_Yb2O3_1

0

5

10

15

20

25

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1Tracer content (wt%)

Car

bony

l ind

ex

Y2O3

CeO2

Nd2O3 Gd2O3

Dy2O3 Er2O3

Yb2O3

A B


68

By comparing in Fig. 6, the area evolution of the carbonyl peak absorption band (in the range [1700-1800 cm-1] between the exposed samples and the unexposed unfilled PP, it is clearly visible that it increases after the UV irradiation treatment. From the FTIR spectrums, the lowest evolution after the UV exposition is observed for the UV-PP_CeO2_1 sample. For the remaining samples the evolution of the carbonyl peak is similar to the UV-PP submitted to weathering cycles.

The results obtained for the carbonyl index show that the carbonyl peak intensities decrease slightly with tracer content, except the CeO2 for which the carbonyl index decreases drastically compared to the UV-PP. Cerium oxide is well known to be an UV absorber (Tessier et al, 2008), and these results show that it plays an important role in PP stabilization by delaying the photo-degradation process. The CeO2 particles act as optical filters by absorbing part of the UV irradiation and releasing the excess energy as heat, thus the UV intensity which can promote the oxidation of the PP chains is reduced. The same explication can be given for the Y2O3 particles which absorb the beginning of the UV irradiation at 300 cm-1 and this is the reason why its carbonyl index has a more important diminution compared to the other tracers.

By comparing the unfilled PP sample with the tracer filled PP samples after the UV irradiation treatment, no new peaks are observed, suggesting that their photo-degradation mechanisms are identical. Similar photo-stabilisation effect of metal oxides has also been reported by Zhao et al, 2006, Tessier et al, 2008 and Sato et al, 2004.

Regarding the remaining tracers (Nd2O3, Gd2O3, Dy2O3, Er2O3 and Yb2O3) their effect on the photo-degradation of PP seems not to play an important role.

II.B.3.3. Effect of tracers on thermal and crystallization behaviours

The thermal and crystallization data obtained from DSC for the unfilled PP and the traced

polymer composites before UV irradiation exposure are summarized in Table 3. As it can be reflected by ∆Hf, Tm, ∆Hc, and Tc the melting and crystallization behaviours of the PP matrix are slightly affected by the tracer addition before UV irradiation treatment. Regarding the PP crystallinity (χc) the values are so close for considering a variation with the tracer addition. Even if it was expected that the rare earth oxides will act as nucleating agents (Liu et al, 1993; Ye et al, 1996; Xiaomin et al, 1996) in polymer crystallization by increasing the crystallinity, their micrometric size seems to prevent them for having a real impact. In fact the average particle size of the rare earth oxides studied by Liu et al, 1993, 1996 was in the range [0.05-0.6 µm], in contrary to the particles used in this study that was superior to 1 µm.

By comparing the data in Tables 3 and 4 and the DSC thermograms displayed in Fig. 8, it can be seen that UV irradiation exposure has a real impact on both melting and crystallization behaviours of the PP matrix. The melting temperature of the unfilled PP dropped from 166.7 to 137.6 °C after 3 months of UV irradiation. This decrease of about 30 °C, is observed for almost all the traced samples, except the one containing CeO2 at 1 wt% (PP_CeO2_1). The chain scission caused by photo-degradation can release the entanglement of molecules and produce more freed segments, thus the recrystallized material (Fig. 8)


69

contains molecules that are smaller and defective due to groups like carbonyls. Consequently, the reduction of molecular weight leads to the decrease of the melting temperature. Table 3 Thermal and crystallization data obtained from DSC before UV irradiation exposure

Sample ∆Hf (J/gPP ) Tm (°C) ΔHc (J/gPP) Tc (°C) χc (%) PP 110.3 167.0 100 105 52.8 PP_Y2O3_0.1 112.5 166.5 102.6 106.9 53.8 PP_CeO2_0.1 109.6 166.1 99.5 108.1 52.4 PP_Nd2O3_0.1 108.1 167.1 98.8 106.3 51.7 PP_Gd2O3_0.1 109.3 166.1 99.3 108.4 52.3 PP_Dy2O3_0.1 108.9 167 99.4 105.1 52.1 PP_Er2O3_0.1 106.9 166.6 96.3 105.5 51.1 PP_Yb2O3_0.1 111.1 166.8 101.1 105.9 53.2 PP_Y2O3_1 111.6 166.6 102.6 105.9 53.4 PP_CeO2_1 106.9 166.3 99 107.4 51.2 PP_Nd2O3_1 110.5 166 98.4 106.3 52.9 PP_Gd2O3_1 110.3 167.2 102.4 108.3 52.8 PP_Dy2O3_1 106.6 167.1 97.7 107.6 51.0 PP_Er2O3_1 107.3 166.4 98.3 105.9 51.3 PP_Yb2O3_1 107.5 166.8 99.2 106.1 51.4

Table 4 Thermal and crystallization data obtained from DSC after UV irradiation exposure

Sample ∆Hf (J/gPP ) Tm (°C) ΔHc (J/gPP) Tc (°C) χc (%) UV-PP 97.8 137.6 88.6 100.3 46.8 UV-PP_Y2O3_0.1 97.9 139.8 86.1 100.8 46.8 UV-PP_CeO2_0.1 93.4 144.8 97.8 106.8 44.7 UV-PP_Nd2O3_0.1 100.5 139.1 89.1 100.0 48.1 UV-PP_Gd2O3_0.1 106.9 138.4 94.3 101.7 51.1 UV-PP_Dy2O3_0.1 102.1 135.8 91.6 100.4 48.9 UV-PP_Er2O3_0.1 107.6 139.6 92.9 100.6 51.5 UV-PP_Yb2O3_0.1 92.4 138.4 92.2 102.1 44.2 UV-PP_Y2O3_1 96.4 138.2 96.7 101.6 46.1 UV-PP_CeO2_1 92.5 164.8 98.8 107.2 44.3 UV-PP_Nd2O3_1 95.2 141.1 88.8 101.7 45.6 UV-PP_Gd2O3_1 99.4 138.2 86.6 101.2 47.6 UV-PP_Dy2O3_1 95.1 145.6 91.2 102.6 45.5 UV-PP_Er2O3_1 95.2 140.6 92.8 101.3 45.6 UV-PP_Yb2O3_1 104.1 141.5 91.3 100.3 49.8


70

Fig. 8. Second heating run DSC thermograms of the PP samples of 1 wt% tracer content exposed to UV

irradiation compared to the unexposed unfilled PP.

II.B.3.4. Effect of tracers on thermal stability

The addition of rare earth oxides in the PP matrix could improve the thermal stability and onset degradation (Xiaomin et al, 1996). The TGA measurement results before and after the UV irradiation treatment, are shown in Fig. 9 and Table 5.

Table 5 TGA measurement results before and after three months of UV irradiation exposure

Sample Before UV irradiation exposure After UV irradiation exposure

*T_5% (°C) *Tmax (°C) T_5% (°C) Tmax (°C) (UV-)PP 431.3 471.7 281.0 460.6 (UV-)PP_Y2O3_0.1 432.5 473.9 271.0 456.7 (UV-)PP_CeO2_0.1 433.5 465.0 363.6 464.5 (UV-)PP_Nd2O3_0.1 434.5 473.6 277.2 456.1 (UV-)PP_Gd2O3_0.1 432.9 475.0 274.1 452.8 (UV-)PP_Dy2O3_0.1 433.6 471.7 243.2 448.9 (UV-)PP_Er2O3_0.1 435.1 473.9 267.1 456.1 (UV-)PP_Yb2O3_0.1 435.9 475.5 285.1 457.9 (UV-)PP_Y2O3_1 433.5 473.4 287.7 462.2 (UV-)PP_CeO2_1 434.4 472.8 405.6 468.9 (UV-)PP_Nd2O3_1 434.7 473.9 288.8 459.5 (UV-)PP_Gd2O3_1 433.9 476.2 271.9 457.8 (UV-)PP_Dy2O3_1 435.1 472.8 260.8 459.5 (UV-)PP_Er2O3_1 433.7 475.6 268.7 453.9 (UV-)PP_Yb2O3_1 433.8 474.5 265.2 453.9

* T_5%: Temperature at 5% weight loss, Tmax: temperature of maximum decomposition rate

UV-PP

UV-PP_Y2O3_1

PP

UV-PP_CeO2_1

UV-PP_Nd2O3_1

UV-PP_Gd2O3_1

UV-PP_Dy2O3_1

UV-PP_Er2O3_1

UV-PP_Yb2O3_1

End

othe

rmic

Temperature (°C)

UV-PP

UV-PP_Y2O3_1

PP

UV-PP_CeO2_1

UV-PP_Nd2O3_1

UV-PP_Gd2O3_1

UV-PP_Dy2O3_1

UV-PP_Er2O3_1

UV-PP_Yb2O3_1

UV-PP

UV-PP_Y2O3_1

PP

UV-PP_CeO2_1

UV-PP_Nd2O3_1

UV-PP_Gd2O3_1

UV-PP_Dy2O3_1

UV-PP_Er2O3_1

UV-PP_Yb2O3_1

End

othe

rmic

Temperature (°C)


71

From the TGA curves of Fig. 9, it is obvious that the thermal decomposition for all the samples is a one-step process. PP thermally degrades to volatile products above 250 °C through a radical chain reaction process propagated by carbon centred radicals due to the lack of oxygen. The formed radicals initiate and propagate the subsequent radical chain reactions (Zhao, 2006; Zanetti et al, 2001).

In Fig. 9-A, it can be seen that the TGA curves of the samples containing the tracers are nearly overlapping with the unfilled PP. By comparing the results before the UV irradiation exposure in Table 5, it results that the temperatures of 5% weight loss (T_5%) and of maximum decomposition rate (Tmax) for the traced materials are superior of 1 to 3 °C than the unfilled PP. This means that rare earth oxides could provide thermal stability to the PP matrix, even though better results could be obtained by increasing their content (Xiaomin et al, 1996).

Fig.9. TGA curves of the PP samples of 1 wt% tracer, before (A) and after (B) UV irradiation exposure. Fig. 9-B shows that the majority of TGA curves of the UV irradiation exposed traced

samples are close to the unfilled UV-PP, except the UV-PP_CeO2_1 sample for which the TGA curve is closer to the unexposed unfilled PP. Moreover, by comparing the results of T_5% for the UV irradiation exposed samples with the unexposed unfilled PP sample, it appears that it drops dramatically of about 125 °C. On the contrary, for the samples containing the CeO2 this diminution is of about 70 °C and 25 °C for the 0.1 and 1 wt% tracer content, respectively. The positive thermal stability provided by cerium oxide can be explained by the protective effect upon UV irradiation due to the UV screening properties of CeO2 particles.

PPPP_Y2O3_1PP_CeO2_1PP_Nd2O3_1PP_Gd2O3_1PP_Dy2O3_1PP_Er2O3_1PP_Yb2O3_1

Wei

ght l

oss (

%)

Temperature (°C)

APPPP_Y2O3_1PP_CeO2_1PP_Nd2O3_1PP_Gd2O3_1PP_Dy2O3_1PP_Er2O3_1PP_Yb2O3_1

Wei

ght l

oss (

%)

Temperature (°C)

A

B

Wei

ght l

oss (

%)

Temperature (°C)

UV-PP

PP

UV-PP_Nd2O3_1

UV-PP_Y2O3_1

UV-PP_CeO2_1

B

Wei

ght l

oss (

%)

Temperature (°C)

UV-PP

PP

UV-PP_Nd2O3_1

UV-PP_Y2O3_1

UV-PP_CeO2_1


72

II.B.3.5. Effect of tracers on mechanical properties

In previous work (Bezati et al, 2010b), it has been shown that the addition of 1 wt% of Nd2O3 and Gd2O3 leads to a slight increase of the Young’s modulus and a slight decrease of the yield stress. Regarding the tensile tests, the most important effect of these rare earth oxides was observed on the elongation at break. Fig. 10 presents the results of elongation at break for all the tracers used in this study before UV irradiation exposure.

Fig.10: Elongation at break as a function of tracer content before UV irradiation treatment.

Predictably, the obtained results reveal that the elongation at break drops upon increasing

the filler fraction (Thio et al, 2002; Leong et al, 2004; Zuiderduin et al, 2003). However, at 0.1 wt% tracer content, the elongation at break slightly decreases from 5 to 25% with a standard deviation of about 5 to 15%. The obtained values are close to the standard deviation of measurements and hence it can be considered that the influence of rare earth oxides in mechanical properties is minor at 0.1 wt%.

As it can be seen in Fig. 11, the ductility of the tensile specimens was reduced considerably after UV irradiation treatment. Indeed, the majority of the samples, except UV-PP_CeO2_0.1 and UV-PP_CeO2_1, were brittle and it was difficult to carry out the tensile tests. The results of elongation at break, Fig. 12, were obtained for the samples submitted to 10 days UV irradiation treatment, the tensile tests results of UV-PP_CeO2_0.1 and UV-PP_CeO2_1 for 10 days and 3 months UV irradiation exposure are displayed in Fig. 13.

Fig.11. Picture of specimens after 3 months UV irradiation treatment.

50

60

70

80

90

100

110


Elo

ngat

ion

at b

reak

(%)

Y2O3 CeO2 Nd2O3Gd2O3

Dy2O3Er2O3Yb2O3

UV-PP_Yb2O3_1

UV-PP_Er2O3_1

UV-PP_Dy2O3_1

UV-PP_Gd2O3_1

UV-PP_Nd2O3_1

UV-PP_CeO2_1

UV-PP_Y2O3_1

UV-PPUV-PP_Yb2O3_1

UV-PP_Er2O3_1

UV-PP_Dy2O3_1

UV-PP_Gd2O3_1

UV-PP_Nd2O3_1

UV-PP_CeO2_1

UV-PP_Y2O3_1

UV-PP


73

Fig.12. Elongation at break as a function of tracer content after 10 days UV irradiation treatment.

Fig.13. Effect of UV irradiation exposure on the elongation at break (A) and yield stress (B) for the CeO2 traced

PP samples. Fig. 12 shows that elongation at break decreases significantly from about 100 % to 20 %

for the samples exposed to UV irradiation. However, for the majority of samples containing the rare earth oxides at 0.1 and 1 wt% the values of elongation at break are similar to the unfilled exposed PP. The unique exception of the traced samples are UV-PP_CeO2_0.1 and UV-PP_CeO2_1 for which elongation at break and yield stress (Fig. 13) increase with CeO2 particle content.

Based on the mechanical tests results, it can be summarized that Y2O3, Nd2O3, Gd2O3, Dy2O3, Er2O3 and Yb2O3 of micrometric mean particle size, have a minor effect on the photo-degradation of PP matrix at 0.1 wt%. On the contrary the addition of 1 wt% of CeO2 can act as an effective UV light screen for reducing the degree of photo-degradation of PP matrix.

II.B.4. Conclusion Results from this research work reveal that the addition of rare earth oxides of

micrometric particle size at 0.1 wt% content into PP matrix has a minor effect on the mechanical and thermal properties of the traced materials as well as in the photo-degradation of the polymer after UV irradiation exposure.

0

10

20

30

40


Elo

ngat

ion

at b

reak

(%)

Y2O3

CeO2

Nd2O3

Gd2O3

Dy2O3 Er2O3

Yb2O3

0

20

40

60

80

100


Elon

gatio

n at

bre

ak (%

)

Before UV exposure After 240 hours of UV exposure After 2160 hours of UV exposure

A

0

10

20

30

40

50

60


Yie

ld st

ress

(MPa

)

Before UV exposure After 240 hours of UV exposure After 2160 hours of UV exposure

B


74

However, addition of 1 wt% of rare earth oxides particles into PP matrix decreases the elongation at break from 10 to 50 % before UV irradiation treatment for a cross-head speed of 250 mm/min. For the same concentration content, the melting and crystallization behaviours are slightly increased, whereas the thermal stability is improved by increasing the temperatures of 5% weight loss and maximum decomposition rate by 1 to 3 °C.

Regarding the UV irradiation treatment, the results of FTIR, DSC and ATG analysis together with mechanical characterization pointed out that the addition of CeO2 particles into PP matrix, which is believed to contain low quantities of UV stabilizers, can lead to significant improvements on the photo-degradation resistance of PP to UV exposure. Indeed, the CeO2 particles act as UV filters by absorbing part of the UV irradiation and thus reducing the UV intensity which can promote the oxidation of the PP chains. For the remaining tracers, the results obtained show the same behaviour on the photo-degradation as in the case of the unfilled PP matrix.

The dispersion effect of tracers into PP matrix was quantitatively characterized by applying image processing to the SEM images with the help of MATLAB. The distance to the first ten neighbouring spheres calculated for the SEM data was compared with the theoretical distance for a cubic and a hexagonal close packing arrangement of the same surface and the same number of particles as in the case of the experimental data. The SEM images and the results obtained from image processing show a homogenous dispersion of tracers into PP matrix.

In order to study the effect of rare earth oxides into PP matrix, a polypropylene homopolymer free of fillers as talc and calcium carbonate and which is believed to contain only low quantities of UV stabilizers was chosen in this study. As it was shown, CeO2 particles can act as UV absorbers by protecting the polymer matrix from photo-degradation. Complementary work can be carried out with other grades of PP, more adapted to outdoor environment applications, for studying probable antagonistic effects between the CeO2 particles and the UV stabilizers already contained in the polymer matrix.

Acknowledgements

The authors would like to thank the French Industry-University Cooperative Research Network on Waste (RECORD – www.record-net.org ), the French Environment and Energy Management Agency (ADEME – www.ademe.fr ) for their contribution to the funding of this work and for providing industrial orientations and scientific supervision to the research.




75

References Ahmad SR. Environ Technol 2004; 25:1143-1149. Atanassov G, Thielsch R, Popov D. Thin Solid Films 1993;223:288-292. Bezati F, Froelich D, Massardier V, Maris E. Waste Manage 2010;30:591-596. Bezati F, Massardier V, Froelich D, Maris E, Balcaen J. Waste Biomass Valor. Accepted June 2010. Bezuidenhout DF, Pretorius R. Thin Solid Films 1986;139:121-132. Blinov NN, Popov AA, Rakovski SK, Stoyanov AK, Shopov DM, Zaikov GY. Polymer Science U.S.S.R. 1989;31:2434-2439. Brandrup J, Immergut EH. Polymer Handbook 3rd edition. New York: Wiley; 1989. Fauchadour D, Jeanson T, Bousseau JN, Echalier B. Application to Coatings Technologies. 2005. Gendron R, Binet D. J Vinyl Addit Techn 1998;4:54-59Labomat : http://www.labomat.com/Produits/QUV.php Karcan I, Taraiya AK, Bower, DI, Wardi M. Polymer 1993;34:2691-2701. Katbab AA, Moushirabadi AY. Radiat Phys Chem 1991;38:295-301. Lauwerys RR, Haufroid V, Huet P, Lison D. : Toxicologie industrielle et intoxications professionnelles. Issy-les-Moulineaux: Elsevier Masson; 2007. Leong YW, Ishak ZAM, Ariffin A. J Appl Polym Sci 2004;91:3327-3336. Liang JZ. Composites 2007;38:1502-1506. Linder C, Herold M. Plastic Waste in European Key Countries. Plastics Europe; 2005. Livanova NM, Zaikov GE. Polym Degrad Stabil 1992;36:253-259. Liu J, Tang G, Qu G, Zhou H, Guo Q. J Appl Polym Sci 1993;47:2111-2116.Morteani G. Eur. J. Mineral. 1991;3:641-650. Morales E, White JR. J Mater Sci 1988;23:3612-3622. Rabello M, White JR. Polym Degrad Stabil 1995;56:55-73. Sastry RLN, Yoganarasimhan SR, Mehrotra PN, Rao CNR. J Inorg and Nucl Chem 1966;28:1165-1177. Sato T, Katakuraa T, Yina S, Fujimotob T, Yabec S. Solid State Ionics 2004;172:377-382. Tessier F, Cheviré F, Muñoz F, Merdrignac-Conanec O, Marchand R, Bouchard M, Colbeau C. J Solid State Chem 2008;181:1204-1212. Thio YS, Argon AS, Cohen RE, Weinberg M. Polymer 2002;43:3661-3674. Xiaomin Z, Jingshu L, Zhihui Y, Jinghua Y. J Appl Polym Sci 1996;62:313-318. Ye C, Liu J, Mo Z, Tang G, Jing X. J. Appl. Polym. Sci. 1996;60:1877-1881 Zanetti M, Camino G, Reichert P, Mulhaupt R. Macromol Rapid Comm 2001;22:176-180. Zhao H, Li RKY. Polymer 2006;47:3207-3217. Zuiderduin WCJ, Westzaan C, Huétink J, Gaymans RJ. Polymer 2003;44:261-275.


76

Conclusion Chapitre II

La dispersion des traceurs dans la matrice polypropylène est étudiée par microscopie électronique à balayage et, ensuite, ces images sont traitées par Matlab pour une analyse quantitative. Ce paramètre est un des très importants dans notre étude puisque la bonne dispersion des traceurs dans la matrice polypropylène permet une bonne détection par SFX. D’autre part, la bonne dispersion de la charge-traceur dans la matrice, lui permet de ne pas trop influencer ses propriétés mécaniques. A partir de ces images MEB et de leur analyse, nous avons remarqué une très bonne dispersion des traceurs dans la matrice polypropylène, sans formation d’agglomérats.

Pour les propriétés mécaniques, en ce qui concerne le module d’Young et la contrainte maximale, les essais de traction ont montré que la présence du traceur dans la matrice polypropylène n’a pas une influence significative et les variations observées sont comprises dans la fourchette d’incertitudes des mesures. Par contre, l’allongement à la rupture diminue en ajoutant des oxydes de terres rares, et plus, spécialement, quand on augmente la teneur en traceur, comparé au PP référence.

Concernant les propriétés physico-chimiques, de légères différences sont observées pour les températures de fusion et de cristallisation, les enthalpies de fusion et de cristallisation ainsi que le taux de cristallinité par rapport au polypropylène de référence. Ces variations sont de l’ordre des incertitudes de l’appareil et des mesures, et, de ce fait, peuvent être considérées comme négligeables.

La très bonne dispersion des oxydes de terres rares dans la matrice polypropylène, citée auparavant, peut expliquer le fait que nous observons de faibles variations dans les propriétés mécaniques et physico-chimiques, avec pour seule exception l’allongement à la rupture.

Pour ce qui est de l’impact des oxydes de terres rares sur la matrice PP après le vieillissement climatique, en suivant l’évolution du pic d’absorption des groupes carbonyles de l’analyse IRTF, il peut être observé que l’irradiation UV provoque une photo-dégradation importante sur le PP référence et les autres échantillons, à l’exception de l’oxyde de cérium. Cependant, en dispersant des particules d’oxyde de cérium dans la matrice polypropylène, l’impact de la photo-dégradation a nettement diminué. Cela est dû au rôle important que jouent les particules de CeO2 dans la stabilisation du polypropylène en retardant le processus de photo-dégradation et en agissant comme des absorbeurs des irradiations UV.

La dégradation induite par les irradiations UV a provoqué une chute significative de la ductilité de tous les échantillons. Les tests mécaniques, réalisés uniquement sur les échantillons contenant l’oxyde de cérium, ont montré que l’allongement à la rupture augmente avec la teneur en CeO2. De même, à partir des thermogrammes DSC, une influence sur la cristallisation et la fusion de la matrice polypropylène a été remarquée sur les échantillons vieillis. Pour le PP référence et la plupart des autres échantillons, à l’exception de celui contenant l’oxyde de cérium, une forte diminution de la température de fusion est observée. La décomposition thermique pour l’UV-PP se produit à une température nettement plus faible et à une vitesse plus rapide que pour l’échantillon contenant l’oxyde de cérium.

Pour conclure, les résultats expérimentaux présentés dans les deux articles ont montré que l’addition de 0,1% d’oxyde de terre rare de taille micrométrique dans une matrice


77

polypropylène a un impact non significatif sur les propriétés mécaniques et physico-chimiques, ainsi que sur la photo-dégradation sous rayonnement UV.

78

CHAPITRE III

Etude de la détectabilité à

haut vitesse par SFX des

traceurs sélectionnés,

dispersés en faible en

concentration dans la matrice

PP

« Ne remettez pas au lendemain ce que vous pouvez faire le jour même »

Benjamin Franklin

79

Addition of tracers into the polypropylene in view of automatic sorting of plastic wastes

using X-ray fluorescence spectrometry

F. Bezati, D. Froelich, V. Massardier, E. Maris

Waste Management 30 (2010) 591–596

Comparison of X-ray detectors for the optimisation of high speed detection of rare earth oxides, used as tracers into a

polymer matrix

F. Bezati, D. Froelich, D. Chambellan, S. Legoupil,

E. Maris, V. Massardier,

Soumis à “Applied Radiation and Isotopes”

80

Sommaire

Part A: Addition of tracers into the polypropylene in view of automatic sorting of plastic wastes using X-ray fluorescence spectrometry .......... 82

III.A.1. Introduction ............................................................................ 82

III.A.2. The identification of tracers .................................................... 83

III.A.3. Experimental part .................................................................... 84

III.A.3.1.Materials ................................................................... 84

III.A.3.2.Dispersion of tracers ................................................... 85

III.A.3.3.X-ray fluorescence device ............................................ 85

III.A.4. Results and discussions ........................................................... 87

III.A.4.1.Detection of tracers and concentration effect ................. 87

III.A.4.2.The overlapping and mass absorption effect ................... 88

III.A.4.3.The effect of the acquisition time .................................. 89

III.A.5. Conclusion and further research .............................................. 90

References ............................................................................................ 92

Part B: Comparison of X-ray detectors for the optimisation of high speed detection of rare earth oxides, used as tracers into a polymer matrix ... 94

III.B.1. Introduction ............................................................................ 94

III.B.2. Theory ..................................................................................... 95

III.B.3. Methods and materials ............................................................ 96

III.B.3.1.Characteristics of detectors and tracers ......................... 96

III.B.3.2.Energy dispersive X-ray fluorescence device .................. 97

III.B.3.3.Software overview ...................................................... 98

III.B.3.4.Calculation of detection limit ........................................ 99

III.B.4. Results and Discussions ........................................................ 101

III.B.4.1.Comparison of the detectors with respect to experimental results 101

III.B.4.2. Modeling results of the commercial detectors more adapted for the high speed detection of rare earth oxides .........................................103

III.B.4.2.1.Detectors specifications ........................................103

III.B.4.2.2.The detector and filter thickness effect ...................104

III.B.4.2.3.The X-ray generator voltage effect .........................107

III.B.4.2.4.The overlapping and mass absorption effect ............108

III.B.4.3.Comparison between experimental and modeling results 109

III.B.5. Conclusion ............................................................................. 110

References .......................................................................................... 112

81

Introduction Chapitre III

Ce Chapitre sera consacré à la détection par SFX des traceurs choisi dans le Chapitre 1. Il est le résultat d’une association de deux articles. Le premier présente les résultats expérimentaux obtenus dans l’entreprise NRT tandis que le deuxième montre les résultats de modélisation ainsi que des tests expérimentaux effectués au Laboratoire Imagerie, Tomographie et Traitement du CEA de Saclay.

Le 1er article, publié dans la revue « Waste Management », présente des tests de détection de traceurs qui ont eu lieu sur un dispositif de l’entreprise NRT, permettant d’obtenir des mesures statiques des échantillons en utilisant l’analyse dispersive en énergie. L’objectif principal de ces tests était de prouver la détection des oxydes de terre rare afin de valider leur choix en tant que traceurs. Dans un deuxième temps, nous avons voulu évaluer quelques paramètres qui peuvent influencer leur détection, tels que la concentration en traceur, le temps d’acquisition et la possibilité de chevauchement des pics entre les traceurs.

Dans le 2e article sont présentés les essais expérimentaux ainsi que les résultats d’une modélisation réalisée au LITT du CEA de Saclay. L’objectif de ces essais était l’estimation de la limite de détection, un paramètre très important d’un point de vue économique, environnemental et pouvant affecter les propriétés du matériau tracé. Cette limite de détection dépend de la concentration en traceur, du temps d’acquisition, ainsi que du taux de comptage net et du bruit de mesure. Le temps d’acquisition est fixé par le fabricant de machine de tri est doit être dans la fourchette [1-10 ms]. En conséquence, les paramètres qui peuvent jouer un rôle sur la limite de détection sont le taux de comptage net et le bruit de mesure. Ces deux derniers dépendent fortement du type de détecteur utilisé et c’est pour cette raison que nous avons étudié trois types de détecteurs (SDD, HPGe et CdTe) afin de déterminer a priori celui qui serait le plus adapté pour la détection des oxydes de terre rare. Il existe des détecteurs proches mais avec des épaisseurs différentes qui pourraient être plus adaptés à notre problématique. Ainsi, nous avons évalué la réponse de ces détecteurs par modélisation afin d’estimer l’ordre de grandeur de la limite de détection. Pour chaque détecteur, nous avons aussi étudié grâce à la modélisation des paramètres secondaires, tels que l’épaisseur du filtre, la tension du générateur de rayons X et la possibilité de chevauchement des pics entre les traceurs.

CHAPITRE III : Etude de la détectabilité à haut vitesse par SFX des traceurs sélectionnés

82

Part A: Addition of tracers into the polypropylene in view of automatic sorting of plastic wastes using X-ray

fluorescence spectrometry

III.A.1. Introduction

Thanks to their technical and economical performances, plastics cover today a large and varied range of uses (packaging, building, automobile, electrical and electronic equipments, etc.). Once they have completed their task, the objects become wastes and so it is necessary to manage their future.

In order to reduce the growing volume of plastic waste, the management of their treatment has become a high priority on the political, economical and environmental agenda of all industrialized nations.

To complete the European Commission directives, dealing with the management of the end-of-life post-consumer products (European Parliament, 2000, 2003), concerned manufacturers are encouraged to develop solutions which reduce the impact of their end-of-life products, and to have a real policy of raw resource management based on a relevant recycling plan. Furthermore, the European legislations and directives (Zoboli, 1998) require producers to use a fraction of their recycled material in their final product.

Currently, the energetic recovery of plastic wastes is achieved through easy options such as incineration, which is often neither economically profitable nor respectful to the environment (Olofsson, Sundberg, Sahlin, 2005; Patel et al., 1998). However, it is generally difficult to recycle these materials because of their contamination with other plastic materials that are incompatible with each other. Therefore, it is necessary to improve technologies such as the sorting of polymer materials so as to make their recycling profitable. For an economically efficient recycling of polymer materials, waste plastics need to be sorted cheaply and automatically into individual types and grades (Froelich et al., 2007).

Techniques based on optical spectroscopy, such as infrared reflection/absorption (Alam, Stanton, Hebner, 1994; Florestan et al., 1994; Moore, 1999) have reached their limits. This technique is not applicable to dark plastics of automotive parts; it cannot identify different grades of the same polymers and cannot be used if the surface of the plastic wastes is wet.

The technique of high resolution imaging using X-rays is limited to the separation of PVC from PET (Kenny and Bruner, 1994) and the elimination of PVC and bromide aromatic compounds contained in electronic waste and substituted combustibles. In the automobile recycling industry this technique is limited to the identification of heavy metals contained in aluminium.

All the above methods do not use a tracer in order to facilitate detection and sorting. In early 1992, Ahmad (1992) developed a new concept of identification of plastics by incorporating a fluorescent tracer into the polymers and giving them fluorescence signatures


83

in Ultra Violet spectroscopy. Recently, he also described a new technology for automatic sorting of plastics, based upon optical identification of fluorescence signatures of dyes, incorporated into such materials in trace concentrations (Ahmad, 2004). The conclusion of this research was that the speed and purity of sorting was limited by the mechanical singulation inadequacy in the conveyor system at high speeds.

The use of tracers is also found in the companies specializing in magnetic sorting. For example, Eriez has filed a patent (Mankosa and Luttrell, 2005) on a magnetic sorting process of polymers in which a magnetic substance is incorporated. The advantage of the magnetic detection is the lack of sensitivity regarding the additives contained in the polymers. However, the magnetic tracer system required amounts of tracers which are of the order of the percent and these quantities could engender homogenisation problems and affect the mechanical properties of the polymer.

The challenge of overcoming the inefficiency of existing sorting technologies associated with the identification of the black plastics has led to the development of new methods of identification of polymeric materials.

The principal objectives of this work are to prove the detection of tracers (rare earth oxides) added to polypropylene through the use of X-ray fluorescence spectrometry and to evaluate some of the parameters which can influence the detection, such as the concentration of tracers, the time of detection and the possible overlapping interferences among the tracers. Dispersion of such selected tracers in thermoplastic polymers, during the compounding process, will, therefore, allow automatic sorting and increase sorting selectivity of the plastic wastes during end-of-life recycling.

III.A.2. The identification of tracers

It turns out that the effectiveness of sorting and in particular the speed of identification of plastic wastes can be improved by the use of a tracer system giving a unique signature to each type and grade of polymer. In the detection system presented below, the excitation of the tracers will be achieved through the use of an X-ray source and detection by X-ray fluorescence spectrometry.

This work of identification technology was firstly proposed by Ahmad (Ahmad, 2000; Simmons et al., 1998), who used fluorescent tracers emitting in ultraviolet spectrometry. The identification system is based on the incorporation of very small quantities of one or more substances emitting in X-ray fluorescence – a ‘‘tracer” – into the plastics requiring identification. The concentration of the tracers must be in the range of 100–1000 ppm, in order not to influence the properties of the polymer matrix.

Due to the great diversity of the nature of plastics, including the types of polymers, as well as the grades, the additives and colorants, it is not practical to have a tracer for each potential variant. To overcome this, tracers can be used in a matrix, such that each combination corresponds to a type, grade or additive (Fig. 1). For example, theoretically, by using only 4 tracers, it is possible to identify 15 (24 - 1) variations of different plastics.

The identification is carried out by exciting the plastics using an X-rays source. This radiation is absorbed by the tracers, who are excited to higher atomic energy levels. Each


84

tracer emits a unique radiation in X-ray fluorescence, which depends on the atomic number of the element. The detection system, coupled with a data processing system, detects the emitted radiation and identifies the signature of the tracer, and thus the nature of the polymer matrix.

Fig. 1. Principle of optical identification by binary code generated from X-ray fluorescent tracers.

III.A.3. Experimental part

III.A.3.1. Materials

ISPLEN PP 050 G1E is a medium melt flow rate polypropylene homopolymer particularly formulated and adapted for injection moulding and extrusion applications. It is intended for applications that require good impact resistance balanced with high stiffness. The original pellets have a melt mass-flow rate of 5.8 g/10 min (2.16 kg at 230°C) and a density of 0.905 g/cm3. The rare earth oxides used as tracers are given in Table 1.

Table 1 Materials specifications.

Materials Chemical Name Supplier Mean particle diameter (µm)

Density (g/cm3)

Melting point (°C)

Toxicity (LD50

*) PP Polypropylene REPSOL 2000 0.905 167 - Y2O3 Yttrium Oxide RHODIA 2.250 5 2415 > 5000 mg/kg CeO2 Cerium Oxide RHODIA 2.250 7.3 1950 > 5000 mg/kg Nd2O3 Neodymium Oxide AMPERE 3.500 7.24 2272 > 1000 mg/kg Gd2O3 Gadolinium Oxide RHODIA 1.850 7.4 2340 > 1000 mg/kg Dy2O3 Dysprosium Oxide RHODIA 2.250 7.8 2340 > 1000 mg/kg Er2O3 Erbium Oxide AMPERE 8.200 8.6 2355 > 1000 mg/kg Yb2O3 Ytterbium Oxide AMPERE 1.000 9.2 2346 > 1000 mg/kg

* Median lethal dose (LD50) of a toxic substance is the dose required to kill half the members of a tested population


85

The rare earth oxides were chosen as tracers because they satisfy the required specifications (Hedrick, 1988; Maestro and Huguenin, 1995; Morteani, 1991). They are compatible, not abrasive and stable in the environment of evolution. They are also non toxic (Lauwerys et al., 2007; Peltier, 1986) during their application and use. Regarding the detection method, they have a high atomic number, facilitating their detection in X-ray fluorescence.

III.A.3.2. Dispersion of tracers All composites were prepared under identical mixing and moulding conditions.

Several samples of polypropylene containing the rare earth oxides in different concentrations were prepared for analysis (Table 2), using a twin screw extruder CLEXTRAL BC21 machine with a screw of L/D = 90 and D = 25 mm. The extrusion temperature and screw speed were 205 °C and 120 rpm respectively. The screw and temperature profiles used in this study are supplied in Fig. 2.

Table 2 Characteristics of the traced polypropylene samples. Code PP/tracer_ppm Tracer Mass (ppm)

Composition Melting

point (°C) PP PP-reference 0 166.7 PP/Nd2O3_1000 Nd2O3 1000 167.1 PP/Gd2O3_1000 Gd2O3 1000 166.1 PP/all_250 Y2O3, CeO2, Nd2O3,

Gd2O3, Dy2O3, Er2O3, Yb2O3

250 166.2 PP/all_1000 1000 166 PP/all_1450 1450 165.7

Fig. 2. Screw and temperature profiles for the extrusion processing. The pellets obtained from extrusion were then injected in a Battenfeld Unilog B2/350

Plus injection moulding machine. A general purpose screw was used in the barrel, which was kept at 220 °C. Specimens of 6 cm of diameter and 2 mm of thickness were fabricated, for the X-ray fluorescence analysis.

III.A.3.3. X-ray fluorescence device The X-ray fluorescence device, developed by NRT, is a test system which allows

taking static measurements of the samples spectrum. A schematic of a typical EDXRF


86

(energy dispersive X-ray fluorescence) spectrometer is shown in Fig. 3. The X-ray radiation from the X-ray source passes through a hole onto the sample and then travels to the detector.

Fig. 3. Schematic of the EDXRF device. X-rays from the tungsten tube pass through a hole on their way to the

sample. X-rays from the sample then travel to the ‘‘2000 impulsions per second” detector. The signal at the detector is then processed by electronics and sent to a computer for analysing the spectra.

The signal from the detector is then processed electronically and sent to the computer,

which also controls the X-ray source. The XRF spectra can then be analyzed and used for the separation of the samples containing the tracers.

This EDXRF technique (Havrilla, 1997) is a two-step process that begins with the removal of an inner shell electron of an atom. The resulting vacancy is filled by an outer shell electron. The second step is the transition from the outer shell electron orbital to an inner shell electron orbital. The transition is accompanied by the emission of X-ray photon. The fluorescent photon is characteristic of the element and is equal to the difference in energy between the two electron energy levels. Because the energy difference is always the same for given energy levels, the element can be identified by measuring the energy of the emitted photon. In turn, the intensity of the emitted photons determines the concentration of the element. Therefore the measure of the photon energy provides the identification of the element and the intensity of the photon emission provides a measure of the amount of the element.

The emission process is similar to other fluorescent measurement techniques, but it is restricted to the X-ray region of the electromagnetic spectrum that ranges from 4 to over 80 keV. The photon energies detected are designated as K, L, or M X-rays, depending on the energy level being filled. For example, NdKb1 represents, for the neodymium element, the transition corresponding to the passage of M to K level and 1, the relative intensity of the transition in the series (1, more intense than 2). There are as many possible X-ray lines as there are inner shell electrons. However, the most analytically useful and most intense X-ray lines are the K shell electrons. Although the multitude of emitted X-ray lines could result in complex spectra, the relative low intensities of the lines below the L level allow obtaining a clear spectrum with minimal interference.

The volume of the analyzed sample was 1 mm3. The distance between the sample and the source was 10 cm and the angle 45°. The detection time varied from 1 to 4 min and the detector could make 2000 counts per second. The X-ray source used was a tungsten tube.

X-ray Source

Sample

Detector Electronics

Computerd

=1

0cm

W tube 2000 imp./s

V=1mm3

X-ray Source

Sample

Detector Electronics

Computerd

=1

0cm

W tube 2000 imp./s

V=1mm3


87

III.A.4. Results and discussions

III.A.4.1. Detection of tracers and concentration effect

The first objective of these tests was to prove the detection of the 7 tracers in the

polypropylene matrix. Table 3 shows the expected energy of the element composing the tracers. The detection of tracers is separated in two domains. The first one, between 7–20 keV (Lα1, Lβ1 of ytterbium and Kα1, Kβ1 of yttrium) and the second between 34–60 keV (Kα1, Kβ1 of cerium to ytterbium).

Table 3 X-ray fluorescent energies of the elements (Peltier, 1986).

Energy lines (keV)

Y2O3 CeO2 Nd2O3 Gd2O3 Dy2O3 Er2O3 Yb2O3 Source Y Ce Nd Gd Dy Er Yb W

Kα1 14.96 34.71 37.36 42.99 45.99 49.12 52.38 59.31 Kβ1 16.74 39.26 42.27 48.72 52.18 55.69 59.35 67.23 Lα1 1.92 4.84 5.23 6.06 6.49 6.95 7.41 8.40 Lβ1 1.99 5.26 5.72 6.71 7.25 7.81 8.40 9.67

The first samples tested, contained all the tracers (Y2O3, CeO2, Nd2O3, Gd2O3, Dy2O3,

Er2O3 and Yb2O3) in three different concentrations (PP/all_250, PP/all_1000 and PP/all_1450). The exposure time for the four samples was 1 min. The obtained spectrum is shown in Fig. 4.

Fig. 4. Comparison of XRF spectra of PP/reference, PP/all_250, PP/all_1000 and PP/all_1450 in order to prove

the detection of tracers and their effect on the concentration.

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70

Energy (keV)

Inte

nsity

Kα Y

Kβ Y

Kα CeKα Nd

Kα GdKα Dy

Kα ErKα Yb

PP/all_250

PP/Reference

PP/all_1450

PP/all_1000

Lα, Lβ-tarcers


88

As shown in the spectrum, five of the tracers (Y2O3, CeO2, Nd2O3, Gd2O3 and Dy2O3), at concentration levels of 1000 and 1450 ppm, are clearly visible and distinguishable from the ‘‘background” sample of 0 ppm by their Kα1 energy line.

For the same five tracers at 250 ppm, the fluorescent peak is not sufficiently different from the background in order to make a positive identification. Concerning the detection of Er2O3 and Yb2O3, their Kα1energy lines (49.12 and 52.38 keV respectively) are emitted in the same range of energy [49–53 keV] as the tungsten source (W Kα1: 59.31 keV), rescheduled of about 8–9 keV. By using another source, gold for example, this problem can probably be avoided. Regarding the detection of the Lα1 and Lβ1 energy lines of CeO2, Nd2O3, Gd2O3, Dy2O3, Er2O3 and Yb2O3 a large fluorescent peak appears in the range [5–9 keV] corresponding to these energies.Consequently it is difficult to distinguish the Lα1 and Lβ1 energy lines of these elements.

Moreover, the detection of the Lα1 and Lβ1 energy lines present three additional difficulties:

– They have significantly less fluorescent yield than the K energy lines. – They are much closer in energy to other elements which might be present (example

cerium versus titanium) and thus require a combination of longer exposure times and more sensitive detectors.

– The lower energy of L X-rays are more easily absorbed by air, plastic dust coverings, thin metal foils, that are present in the construction of an ‘‘industrial” sorting environment.

III.A.4.2. The overlapping and mass absorption effect The second objective of these tests was to assess the probable overlapping and mass

absorption of energy lines of different tracers. Four different spectra corresponding to polypropylene with all the tracers, with gadolinium oxide and with neodymium oxide at 1000 ppm and the polypropylene reference are presented in Fig. 5.

The exposure time was 1 min. The samples of neodymium oxide and gadolinium oxide were chosen for the reason that the Kβ1 energy line of neodymium (42.27 keV) is close to the Kα1 energy line of gadolinium (42.99 keV) and this small energy difference could lead to overlapping interferences.

In the range [41.5–43.5 keV] a large fluorescent peak is observed. By taking in consideration that the Kb1 energy line of neodymium is 42.27 keV and the Kα1 energy line of gadolinium is 42.99 keV, it is difficult to distinguish which element is contained in the polymer matrix. The overlapping is caused by the small energy difference between Kβ1 energy line of neodymium and the Kα1 line of gadolinium.

This is a problem associated with the detector and which can be resolved by either of the following two options. The first one is to use a detector which has a better resolution than that used for these trials. The other solution may be the simultaneous detection of Kα1 and Kβ1 energy lines of gadolinium in order to identify it.


89

Fig. 5. Comparison of XRF spectra of PP/reference, PP/Nd2O3_1000, PP/Gd2O3_1000 and PP/all_1000 so as to

assess the probable overlapping and mass absorption of energy lines between different tracers.

Regarding the mass absorption effect, in the energy range [36.5–38 keV], corresponding to the Ka1 energy line of neodymium, the surface of the fluorescent peak is larger in the case of neodymium oxide at 1000 ppm alone than the peak of polypropylene containing all the tracers. This means that the presence of several rare earth oxides at the same time could influence their intensity. This factor should be studied in more details in order to have a better understanding and take it in consideration for the detection, if necessary.

III.A.4.3. The effect of the acquisition time During these tests, the effect of the acquisition time in the detection of tracers was also

studied. The fluorescent emissions of a sample with 250 ppm of all tracers exposed for 1 and 4 min were compared to 1000 ppm sample exposed for 1 min (Fig. 6).

The main aim was to look at the difference between a sample exposed for 1 min and 1 for 4 min, in order to study the effect of the acquisition time. Secondly, we wanted to observe if there is a linear evolution between the detection time and the concentration of the tracer. For that reason, the graph containing the sample of all the tracers at 250 ppm exposed for 4 min is compared to the sample of all tracers at 1000 ppm exposed for 1 min.

By comparing the graphs of the samples containing all the tracers at 250 ppm at 1 and 4 min exposure time, it appears that for the sample exposed for 4 min, five of the tracers (Y2O3, CeO2, Nd2O3, Gd2O3 and Dy2O3) are clearly visible and distinguishable from the ‘‘background” sample of 0 ppm by their Kα1 energy line compared to the sample exposed for 1 min, in which only three tracers (Y2O3, CeO2 and Nd2O3) are distinguishable.

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70

Energy (keV)

Inte

nsity

Kα NdKα Gd

PP/Nd2O3_1000

PP/Reference

PP/all_1000

PP/Gd2O3_1000


90

Fig. 6. Comparison of XRF spectra of PP/reference exposed for 1 min, PP/all_250 exposed for 1 min,

PP/all_1000 exposed for 1 min and PP/all_250 exposed for 4 min, with the aim of studying the effect of the acquisition time.

For the samples containing all the tracers at 250 ppm for 4 min and at 1000 ppm for 1

min, in both case five of the tracers (Y2O3, CeO2, Nd2O3, Gd2O3 and Dy2O3) are clearly visible and distinguishable from the background by their Kα1 energy line. The difference between them is that in the case of the sample containing all the tracers at 1000 ppm for 1 min, the surface of the fluorescent peak is superior to the one of the sample of all the tracers at 250 ppm for 4 min. The advantage of this last one is that the Kβ1 energy line of neodymium and the Kα1 of gadolinium are more easily distinguished between them compared to the sample containing all the tracers at 1000 ppm for 1 min exposition time.

The conclusion reached from these tests is that by increasing the acquisition time, the fluorescent peak is more distinguishable from the background and therefore the level of accuracy in identification is increased.

III.A.5. Conclusion and further research The majority of post-consumer plastic wastes is sent to landfill sites for disposal. The

automotive and electrical industries are currently the worst performers, concerning the recycling of plastics, partly due to the complexity of the waste material that these sectors produce. The existing automatic sorting techniques are not applicable to dark plastics and cannot provide sorted plastics into individual types and grades.

The originality of this work is to propose an X-ray fluorescence detection system based on the use of tracers, for the identification of plastics. The use of a tracer system offers many benefits in the sorting of plastics. The tracer system can provide specific/positive

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70

Energy (keV)

Inte

nsity

Kα Y

Kβ Y

Kα CeKα Nd

Kα GdKα Dy

Kα ErKα Yb

PP/all_250 for 1min

PP/Reference

PP/all_250 for 4min

PP/all_1000 for 1min


91

identification versatility and increase the efficiency of sorting and purity of the sorted fractions. It can also provide high speed identification of plastic material and solve the problem of identifying dark plastics.

This work shows that by using the XRF test system device, it was possible to detect 5 of the 7 tracers tested at the concentration level of 1000 ppm. The two parameters, concentration level and acquisition time, will play an important role in the development of an industrial device. Some improvements have to be made in order to reduce the acquisition time and the concentration level of the tracers. For this reason further work is under progress with the objective of developing a pilot plant devoted to the optimization of these two parameters.

Acknowledgements

The authors would like to thank National Recovery Technologies Inc. for helping us to carry out the tests in their X-ray fluorescence device, the French Environment and Energy Management Agency (ADEME – www.ademe.fr) and the French Industry-University Cooperative Research Network on Waste (RECORD – www. record-net.org) for their contribution to the funding of this work and for providing industrial orientations and scientific supervision to the research.


92

References

Ahmad, S.R., 2004. A new technology for automatic identification and sorting of plastics for recycling. Environ. Technol. 25 (10), 1143–1149. Ahmad, S.R., 2000. Marking of products with fluorescent tracers in binary combinations for automatic identification and sorting. Assem. Autom. 20 (1), 58–64. Ahmad, S.R., 1992. Partners wanted for polymer research project. Mater. Recycl. Weekly (UK) 14. Alam, M.K., Stanton, S.L., Hebner, G.A., 1994. Near-infrared spectroscopy and neural networks for resin identification. Spectroscopy 9 (2), 30–40. European Parliament, 2003. Directive 2002/96/EC of the European parliament and of the council of 27 January 2003 on waste electrical and electronic equipment. European Parliament, 2000. Directive 2000/53/EC of the European parliament and of the council of 18 September 2000 on end-of-life vehicles. Florestan, J., Lachambre, A., Mermilliod, N., Boulou, J.C., Marfisi, C., 1994. Recycling of plastics: automatic identification of polymers by spectroscopic methods. Resour., Conserv. Recycl. 10 (1–2), 67–74. Froelich, D., Maris, E., Haoues, N., Chemineau, L., Renard, H., Abraham, F., Lassartesses, R., 2007. State of the art of plastic sorting and recycling: feedback to vehicle design. Miner. Eng. 20 (9 SPEC. ISS.), 902–912. Havrilla, G.J., 1997. Handbook of Instrumental Techniques for Analytical Chemistry, pp. 459–479. Hedrick, J.B., 1988. Availability of rare earths. Am. Ceram. Soc. Bull. 67 (5), 858–861. Kenny, G.R., Bruner, R.S., 1994. Experience and advances in automated separation of plastics for recycling. J. Vinyl Addit. Technol. 16 (3), 181–186. Lauwerys, R.R., Haufroid, Vés, Huet, Pés, Lison, Dés. impr. 2007. Toxicologie industrielle et intoxications professionnelles [texte imprimé]. Issy-les- Moulineaux: Elsevier, Masson. Maestro, P., Huguenin, D., 1995. Industrial applications of rare earths: which way for the end of the century? J. Alloys Compd. 225, 520–528. Mankosa, M.J., Luttrell, G.H., 2005 inventors; Plastic material having enhanced magnetic susceptibility, method of making and method of separating. Moore, S., 1999. Infrared scanner affords easy plastic identification. Mod. Plast. (NY) 76, 32–33. Morteani, G., 1991. The rare earths: their minerals, production and technical use. Eur. J. Mineral. 3 (4), 641–650. Olofsson, M., Sundberg, J., Sahlin, J., 2005. Evaluating waste incineration as treatment and energy recovery method from an environmental point of view. In: 13th Annual North American Waste to Energy Conference, NAWTEC13; 23 May 2005 through 25 May 2005; Orlando, FL, 175 p. Patel, M.K., Jochem, E., Radgen, P., Worrell, E., 1998. Plastics streams in Germany – an analysis of production consumption and waste generation. Resour. Conserv.


93

Recycl. 24 (3–4), 191–215. Peltier, A., 1986. Exposition aux poussières de terres rares. Report nr. 122. Simmons, B.A., Overton, B.W., Viriot, M., Ahmad, S.R., Squires, D.K., Lambert, C., 1998. Fluorescent tracers enable automatic identification and sorting of waste plastics. Br. Plast. Rubber 4, 6, 8. Zoboli, R., 1998. Implications of environmental regulation on industrial innovation: The case of end-of-life vehicles. Implications of Environmental Regulation on Industrial Innovation: The Case of End-of-Life Vehicles.


94

Part B: Comparison of X-ray detectors for the optimisation of high speed detection of rare earth

oxides, used as tracers into a polymer matrix

III.B.1. Introduction

In 2000, the European Commission has decided to apply a real policy of raw resource management by setting up a directive (European Parliament, 2000) dealing with the end of life of post consumer products coming from end-of-life vehicles (ELVs). The European Directive 2000/53/CE has established as objective that from January 1st 2015, the reuse and recovery rate shall be increased to a minimum of 95% of average weight per vehicle and year.

In the current treatment methods of ELVs, the metals, representing 75% by mass per vehicle are completely recycled whereas the remaining materials, commonly called automotive shredder residue (ASR), representing 25% by mass per vehicle, are in general disposed to landfills or incinerated. The ASR consists of about 50% of plastics (Rausa and Pollesel, 1997), and it has been estimated that only 3% to 4% of these plastic materials are recycled (Froelich et al, 2007). The recycling problem is directly related to the complexity of current end-of-life products and available sorting techniques. Today, it is impossible to discriminate different families of black plastics, the main polymer materials in car parts, with the existing optical sorting technologies. Therefore, the development of advanced sorting techniques which could be used to recover the polymer materials from ASR in order to achieve 95% recovery rate per vehicle and year, imposed by the European Commission, has become an important issue.

In previous works [Bezati et al, 2010a, 2010b], the authors have already reported a high-speed automatic sorting system based on the detection by X-ray fluorescence spectrometry of tracers dispersed into the polymer materials during the compounding process. The use of tracers, rare earth oxides, increases the sorting selectivity of the plastic wastes during end-of-life recycling by providing specific and positive identification versatility as well as a possible codification of the plastic materials.

For an economic and efficient sorting of plastic wastes, the tracer concentration must be lower than 1000 ppm for an acquisition time of about 10 ms. Based on these restrictions, the main objective of this research work is the estimation of their detection limit by experimental and modeling results. The detection limit depends on the tracer concentration, the acquisition time, as well as on the detection efficiency of the measurement system. Since the first two parameters are fixed, thus the main factor which will influence the detection limit will be the quality of detection, which depends mainly on the detector used. In this paper, three different semiconductor detectors, a SDD, a GeHP and a CdTe, were experimentally studied in order to compare them. Commercially, there are detectors which can be more adapted for the high speed identification of rare earth elements. The detection limit of these commercial detectors was estimated through modeling results obtained by using MACALU, software based on Monte-Carlo methods developed by CEA. Except the detectors, secondary parameters such as the X-ray filter material and thickness, the voltage of the X-ray generator


95

as well as the probable overlapping interferences and mass absorption effect of tracers may also play an important role in the detection limit. The modeling study of these parameters is also reported in this paper.

III.B.2. Theory

Some general physical properties and spectroscopic features of the 3 semiconductor detectors are given in Table 1 (Owens and Peacock, 2004). From an electronic point of view, the semiconductor material must be chosen regarding the following parameters:

- for an intense measured signal, the pair creation energy must be the lowest

possible, which means that the band gap should be low. - the noise coming from the leakage current should be also low, which means that

semiconductor resistivity must be high. Material resistivity depends on carrier mobility and band gap width.

- an important parameter which influences the charge collection, is the carrier mobility and their lifetime. The best collection is obtained for a high lifetime and charge mobility, which means that the µτ product must be maximal.

By comparing the results in Table 1, it can be concluded that the Ge semiconductor

has the best performance regarding the band gap (significantly smaller than the other semiconductors), the carrier mobility and lifetime as well as for the material resistivity. The Ge detector is followed by the SDD and then the CdTe which has the lowest performance.

Table 1 Compilation of the general properties and spectroscopic features of semiconductors, grouped according to the average atomic number (Gros d’Aillon, 2005; Sakai, 1982)

Detectors Average atomic number

Density (g.cm-3)

Resistivity (Ohm.cm-1)

Band gap (eV)

Pair creation energy

(eV)

Mobility (cm².V-1.s)

Lifetime (µs)

electron hole Electron hole Si (SDD) 14 2.33 <106 1.12 3.62 1400 1900 >1000 1000 Ge 32 5.33 50 0.67 2.96 3900 1900 >1000 2000 CdTe 50 5.85 109 1.44 4.43 1100 100 3 2

Regarding the high speed detection of rare earth oxides, the detector must be chosen

by taking into account additional parameters such as the following ones: - a high efficiency in the domain of the energetic lines of tracers, for having a high

net counting rate - a high resolution, for reducing the background noise


96

Table 2 Advantages and drawbacks of SDD, Ge and CdTe detectors (Sato et al, 2002; Takashi et al, 2001, 2002; Gatti and Rehak, 2005; Castoldi et al, 1999. Morris et al.) Detectors Advantages Drawbacks

SDD

- excellent energy resolution - Peltier cooling - detector saturation : 1 000 000 cps - performed technology fabrication

- low efficiency

HPGe

- excellent energy resolution - high efficiency - detector saturation : 1 000 000 cps - performed technology fabrication

- very expensive - complicated cooling system - fragile

CdTe - high efficiency - Peltier cooling - low cost

- medium energy resolution - detector saturation : 200 000 cps - high noise in the range of [30-50 keV] due to the escaping peaks - low hole mobility

By comparing from Table 2, the three detectors with respect to their advantages and

drawbacks, the HPGe seems to be the most adapted for high speed identification of rare earth oxides. Its high efficiency in the range [10-60 keV] as well as its high resolution and maximum saturation counting rate make it the best candidate for our project. The CdTe has the same efficiency as the HPGe detector in the range [10-60 keV], but its medium energy resolution and high noise in the range [30-50 keV], will probably be a problem for detecting the rare earth oxides in low concentrations. The SDD has an excellent energy resolution and a high maximum saturation counting rate, but its low efficiency will possibly influence the detection of high atomic number elements composing the tracers.

III.B.3. Methods and materials

III.B.3.1. Characteristics of detectors and tracers The characteristics of the detectors (SDD, HPGe and CdTe) reported in this paper are

given in Table 3. By comparing the three detectors, SDD has the best energy resolution, based on the low thickness and Peltier cooling, whereas the CdTe and HPGe detectors have the best efficiency, which is explained by their high stopping power provided by their elevated density and average atomic number.

Table 4 provides the tracers specifications. As XRF is a volume, non-destructive spectroscopic method enabling elemental analysis of materials; hence the choice of tracers is directly relied on the elements of Mendeleyev’s Periodic Table. Once from Mendeleyev’s Periodic Table are eliminated the toxic and radioactive elements, in addition to elements presented in polymer additives, low atomic number elements; which do not have an intense


97

signal in XRF and elements which do not have important reserves the selection leads to the choice of rare earth elements. For these last ones we have chosen the most stable chemical molecules, the oxides (Bezati, 2010b).

Table 3 Characteristics of SDD, Ge and CdTe detectors

Detectors_ thickness

Area (mm²)

Thickness (mm)

Window (µm)

FWHM [14 - 60 keV] (eV)

Photoelectric efficiency (%) 15 keV 30 keV 40 keV 50 keV

SDD_0.3 10 0.3 Be (8) [230 – 370] 52.4 9.9 4.9 3.1 HPGe_10 200 10 Al (50) [700 – 800] 99.4 99.7 99.7 99.7 CdTe_1 25 1 Be (100) [1000 – 1300] 99.4 99.7 99.7 99.5

Table 4 Tracers specifications

Tracers Y2O3 CeO2 Nd2O3 Gd2O3 Dy2O3 Er2O3 Yb2O3 Chemical Name

Yttrium Oxide

Cerium Oxide

Neodymium Oxide

Gadolinium Oxide

Dysprosium Oxide

Erbium Oxide

Ytterbium Oxide

Supplier Rhodia Rhodia Ampere Rhodia Rhodia Ampere Ampere

III.B.3.2. Energy dispersive X-ray fluorescence device Fig. 1 shows the experimental device composed of an X-ray generator produced by

CEA, a SDD detector of 10 mm² active area, 0.3 mm thickness and a copper (Cu) filter of 1 mm thickness.

Fig.1. The X-ray fluorescence device established at LIST

SDD detector with Peltier cooling

X-ray Generator

Trap of Pb used for photons capture

Sample holder

Cu filter of 1 mm

SDD detector with Peltier cooling

X-ray Generator

Trap of Pb used for photons capture

Sample holder

Cu filter of 1 mm


98

The angle between the X-ray generator and the detector is 53°, the distance between the X-ray generator and the sample is 120 mm and the filter is placed at 80 mm from the generator. The distance between the detector and the sample is 120 mm and the detector shielding is composed of lead (Pb). The sample contains 1000 ppm of each tracer; its thickness is 2 mm and its diameter 60 mm. The length of the sample viewed by the detector is 20 mm, which also corresponds to an average length of shredder residue. The same configuration was used for the HPGe and CdTe detectors. Table 5 summarizes the specifications of each of them. Spectrum analysis was carried out using Interwinner software developed by Camberra.

Table 5 Detection device specifications

Detectors

Copper filter thickness (mm)

X-ray generator voltage (kV)

X-ray generator courant (mA)

Detector shielding Diameter (mm) Length (mm)

SDD 1 150 18 14 50 HPGe 2 150 20 10 50 CdTe 1 150 9 14 50

III.B.3.3. Software overview MACALU, developed by F. Tola (Tola 1994, 1995), is a software using Monte-Carlo

methods, comprising a knowledge base as well as a physical and nuclear database. This tool is a mean of assistance for designing radiogauges (thickness, density …) based on the attenuation or diffusion of γ or X-rays by radioactive source or X-ray generator between 1 keV and 10 MeV. It provides the optimization of different components of a transmission gauge (detector, radioactive source, collimators) by taking into account parameters and constraints linked to a given configuration (nature and composition of materials, presence of shields and walls) as well as users requirements (counting time, beam collimation, active surface of detector, etc.).

Fig. 2 shows the geometric setting, composed of 3 sections: - the source, assumed to be punctual and its collimator, is considered to be perfect.

The system only takes into consideration the photons emitted in the solid angle defined by the collimator.

- the walls, considered to be infinite and homogenous. The thickness and the composition are given by the user.

- the detector bloc, composed of a cylindrical detection volume for which the composition and dimensions can be set by the user, and of a collimator which can be used as a shielding.


99

Fig. 2. Set up of the detection device

The source and the detector can be placed on the same side or on opposite sides of the

wall. Their angles can be set by the user and their position is defined with respect to their distance from the focal point F. The configuration can be graphically visual and modified after the data input.

The software provides the following results:

- the energy spectrum of the incident photons - the energy spectrum of the detected photons - the total counting rate - the counting rate for a given energy range

III.B.3.4. Calculation of detection limit Table 6 gives the energies and relative intensities of the Kα2, Kα1, Kβ1, and Kβ2 lines

expected for the elements composing the tracers, the detectors and the anode of X-ray generator. Intensities are normalized to 100% for the Kα1 energetic line.

The energies between the couples "Kα1 and Kα2" and "Kβ1 and Kβ2" are very close, thus

not allowing their individual identification with the energetic resolution supplied by the commercial detectors. For this reason, the elements will be identified by measuring the entire counting rate between the two energetic lines Kα1 and Kα2. Kα1 energetic line is two times more intense than Kα2 line and about three times more intense than Kβ1 line.

Source

Detector bloc Walls

F

Crystal

Shielding

The flow path of a photon


100

Table 6 Kα1, Kβ1, Kα2 and Kβ2 energetic lines and their intensities for the elements composing the tracers, the anode of X-ray generator and the detectors

Lines Kα1 Kα2 Kβ1 Kβ2

Energy (keV)

Intensity (%)

Energy (keV)

Intensity (%)

Energy (keV)

Intensity (%)

Energy (keV)

Intensity (%)

Tracers

Y 14.96 100 14.88 50 16.7 22 17 3.1 Ce 34.72 100 34.28 52 39.2 29 40.2 6 Nd 37.36 100 36.85 52 42.2 29 43.3 6 Gd 43.00 100 42.31 53 48.7 30 50.0 7 Dy 46.00 100 45.21 53 52.1 31 53.5 7 Er 49.13 100 48.22 53 55.6 32 57.2 7 Yb 52.39 100 51.35 54 59.3 32 61.0 7

Anode W 59.32 100 57.98 54 67.2 33 69.1 8

Detectors

Si 1.74 100 1.74 100 1.84 100 - - Ge 9.89 100 9.86 50 11.0 19 11.1 0.6 Cd 23.17 100 22.98 51 26.1 26 26.6 5 Te 27.47 100 27.2 51 31.0 27 31.7 6 As mentioned above, MACALU or Interwin provides the energy spectrum of the

detected photons and the total counting rate (Rtotal). Fig. 3 shows the way the raw counting rate (Rraw) and the background counting rate (R0) are calculated. The difference between them is defined as the net counting rate (Rnet = Rraw – R0).

Fig. 3. The energy spectrum of background and a sample containing the Y2O3, Nd2O3 and Dy2O3, showing the

way the raw and background counting rate are calculated In our study, a Poisson distribution is applied for the calculation of the detection limit

by considering the following hypothesis: - the expected counts N are in the range [10-500], - the risk α and β are taken equal to 2.5%,

Compton scattering

R0

Rraw W

Pb


101

- the net counting rate of the fluorescent peak of tracers follows a linear relation for low concentrations,

- the tracer does not undergo an auto-absorption. Therefore, the detection limit can be assimilated to the following formula by adjusted the Poisson Tables to a power function (Neuilly, 1993):

454.00 )(6.7 tR

tRCDL

net

=

Where C is the tracer concentration in the polymer matrix, given in ppm and t the

acquisition time. Consequently, the unit of the detection limit (DL) is also in ppm. Equation 1 shows that the detection limit depends on the tracer concentration, the acquisition time as well as the net and background counting rate. For these four parameters influencing the detection limit, the acquisition time must be in the range [1-10 ms] in order to obtain a high speed automatic identification and the tracer concentration must be lower than 1000 ppm for having an economically efficient sorting of plastic wastes without affecting the polymer properties (Bezati, 2010b). As the range of acquisition time and of tracer concentration is imposed, the only parameters which can reduce the detection limit are the net counting rate and the background counting rate, which depend on the detector used. The aim of this work is to choose the proper parameters for the detection system in order to obtain the lowest background counting rate and the highest net counting rate for reducing the detection limit.

III.B.4. Results and Discussions

III.B.4.1. Comparison of the detectors with respect to experimental results

Fig. 4 shows spectra from the background and the sample containing the tracers

obtained from SDD_0.3, HPGe_10 and CdTe_1 detectors, whereas Fig. 5 shows the data of the net and background counting rate from these three detectors. The data were obtained for an angle between the X-ray generator and the detector of 53°, a distance between the X-ray generator and the sample of 120 mm and a distance between the detector and the sample of 120 mm. The copper filter thickness was 1 mm for the SDD_0.3 and the CdTe_1 detectors and 2 mm for the HPGe_10.

From both figures, it is clear that the HPGe_10 and the CdTe_1 detectors have excellent efficiency and sensitivity in the range [30-60 keV]. Their noise from the Compton scattering is much more important than the one of the SDD_0.3 detector. This last one has considerably better energy resolution and lower Compton scattering than HPGe_10 and CdTe_1, but its photopeak efficiency in the range [40-60 keV] is much lower.


102

Fig. 4. Comparison of energy spectrum with SDD_0.3, HPGe_10 and CdTe_1

Fig. 5. Net and background counting rate experimental results with SDD_0.3, HPGe_10 and CdTe_1

Fig. 6 compares the detection limit calculated for the three detectors from the data of

Fig. 5. As expected, the best results are obtained for the HPGe_10 and the CdTe_1. For these two detectors the experimental results are similar and the value of the detection limit is in the range [1000-2000 ppm] for both of them for all the tracers tested, except the Y2O3 in the case of HPGe_10. As can be seen in Fig. 5, the net counting rate of Y2O3 for the HPGe_10 is negligible compared to the background counting rate. The aluminium filter in front of the HPGe_10 detector prevents the detection of low energies such as those of yttrium, hence the identification and the calculation of the detection limit of yttrium oxide is impossible.

For all the tracers tested, the experimental results obtained by the SDD_0.3 detector are worse than the one of HPGe_10 and CdTe_1. However, the detection limit of Y2O3, CeO2 and Nd2O3 is lower than 2000 ppm and, by using a thicker detector, the value of the detection limit of these three tracers will probably decrease so as to be lower than 1000 ppm. For the

0

1000

2000

3000

4000

5000

6000

7000

8000

0

1000

2000

3000

4000

5000

6000

7000

8000

Y2O3 CeO2 Nd2O3 Gd2O3 Dy2O3 Er2O3 Yb2O3

Back

grou

nd c

ount

ing

rate

R0

(cps

)

Net

coun

ting

rate

Rne

t(cp

s)

Tracers

HPGe_R0

SDD_R0

CdTe_RnetCdTe_R0

HPGe_Rnet

SDD_Rnet


103

remaining tracers, the use of a SDD detector for high speed identification does not seem to be an appropriate choice.

Fig. 6. Detection limit experimental results of SDD_0.3, HPGe_10 and CdTe_1

III.B.4.2. Modeling results of the commercial detectors more adapted for the high speed detection of rare earth oxides

III.B.4.2.1. Detectors specifications

The three detectors reported previously are not completely adapted for the high speed identification of rare earth oxides. The SDD_0.3 has a thin thickness and a low active area which prevents it from a high efficiency for the high atomic number elements. The HPGe_10 detector possesses an aluminum window which blocks the yttrium oxide detection whereas its low energy resolution compared to current commercial detectors prevents us to estimate its real performance. For the CdTe_1 detector, we have considered that the thickness reduction could have a positive impact in energy resolution as well as in Compton scattering and escape peaks. In order to achieve a homogenous comparison of detectors’ behavior, we have considered that they have the same active area and window. Their thickness and energy resolution (FWHM) were chosen according to actual commercial detectors. The characteristics of the commercial detectors considered for the modeling study are given in Table 7.

As can be seen in Fig. 7 and Table 7, the thickness of the SDD detector was increased in order to improve its efficiency and thus its selectivity for the high atomic number tracers. On the contrary the thickness of the HPGe and CdTe detectors was decreased to reduce the Compton scattering.

For these three detectors we have studied some of the parameters which can influence the detection limit, such as the filter thickness, the X-ray generator voltage and the possible overlapping interferences and mass absorption effect among the tracers.

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000


Det

ectio

n lim

it fo

r 10

ms (

ppm

)

Tracers

HPGe

SDD

CdTe


104

Table 7 Characteristics of the commercial detectors more adapted for the high speed detection of rare earth elements

Detectors Area

(mm²) Thickness

(mm) Window

(µm) FWHM at 40 keV

(eV) SDD_0.5 100 0.5 Be (100) 200 HPGe_5 100 5 Be (100) 200 CdTe_0.5 100 0.5 Be (100) 400

Fig. 7. Log-log plot of photoelectric efficiency of SDD, Ge and CdTe detectors

III.B.4.2.2. The detector and filter thickness effect The two major parameters influencing the net and background counting rate and thus

the detection limit are the detector and the filter thickness. The detector’s efficiency will influence the net counting rate whereas the filter thickness will play an important role on the Compton scattering reduction and on the optimization of the ratio between the background and net counting rate. Fig. 8 shows the influence of the detectors and filter thickness on the detection limit.

By comparing the modeling results obtained for the three detectors given in Fig. 8, the following conclusions are established. First, the HPGe detector gives the best results for the high speed detection of rare earth oxides (except the case of Y2O3), with the detection limit shifting from 30 ppm for the CeO2 to 350 ppm for the Yb2O3 for an acquisition time of 10 ms.

Second, for the SDD and CdTe detectors, the tracers are separated in two groups. The SDD gives the best results for the Y2O3, CeO2, Nd2O3 and Gd2O3, whereas the CdTe performs better for the Dy2O3, Er2O3 and Yb2O3.

Finally, regarding the filter thickness, the SDD behaves in a different way than the HPGe and CdTe. Considering all the tracers, the best compromise for the thickness, for the SDD is 1 mm, whereas for the HPGe and CdTe the best results are obtained for 2 mm. The 2 mm filter thickness is the maximum value accepted in order to have a sufficient counting rate while applying the Poisson distribution.

1

10

100

1 10 100

Phot

oele

ctri

c Eff

icie

ncy

(%)

Energy (keV)

Be window of 100 µm thickness

SDD_0.3

HPGe_10

CdTe_0.5

HPGe_5

SDD_0.5

CdTe_1


105

Fig. 8. Detection limit as a function of tracers, detectors and filter thickness for 10 ms

0

100

200

300

400

500

600

700

0 0.5 1 1.5 2

Det

ectio

n lim

it (p

pm/1

0ms)

Filter thickness (mm)

Tracer : Y2O3

SDD_Y

HPGe_Y

CdTe_Y

0100200300400500600700800900

1000

0 0.5 1 1.5 2

Det

ectio

n lim

it (p

pm/1

0ms)


Tracers : CeO2, Nd2O3

SDD_Ce

HPGe_Ce

CdTe_Ce

CdTe_Nd

SDD_Nd

HPGe_Nd

0

200

400

600

800

1000

1200

1400

0 0.5 1 1.5 2

Det

ectio

n lim

it (p

pm/1

0ms)


Tracers : Gd2O3, Dy2O3

SDD_Gd

HPGe_Gd

CdTe_Gd

CdTe_Dy

SDD_Dy

HPGe_Dy

0

500

1000

1500

2000

2500

3000

3500

4000

0 0.5 1 1.5 2

Det

ectio

n lim

it (p

pm/1

0ms)


Tracers : Er2O3, Yb2O3

SDD_Er

HPGe_Er

CdTe_Er

CdTe_Yb

SDD_Yb

HPGe_Yb


106

Fig. 9 gives the energy spectrum of the background and of a sample containing all the tracers for the SDD detector with 1 mm copper filter thickness and the HPGe and CdTe with a filter thickness of 2 mm. For the 7 tracers tested, the HPGe detector gives the best results for 6 of them; especially for the tracers of high atomic number elements which come out at high energies. This performance can be explained by its elevated efficiency in the range [30-60 keV] as well as its high resolution and maximum saturation counting rate. The high efficiency as well as the high maximum counting rate contribute to an elevated net counting rate, whereas the high resolution permits to reduce the background.

Fig. 9. Comparison of energy spectrum of the three detectors for background and samples

The SDD detector is well known for its performances at low energy and this is the

reason why it gives the best results for the Y2O3. Furthermore, its high energy resolution and maximum saturation counting rate as well as its low cross-section for Compton scattering from collimators, samples and other materials compared to the CdTe, allowed the achievement of detection limits similar to the HPGe for the CeO2 and Nd2O3. However, its sensitivity falls drastically from energies above 40 keV with a yield of around 5% by reducing the peak signal of Gd2O3, Dy2O3, Er2O3 and Yb2O3 as can be seen in the Fig. 7. Consequently, the detection limit of these tracers is 2 to 3 times higher than that of the HPGe detector. The CdTe detector is constituted of higher atomic number elements than the SDD (Si) and HPGe, and it has a density of 5.85 g/cm3, which means a high stopping power and high photoelectric probability of interaction (Fig. 7) for the K lines of the tracers. However, its low resolution and low maximum saturation counting rate limit its performances.

0 10 20 30 40 50 60 70 80 90 100

Cou

nts

Energy (keV)

SDD

CdTe

HPGe

Background

0 10 20 30 40 50 60 70 80 90 100

Cou

nts

Energy (keV)

Y

Ce Nd GdDyErYb Sample

SDD

CdTe

HPGe


107

Moreover, the escape peaks have a much more pronounced effect in CdTe than in SDD and HPGe detectors. The secondary X-rays produced in the detector by the interaction with the elements composing the detector escape it by reducing the measured energy. In the case of the CdTe detectors, three intrinsic properties of Cd and Te make the difference more important (Redus et al, 2002, 2009). Firstly, in Cd and Te the yield emission is near to 85 % (Markowitz, 2002) since it produces more X-rays compared to the Si where only 5% of the atoms decay with emission of an X-ray. Secondly, in CdTe, both the Cd and the Te produce both Kα and Kβ peaks. Therefore, each primary photopeak produces four escape peaks, while in Si only the Kα peak is usually visible. Thirdly, in Si the characteristics X-rays have low energy, 1.74 keV, so only exit the detector for interactions near the surface. In Ge, these energies are higher than the Si, 9.9 and 11.0 keV, whereas in CdTe, the characteristic energies are 23.2 and 26.1 keV for the Cd Kα and Kβ, and 27.5 and 31.0 keV for the Te peaks. These have longer range so are important at higher energies by bringing an important noise at energies superior to 30 keV (Fig. 9).

III.B.4.2.3. The X-ray generator voltage effect During the modeling tests the effect of the X-ray generator voltage on the net counting

rate was also studied. For comparing its effect, the probability of detection was measured in function of several voltages as shown in Fig. 10.

The probability of detection is defined as the probability of detecting one fluorescent photon to the number of photons exited a surface of 1cm² at a distance from the source of 1 m for a given voltage of the X-ray generator and a current of 400 µA.

Fig. 10. Probability of detection of the tracers as a function of the X-ray generator voltage

Figure 10 shows that the probability of detection increases while the X-ray generator

voltage increases, by following a second order equation of the type. Consequently, for reducing the detection limit the device has to operate at elevated voltages of 150 to 160 kV.

0

2E-12

4E-12

6E-12

8E-12

1E-11

1.2E-11

1.4E-11

70 80 90 100 110 120 130 140 150 160

Prob

abili

ty o

f det

ectio

n

X-ray generator tension (kV)

Y2O3CeO2

Nd2O3

Gd2O3Dy2O3

Er2O3


108

From Figure 10, it can also been concluded that the probability of detection increases with the atomic number of the elements composing the tracers.

III.B.4.2.4. The overlapping and mass absorption effect

All the results given in Fig. 8 were obtained by considering that the samples contain only one tracer for each test. The idea of codification is based on the simultaneous presence of several tracers in the polymer matrix. Table 6 shows that Kβ1 energy lines of Nd, Gd and Er are close to the Kα1 energy lines of Gd, Er and Yb respectively. Thus by the simultaneous presence of these tracers, the raw counting rate of Gd, Er and Yb will probably be overvalued by the additional counts of Kβ1 energy lines of Nd, Gd and Er respectively.

Moreover, the detected fluorescence peaks in the case of simultaneous presence of the tracers could be affected to a small extent by absorption and enhancement effects caused by the other elements composing the tracers present in the matrix (AMC, 2006, Bao, 1999).

In order to estimate the probable overlapping interferences of Nd, Gd and Er to Gd, Er and Yb respectively, as well as the mass absorption effect of the simultaneous presence of tracers in the same sample, Fig. 11 compares the simulation results of probability of detection as a function of tracers for a sample containing at 1000 ppm all the tracers simultaneously and one with a single tracer.

Fig. 11. Comparison of modeling results for a sample containing all the tracers simultaneously and one with a

single tracer at each test For the three detectors tested, the probability of detection is higher for the Gd2O3, Er2O3 and Yb2O3 in the case of the sample containing all the tracers simultaneously, which proves indeed the overlapping interferences of Kβ1 energy lines of Nd2O3, Gd2O3 and Er2O3. Regarding the mass absorption effect of the simultaneous presence of all the tracers in the sample it seems that for the tracers not concerned by the overlapping interferences the values of the raw counting rate are similar in both cases.

0.E+002.E-124.E-126.E-128.E-121.E-111.E-111.E-112.E-112.E-112.E-11


Prob

abili

ty o

f det

ectio

n

Tracers

SDD_all

HPGe_all

CdTe_singleCdTe_all

SDD_single

HPGe_single

0


109

III.B.4.3. Comparison between experimental and modeling results

Table 8 compares the detection limit for 10 ms between the experimental data and the

modeling results. As it can be noticed, the detection limit for the CdTe detector slightly improves even if its thickness was reduced. As explained above, the escape peaks of Cd and Te, prevent the CdTe detector to better performed. By using this detector the best results for the detection limit at 10 ms will is the order of 500 ppm.

Regarding the SDD detector the modeling results show it could be a good candidate to fulfil the need. This was partly expected since as it can be seen in Fig. 7 the photoelectric efficiency is higher for the SDD_0.5 compared to the SDD_0.3; hence the increase of SDD thickness has positively reacted by decreasing the detection limit. The thick SDD detector could be a solution for the identification of Y2O3, CeO2, Nd2O3 and Gd2O3. The development of new thicker SDD detectors could be a solution for the high speed detection of elements having energies coming out in the range of [15-50 keV].

The HPGe detector has given the best modeling results among the three detectors. By comparing the experimental and modeling results, HPGe_5 seems to perform approximately 10 times better than HPGe_10, even if its efficiency was reduced (Figure 7). However, for the Monte Carlo method used by MACALU, in the case of the HPGe detector, the calculated efficiency is very sensitive to the thickness of the germanium dead layer; a layer of inactive germanium that is not useful for detection, but which strongly attenuates the photons and acts as a filter (Rodenas et al, 2003). The dead layer contributes to the existence of a transition zone between the inactive layer and the active germanium crystal. Photons absorbed in the transition zone do not contribute to the full energy peak count rate and thus the modeling results of the net counting rate is overestimated and this explains why the Ge_5 performed better than the Ge_10. Nevertheless, a real commercial HPGe detector of 5 mm thickness has to be tested experimentally for confirming the modeling results.

Table 8 Comparison of detection limit for 10 ms between experimental and modeling results

Detection limit for 10 ms (ppm) Experimental results Modeling results SDD_0.3 HPGe_10 CdTe_1 SDD_0.5 HPGe_5 CdTe_0.5 Y2O3 < 1600 - < 1400 < 100 < 300 < 500 CeO2 < 2000 < 1300 < 1100 < 150 < 100 < 500 Nd2O3 < 2300 < 1100 < 1000 < 250 < 100 < 500 Gd2O3 < 3800 < 1000 < 1000 < 500 < 100 < 500 Dy2O3 < 6700 < 1500 < 1500 < 650 < 150 < 500 Er2O3 < 6400 < 1500 < 1300 < 1000 < 200 < 600 Yb2O3 < 10000 < 2000 < 2400 < 2000 < 400 < 1000


110

III.B.5. Conclusion

In this study we compared 3 X-ray semiconductor detectors (SDD, HPGe and CdTe) and some secondary parameters, which could play an important role in the identification of rare earth elements in the content of high speed sampling time (Fig. 12).

Fig. 12. Summary of the parameters studied in this work

The experimental results show that CdTe and HPGe detectors clearly have much

higher sensitivity than SDD detector at energies above 30 keV, corresponding to K lines of rare earth elements. This high efficiency allowed the detection of rare earth oxides, added to a polymer matrix, for a concentration range of [1000-2000 ppm] for 10 ms acquisition time, for a distance between the X-ray generator and the sample of 120 mm and a distance between the detector and the sample of 120 mm. The thin thickness of the SDD detector tested, does not allow the detection of tracers to the required concentrations.

For a detection limit lower than 1000 ppm the modeling results are more contrasted. These results pointed out that HPGe detector could be the best detector for the high speed identification of rare earth elements, thanks to its high efficiency and energy resolution by detecting the majority of tracers at 100 ppm. However, these modeling results have to be taken with caution since the dead layer of the HPGe detector could have an effect on the Monte Carlo method by overestimating the counted photons. For the SDD detector, the increase of the thickness decreases the detection limit of tracers. According to the modeling results the SDD could detect Y2O3, CeO2, Nd2O3 and Gd2O3 to a concentration range of [100-500 ppm]. Regarding the CdTe detector, the modeling results showed that the detection limit is limited by the escape peaks effect in the range [500-1000 ppm] for all the tracers tested.

In addition to the detectors effect on the detection limit, we have also studied the influence of other parameters such as the filter thickness and the voltage of the X-ray generator in the detection limit as well as the probable overlapping and mass absorption effect

S

D

*

F

Cu filter thickness: [0 - 2 mm]

thickness= 3 mm

width = 20-30 mm

Source Filter Traced material

Detector

Voltage of X-ray generator :

[70 – 160 keV]

Tracers :Y2O3, CeO2, Nd2O3, Gd2O3,

Dy2O3, Er2O3 and Yb2O3

Detectors :

SDD HPGe CdTe

S t10 0.3

100 0.5

S : Surface (mm²)

t: thickness (mm)

S t200 10

100 5

S t25 1

100 0.5

Experimental

Modeling


111

of tracers with respect to their identification. The filter thickness depends on the semiconductor thickness and its efficiency. For example, for a 5 mm thick SDD detector the best value for the copper filter is 1 mm, whereas for the HPGe of 5 mm and the CdTe of 0.5 mm, this value is in the range [1-2 mm]. The results obtained for the influence of the X-ray generator voltage showed that for reducing the detection limit the device has to perform at 150 to 160 kV. Regarding the overlapping interferences of some tracers between them it has been shown that Kβ1 energy lines of Nd, Gd and Er affect the Kα1 energy lines of Gd, Er and Yb respectively by increasing the net counting rate. Acknowledgements

The authors would like to thank the French Industry-University Cooperative Research Network on Waste (RECORD – www.record-net.org ), the French Environment and Energy Management Agency (ADEME – www.ademe.fr ) for their contribution to the funding of this work and for providing industrial orientations and scientific supervision to the research.




112

References AMC (Analytical Methods Committee), 2006. Evaluation of analytical instrumentation. Part XX Instrumentation for energy dispersive X-ray fluorescence spectrometry. Accreditation and Quality Assurance: Journal for Quality, Comparability and Reliability in Chemical Measurement 11, 610-624. Bao, S., 1999. Combination of Corrections for Absorption, Overlap and Background in XRF Spectrometry. X-Ray Spectrom. 28, 141-144. Bezati, F., Froelich, D., Massardier, V., Maris, E., 2010. Addition of tracers into the polypropylene in view of automatic sorting of plastic wastes using X-ray fluorescence spectrometry. Waste Manag. 30, 591–596. Bezati, F., Massardier, V., Froelich, D., Maris, E., Balcaen, J., 2010. Elaboration and characterization of traced polypropylene with rare earth oxides for automatic identification and sorting of end-of-life plastics. Waste Biomass Valor. Published online June 2010. Castoldi, A., Fiorini, C., Guazzoni, C., Longoni, A., Strüder, L., 1999. Semiconductor drift detectors: Applications and new devices. X-Ray Spectrom. 28, 312-316. European Parliament, 2000. Directive 2000/53/EC of the European Parliament and of the Council of 18 September 2000 on End-Life Vehicles. Brussels, Belgium. Froelich, D., Maris, E., Haoues, N., Chemineau, L., Renard, H., Abraham, F., 2007. State of the art of plastic sorting and recycling: feedback to vehicle design. Miner. Eng. 20, 902–912. Gatti, E., Rehak, P., 2005. Review of semiconductor drift detectors. Nuclear Nucl. Instrum. Meth. A 541, 47-60. Gros d’Aillon, E., 2005. Etude des performances spectrométriques des détecteurs gamma CdZnTe / CdTe monolithiques. Université Joseph Fourier, Grenoble, France. Kamae, T., 1999. Developments in semiconductor detector technology and new applications - symposium summary. Nucl. Instrum. Meth. A. 36, 297-303. Markowicz, A., 2002. X-ray physics, in Handbook of X-Ray Spectrometry, 2nd Edition, ed. R.A. Van Grieken, A.A. Markowicz, Marcel Dekkerm. New York and Basel. 58. Morris, K.E., Mueller, W.F., Blanc, P., Bronson, F., Croft, S., Field, M.B., Nakazawa, D.R., Venkataraman R., Zhu, H., Efficiency characterization of germanium detectors at energies less than 45keV. Canberra Industries Inc. http://www.canberra.com. Neuilly, M., 1993. Modélisation et estimation des erreurs de mesure. Technique et Documentation – Lavoisier, Paris. Owens, A., Peacock, A., 2004. Compound semiconductor radiation detectors. Nucl. Instrum. Meth. A 531, 18–37. Rausa, R., Pollesel, P., 1997. Pyrolysis of automotive shredder residue (ASR) influence of temperature on the distribution of products. J. Anal. App. Pyrol. 41, 383-401. Redus, R., Pantazis, J., Huber, A., Pantazis, T., 2002. Improved sensitivity X-ray detectors for field applications. IEEE T. Nucl. Sci., 49, 3247-3253. Redus, R. H., Pantazis, J. A., Pantazis, T. J., Huber, A.C., Cross, B. J., 2009. Characterization of CdTe detectors for quantitative X-ray spectroscopy. IEEE T. Nucl. Sci, 56, 2524-2532. Ródenas, J., Pascuala, A., Zarzaa, I., Serradellb, V., Ortizb, J., Ballesterosb, L., 2003. Analysis of the influence of germanium dead layer on detector calibration simulation for

http://www.canberra.com/


113

environmental radioactive samples using the Monte Carlo method. Nucl. Instrum. Meth. A 496, 390-399. Sakai E., 1982. Present status of room temperature semiconductor detectors. Nucl. Instrum. Meth. A. 196, 121-130. Sato, G., Takahashi, T., Sugiho, M., Kouda, M., Mitani, T., Nakazawa, K., 2002. Characterization of CdTe/CdZnTe detectors. IEEE T. Nucl. Sci. 49, 1258-1263. Takahashi, T., Mitani, T., Kobayashi, Y., Kouda, M., Sato, G., Watanabe, S., 2002. High-resolution schottky CdTe diode detector. IEEE T. Nucl. Sci. 49, 1297-1303. Takahashi, T., Watanabe, S., 2001. Recent progress in CdTe and CdZnTe detectors. IEEE T. Nucl. Sci. 48, 950-959. Tola, F., 1995. MACALU - Code Monte-Carlo de simulation du transport de photons γ ou X. Technical note N° R95-118, CEA Saclay, France. Tola, F., 1994. An expert system for the conception of industrial gauges based on beta, gamma or X-rays transmission (JANU). Nucl. Instrum. Meth. A 353, 706-709.


114

Conclusion Chapitre III

Que cela soit les essais effectués chez NRT ou ceux au LITT du CEA de Saclay, il a été prouvé que la détection des oxydes de terre rare, utilisés en tant que traceurs dans une matrice PP, est possible et à de faibles concentrations. Cependant la limite de détection dépend fortement du choix du détecteur.

En comparant les résultats expérimentaux de trois types de détecteurs différents (SDD, HPGe et CdTe), utilisés dans cette étude, nous remarquons que le CdTe et le HPGe ont nettement la meilleure sensibilité pour la détection des énergies supérieures à 30 keV, correspondant aux raies énergétiques K de la majorité des traceurs. Cette efficacité élevée dans ce domaine permet la détection d’oxydes de terre rare dans une gamme de concentration de [1000-2000 ppm] pour un temps d’acquisition de 10 ms, pour une distance entre le générateur de rayons X et l’échantillon de 120 mm et une distance entre le détecteur et l’échantillon de 120 mm également. La faible épaisseur du détecteur SDD testé n’a pas permis la détection des traceurs dans la gamme de concentration requise.

En ce qui concerne les résultats de la modélisation, ceux-ci ont montré que le détecteur HPGe peut être la meilleure solution pour l’identification rapide des oxydes de terre rare, grâce à son efficacité élevée et à la résolution en énergie. En se basant sur les résultats de la modélisation, on peut penser qu’il peut détecter la majorité des traceurs à une concentration de 100 ppm. Cependant ces résultats doivent être pris avec précaution puisque la couche inactive du HPGe peut avoir un impact sur les calculs de la méthode Monte Carlo en surestimant le nombre de photons comptés.

Pour le détecteur SDD, l’augmentation de l’épaisseur du semi-conducteur peut nettement améliorer la limite de détection en la descendant en des sous de 500 ppm pour l’Y2O3, le CeO2, le Nd2O3 et le Gd2O3. Par contre, dans le cas du CdTe, la limite de détection est limitée dans le domaine entre [500-1000 ppm] due aux pics d’échappements du cadmium et du tellure, qui créent un bruit de fond très fort dans le domaine de sortie des pics des traceurs.

Concernant les paramètres secondaires, tel que l’épaisseur du filtre, cela dépend de l’épaisseur du semi-conducteur ainsi que de son efficacité. Pour les détecteurs utilisés dans cette étude le meilleur compromis se trouve dans le domaine entre [1-2 mm]. Pour ce qui en est de la tension du générateur de rayons X, les résultats de modélisation ont montré que celui-là fonctionne mieux pour des tensions élevées de l’ordre de 150 à 160 keV.

Pour finir, les résultats obtenus par les deux articles ont montré qu’il peut avoir des chevauchements de pics entre les traceurs qui conduisent à une surestimation du taux de comptage net si les traceurs interférents sont présentés simultanément. Néanmoins, cette problématique peut être résolue en choisissant bien les traceurs et les combinaisons possibles.

115

Conclusion Générale

Cette thèse a débuté suite à deux études financées par ReCoRD et confiées conjointement à l’IMP et au LCPI sur l’ajout de traceurs dans le polypropylène en vue de l’automatisation du tri des déchets plastiques. Ces deux travaux ont conclu qu’une technique spectroscopique prometteuse pour la récupération des matériaux plastiques contenus dans les VHU et des DEEE peut être la spectrométrie de fluorescence X.

L’objectif principal de ce projet était de prouver la faisabilité technique de détection par la SFX de traceurs ajoutés dans une matrice polymère afin de développer un tri efficace et rentable. Pour atteindre cet objectif, nous avons tout d’abord proposé les traceurs qui conviendraient le mieux à notre problématique et, suite à leur choix, nous avons articulé notre recherche sur deux axes :

- le premier était la validation de ces traceurs par rapport à leur impact sur les propriétés du matériau tracé. - le deuxième était leur validation par rapport à la détection par SFX. La sélection des traceurs a été accomplie par un processus en deux étapes. Puisque la

SFX est une méthode d’analyse élémentaire, la première étape était de choisir les éléments les plus adaptés du Tableau Périodique de Mendeleïev, puis de choisir la formule chimique qui conviendrait le plus au niveau de la stabilité et de la disponibilité. En appliquant des critères tels que la toxicité et la radioactivité, l’intensité forte du signal de détection, la singularité du pic, la disponibilité des réserves ainsi que des contraintes économiques, nous avons pu trouver que les traceurs a priori les plus adaptés pour notre problématique étaient des oxydes de terres rares.

Le nombre d’études traitant l’ajout de telles particules sur des matrices polymères et à des concentrations de 100 à 1000 ppm est très faible. En conséquence, nous avons dû caractériser les matériaux tracés pour connaitre l’impact des oxydes de terres rares sur la matrice PP et les concentrations pour lesquelles ces propriétés ne sont pas significativement affectées.

Les résultats de ces travaux ont montré que l’addition de 1000 ppm de particules d’oxydes de terres rares dans la matrice PP peut réduire l’allongement à la rupture dans une fourchette [10-50%] pour une vitesse de 250 mm/min. Pour la même concentration, les caractéristiques physico-chimiques, analysées par DSC, ont légèrement augmenté, tandis que la stabilité thermique (ATG) a été améliorée.

En ce qui concerne la photo-dégradation sous rayonnement UV, les résultats obtenus par FTIR, DSC, ATG et caractérisation mécanique ont montré que l’addition de particules de CeO2 dans la matrice PP peut améliorer son comportement en résistant plus longtemps à la photo-dégradation sous rayonnement UV. En effet, les particules de CeO2 jouent un rôle important dans la stabilisation du polypropylène en retardant le processus de photo-dégradation et en agissant comme des absorbeurs d’irradiations UV. Pour les traceurs restants, les résultats obtenus ont montré le même comportement sous photo-dégradation que la matrice PP de référence.

116

Un des paramètres clés pour la bonne détection des traceurs par SFX et la minimisation de leur influence sur les propriétés de la matrice polymère est leur dispersion. Celle-ci a été caractérisée quantitativement en appliquant un traitement d’image par Matlab sur les images MEB. Nous avons comparé la distance entre une particule de traceur et dix voisins à la distance théorique pour un empilement cubique et hexagonal. Les résultats de traitement d’image ainsi que les images MEB ont montré une dispersion homogène des traceurs dans la matrice PP, sans formation d’agglomérats pour des concentrations unitaires de traceurs variant de 1000 à 10000ppm.

Suite à la caractérisation des matériaux tracés, nous avons défini une contrainte sur la concentration en traceur. Pour avoir un faible impact sur les propriétés mécaniques et physico-chimiques, ainsi que sur la photo-dégradation sous rayonnement UV celle-ci doit être inférieure à 1000 ppm.

Dans le cadre de la détection des traceurs et de l’optimisation de leur concentration nous avons testé expérimentalement des échantillons avec trois types de détecteurs différents. Les résultats expérimentaux ont montré que le CdTe et le HPGe ont nettement la meilleure sensibilité pour la détection des énergies supérieures à 30 keV, correspondant aux raies énergétiques K de la majorité des traceurs. La gamme de concentration pour laquelle ces détecteurs ont identifié les oxydes de terre rare est [1000-2000 ppm] pour un temps d’acquisition de 10 ms. Concernant le détecteur SDD étudié, son faible épaisseur limite son rendement et en conséquence la détection de la majorité des oxydes de terre rare, qui ont des raies énergétiques supérieurs à 30 keV, est très compliquée.

Les résultats de la modélisation ont montré que le HPGe peut être la meilleure solution pour la détection des oxydes de terre rare. Par contre la différence entre un SDD, plus épais que celui utilisé pour l’obtention des résultats expérimentaux, et un CdTe est que le premier est plus performant pour les traceurs plus légers, tels que Y2O3, CeO2 et Nd2O3 tandis que le deuxième donne de meilleurs résultats pour les éléments plus lourds.

Ce qui est très intéressant est de comparer les résultats expérimentaux avec ceux de la modélisation. Dans le cas du CdTe, nous remarquons que la limite de détection est limitée dans le domaine [500-1000 ppm] principalement à cause de pics d’échappements du cadmium et du tellure, qui créent un bruit de fond très important dans le domaine de sortie des pics de traceurs. En utilisant ce type de détecteur, les meilleurs résultats seront limités à un seuil de détection de 500 ppm pour un temps d’acquisition de 10 ms.

Dans le cas du HPGe, la comparaison entre les résultats expérimentaux et ceux de la modélisation montre une amélioration d’un facteur dix environ. Celle-ci doit être prise avec précaution puisque la couche inactive du HPGe peut avoir un impact sur les calculs de la méthode Monte Carlo en surestimant le nombre de photons comptés dans le taux de comptage net.

Pour finir, dans le cas du SDD, la limite de détection obtenue par modélisation pour un semi-conducteur plus épais est nettement plus basse que pour celui utilisé dans le banc d’essai. L’augmentation de l’épaisseur a amélioré le rendement du détecteur pour les énergies supérieures à 30 keV, ainsi le SDD peut être une solution pour la détection d’Y2O3, CeO2, Nd2O3 et Gd2O3.

Pour revenir sur le choix du détecteur le plus adapté pour l’identification rapide des oxydes de terre rare la réponse n’est pas simple. Si on se base sur les résultats expérimentaux,

117

il semble que le HPGe et le CdTe sont les détecteurs donnant les meilleurs résultats pour tous les traceurs testés. Pour ce qui est des tests en statique il ne faut pas négliger que le SDD utilisé est d’une très faible épaisseur et, que, actuellement sur le marché il existe des capteurs SDD plus performants.

Si maintenant, on prend en compte les résultats de la modélisation, le HPGe donne les meilleurs tandis que, dans le cas du CdTe, la limite de détection est limitée par les pics d’échappement. Afin d’avoir une réponse claire sur le HPGe, il faudra tester un nouveau détecteur, un peu moins épais que celui des tests, pour pouvoir conclure. Concernant le SDD, via les résultats de la modélisation nous avons vu qu’il donne de très bons résultats pour l’Y2O3, le CeO2, le Nd2O3 et le Gd2O3. En prenant en considération les réserves des tous les traceurs proposés, il apparaît que ces quatre là ont les disponibilités les plus importantes. Par conséquent, si dans le futur, il est décidé d’identifier les traceurs qui ont des réserves importantes (Gd, Y, Ce, Nd), le SDD peut être le meilleur choix tant d’un point de vue économique et que des performances.

118

Perspectives Les résultats de ce travail de thèse ont donné lieu à un programme ECOTECH de 45

mois financé par l’Agence Nationale de la Recherche, sous le nom de TRIPTIC, Traceurs Répartis pour l’Identification des Polymères et le Tri Industriel en Cadence. L’objectif du projet TRIPTIC est de développer une technologie de marquage des polymères grâce à des systèmes traceurs, que cela soit en SFX ou fluorescence UV, ainsi que le développement de pilotes de tri. Il fixe aussi comme objectif final du projet, l’arrivée à une normalisation du traçage des matériaux polymères avec l’AFNOR.

En ce qui concerne le projet de thèse en lui-même pour la caractérisation des matériaux tracés, des études supplémentaires peuvent être envisagées comme complément aux résultats présentés.

Tout d’abord, il conviendra de valider le cahier des charges lié aux matériaux tracés, comme le coût de l’ajout de traceurs, la mise en peinture, l’aspect ainsi que l’impact du traceur sur les matériaux tracés recyclés

Deuxièmement, le protocole d’addition des traceurs peut être optimisé par la recherche d’autres procédés de mise en œuvre qui peuvent faciliter leur dispersion tout en optimisant la quantité de composés organiques volatils émis lors des opérations. Il serait aussi intéressant d’étudier la distribution des traceurs dans des pièces réelles complexes, des phénomènes de ségrégation pouvant intervenir pendant la mise en œuvre.

Pour finir, des études supplémentaires sur les propriétés mécaniques, physico-chimiques et du vieillissement climatique peuvent être considérées sur d’autres grades de PP (dédiées à des applications pour l’industrie automobile) ou d’autres familles de polymères tels que l’acrylonitrile butadiène styrène (ABS) fortement utilisé dans les industries automobiles, électriques et électroniques.

Pour la partie détection, en sachant que d’un point de vue des performances et du prix, le

détecteur reste un paramètre clé, il faudra tester un SDD plus épais ainsi qu’un HPGe plus mince afin de confirmer les résultats de la modélisation.

Une autre possibilité pour la détection des oxydes de terres rares est l’utilisation de la SFX en longueur d’onde (WD-XRF). L’avantage de celle-ci est sa très bonne résolution, obtenue grâce à l’utilisation d’un cristal faisant office de réseau. Cependant, le rendement de cette technique est très faible, ce qui signifie qu’il faudrait augmenter le temps d’acquisition et/ou avoir un générateur très puissant.

Les tests ayant lieu chez NRT et le LITT du CEA de Saclay étaient en statique. La prochaine étape est de réaliser des analyses en dynamique et cela sera possible grâce au pilote qui va être construit pour le projet TRIPTIC. Pour arriver à la réalisation du pilote, il faudra continuer les tests sur le banc d’essais du LITT, afin de déterminer le cahier des charges de la détection pour l’applicabilité sur une machine de tri industrielle. Pour aboutir à cela, il faudra trouver la distance et la largeur optimale de détection, la nature et la puissance du générateur de rayons X, la granulométrie des produits, ainsi que la vitesse de défilement et le temps de remplissage du convoyeur pour pouvoir déterminer la cadence de tri en kg/h.

119

En plus de l’utilisation de la SFX avec un système traceur, nous pouvons envisager son application pour l’identification des contaminants présents dans des matériaux d’ancienne génération, tels que les métaux lourds et les retardateurs de flamme. En effet, l’usage de Cadmium ou de Plomb dans les colorants de plastiques, ou de retardateurs de flamme de type PBDE (Polybromodiphényléther) était courant il y a une quinzaine d’années et en détectant ces éléments nous pourrons surement les trier, dans les résidus de broyage. Cependant, quelques analyses de modélisation, non présentées dans cette thèse, ont montré que le temps d’acquisition est trop long pour détecter 100 ppm de Cadmium, tandis que dans le cas du Plomb, le pic du contaminant est confondu avec le pic du piège et du blindage, qui sont composés du même élément. Donc, pour envisager la récupération des contaminants contenant du Plomb il faudra penser à un autre type de blindage.

Nous avons vu qu’en utilisant la SFX, le nombre de traceurs potentiels est limité par le

Tableau Périodique de Mendeleïev. En sachant que la cible à terme est une normalisation des traceurs pour le tri et le recyclage des polymères, il serait bon d’avoir une vision globale de la problématique du traçage à court et à long terme, afin de savoir le nombre de traceurs nécessaires. Pour résoudre ce problème il faudra définir quelles seraient les familles et les grades de matériaux pertinentes pour le recyclage aujourd’hui et à terme. Ce choix doit tenir compte de critères industriels comme l’existence ou pas de procédés ou de filières de recyclage de ces matériaux à terme, des gisements potentiels de ces matières, des performances mécaniques de ces matériaux une fois recyclés, les évolutions prévisibles de la réglementation et enfin de l’intérêt environnemental et économique d’un tel recyclage.

120

Annexes Annexe I : La spectrométrie de fluorescence X [1, 2] Principe de base

La fluorescence X est une propriété spectrale des atomes exploitée couramment en analyse pour obtenir des renseignements qualitatifs ou quantitatifs sur la composition élémentaire de toutes sortes de mélanges.

Le principe consiste à irradier l’échantillon soit par un faisceau de rayons X, soit par bombardement avec des particules, généralement des électrons ayant suffisamment d’énergie, pour que les atomes ainsi ionisés émettent un rayonnement de fluorescence également dans le domaine des rayons X.

Le caractère universel du phénomène et la possibilité de faire un examen rapide sur des échantillons restés le plus souvent dans leur état d’origine, expliquent le succès de cette méthode d’analyse non destructive.

Quand on irradie avec des photons ou qu’on bombarde avec des particules de grande énergie (entre 5 et 100 keV) un matériau servant de cible, celui-ci émet une fluorescence située dans le domaine des rayons X. Le spectre de cette photoluminescence comporte des radiations dont les longueurs d’ondes sont caractéristiques des atomes de ce matériau.

L’échantillon, soumis à l’excitation d’une source primaire de rayons X émet un rayonnement de fluorescence qui peut conduire à deux types de spectres :

- Spectre en énergie (ED-XRF) obtenu directement au moyen d’une diode dont le signal diffère selon l’énergie de chaque photon incident ;

- Spectre en longueur d’onde (WD-XRF) obtenu par rotation d’un cristal faisant office de réseau (montage goniométrique comportant un ou plusieurs détecteurs mobiles). Cependant énergie et longueur d’onde étant reliées, on présente les spectres en unités d’énergie (keV), quel que soit le mode de détection.

Dans le cas de notre étude nous sommes uniquement intéressés sur les spectres en énergie en utilisant des semi-conducteurs. Le spectre de fluorescence X La fluorescence X d’un atome isolé résulte d’un processus en deux temps : photoionisation de l’atome, au cours de laquelle l’impact du photon extérieur incident se

traduit par l’arrachement d’un électron interne de l’atome, tel un électron K si le photon a suffisamment d’énergie. Cet effet photoélectrique conduit à l’émission d’un photoélectron (Figure 1) et à un atome ionisé par suite d’une lacune interne.

stabilisation de l’atome ionisé, qui correspond à la ré-émission de tout ou partie de l’énergie acquise au cours de l’excitation. La lacune créée précédemment est suivie d’une réorganisation quasi instantanée (en 10-16s) des électrons situés dans les différents niveaux de cet atome ionisé, ce qui le ramène très vite vers un état de faible énergie. Des réarrangements en cascade sont observés pour les atomes lourds, à la différence des

121

éléments légers dont les électrons se répartissent sur un nombre plus restreint de niveaux de base.

Figure 1 : Schémas simplifiés montrant l’origine de quelques transitions de fluorescence.

Cette réorganisation fait naître des photons de fluorescence. En désignant par E1

l’énergie de l’électron qui occupait la lacune considérée et E2 l’énergie de l’électron qui vient combler cette lacune, il pourra apparaître (probabilité comprise entre 0 et 1) un photon de fluorescence caractérisé par une fréquence ν, telle que :

hν = | E2 - E1 | Chaque atome, à partir de Z = 3, conduit à un ensemble de radiations spécifiques, obéissant à des règles de sélection quantiques de ses électrons (Δn > 0, Δl = ±1). Les éléments hydrogène et hélium ne peuvent avoir de spectre de fluorescence X, n’ayant pas d’électron dans le niveau L. Du béryllium au fluor on note une seule transition de type Kα. Puis les atomes devenant plus gros, le nombre de transitions possibles croît (75 pour le mercure) mais la probabilité de certaines est très faible. Heureusement, il suffit pour caractériser un élément de repérer les quelques transitions les plus intenses. Pour l’ensemble des éléments, la fluorescence se situe dans une large plage allant de 40 eV à plus de 100 keV (31 à 0.012nm). La désignation précise des différentes transitions électroniques possibles fait référence aux niveaux d’énergie des orbitales de l’atome considéré, mais on utilise aussi une nomenclature simplifiée due à Siegbahn. Ainsi CeKβ2 symbolise, pour l’élément cérium, la transition qui correspond au passage M – K et 2, l’intensité relative de la transition dans la série (1, plus intense que 2). Les transitions Kβ sont approximativement six fois moins probables (donc moins intenses) que les transitions Kα correspondantes, α marquant la « distance » la plus proche. Pour les niveaux L, M, qui sont multiples, la notation selon Siegbahn n’est pas toujours suffisamment explicite. Référence

[1] Rouessac F, Rouessac A, Bertrand MJ, Waldron KC. Chemical analysis: Modern instrumental methods and techniques. Chichester; New York ; Weinheim [etc.: J. Wiley & sons, cop. 2000; op. 2000.

[2] Van Grieken, René Editeur scientifique, Markowicz. Handbook of X-Ray spectrometry: methods and techniques. New York ; Basel ; Hong Kong: M. Dekker, cop. 1993; op. 1993.

122

Annexe II : Les directives liées aux oxydes de terres rares

Dénomination CAS

Classement annexe I directive

67/548/CEE

Phrases de risques DL 50 oral, rat (mg/kg)

Concentration (%) au-delà de

laquelle un mélange est dangereux

selon arrêté du 09-11-2004

modifié Oxyde d'yttrium 1314-36-9 non dangereux sans objet > 5 000 sans objet

Oxyde de cérium 1306-38-3 non dangereux sans objet > 5 000 sans objet

Oxyde de néodyme 1313-97-9 non dangereux sans objet > 5 000 sans objet

Oxyde de gadolinium 12064-62-9 Irritant R36 - irritant pour les

yeux > 5 000 20

Oxyde de dysprosium 1308-87-8 non dangereux sans objet > 5 000 sans objet

Oxyde d'ytterbium 1314-37-0 non dangereux sans objet > 5 000 sans objet

Références

[1] Directive 98/98/CE du 15 décembre 1998, portant vingt-cinquième adaptation au progrès

technique de la directive 67/548/CEE, JOCE L355, 30 décembre 1998.

[2] Directive 98/73/CE du 18 septembre 1998, portant vingt-quatrième adaptation au progrès

technique de la directive 67/548/CEE, JOCE L305, 16 novembre 1998.

[3] Directive 97/69/CE du 5 décembre 1997, portant vingt-troisième adaptation au progrès

technique de la directive 67/548/CEE, JOCE L343, 13 décembre 1997.

[4] Directive 96/54/CE du 30 juillet 1996, portant vingt-deuxième adaptation au progrès

technique de la directive 67/548/CEE, JOCE L248, 30 septembre 1996.

[5] Directive 94/69/CE du 19 décembre 1994, portant vingt et unième adaptation au progrès

technique de la directive 67/548/CEE (volumes I et II), JOCE L381, 31 décembre 1994.

[6] Directive 93/101/CE du 11 novembre 1993, portant vingtième adaptation au progrès

technique de la directive 67/548/CEE, JOCE L13, 15 janvier 1994.

[7] Directive 93/72/CEE du 1er septembre 1993, portant dix-neuvième adaptation au progrès

technique de la directive 67/548/CEE (volumes I et II), JOCE L258 et L258A, 16 octobre

1993.

123

[8] Directive 67/548/CEE du 27 juin 1967 modifiée , concernant le rapprochement des

dispositions législatives, réglementaires et administratives relatives à la classification,

l’emballage et l’étiquetage des substances dangereuses, Code Permanent Environnement et

Nuisances, EDITIONS LEGISLATIVES .

[9] Directive 1999/45/CE du 31 mai 1999 modifiée concernant le rapprochement des

dispositions législatives, réglementaires et administratives des états membres relatives à la

classification, à l’emballage et à l’étiquetage des préparations dangereuses, Dictionnaire

Permanent Sécurité et Conditions de Travail, EDITIONS LEGISLATIVES .

[10] Arrêté du 9 novembre 2004 relatif aux critères de classification et aux conditions

d’étiquetage et d’emballage de préparations dangereuses, Code Permanent Environnement et

Nuisances, EDITIONS LEGISLATIVES.

[11] Directive 2000/32/CE du 19 mai 2000 portant vingt-sixième adaptation au progrès

technique de la directive 67/548/CEE, JOCE L 136, 8 juin 2000 .

[12] Directive 2001/59/CE du 6 août 2001 portant vingt-huitième adaptation au progrès

technique de la directive 67/548/CEE, JOCE L 225, 21 août 2001 .

[13] Directive 2004/73/CE du 29 avril 2004 portant vingt-neuvième adaptation au progrès

technique de la directive 67/548/CEE, JOUE L 216, 16 juin 2004 .

[14] Base de données MERCK, VWR

124

Annexe 3: Description de MACALU [1, 2] Présentation du logiciel MACALU

MACALU est un code de calcul Monte-Carlo, développé par M. François TOLA, destiné à la modélisation de la réponse de chaînes de mesures nucléaires dans une configuration expérimentale simple. Cet outil est notamment un moyen d'aide à la conception de jauges (d'épaisseur, densité, …), basées sur l'atténuation ou la diffusion d'un rayonnement γ ou X émis par une source radioactive ou un générateur X, entre 1 keV et 10 MeV.

Son intérêt est de disposer d'une base de données complète et validée (sections efficaces, propriétés physiques des éléments, spectre énergétique des principales sources γ ou X et de générateurs X, détecteurs à scintillation, matériaux de blindage,…), d'être très convivial et accessible au non-spécialiste. L'interface utilisateur est en Prolog, tandis que le noyau de calcul Monte-Carlo a été développé en langage C, afin d'améliorer ses performances. Dispositif de mesure et géométrie d’installation La Figure 1 montre la géométrie de l'installation. Celle-ci comporte 3 parties :

– la source, supposée ponctuelle et son collimateur, supposé parfait. Seuls sont pris en compte les photons émis dans l'angle solide défini par le collimateur ;

– des écrans infinis et homogènes, d'épaisseur donnée, dont le nombre et la composition peuvent être quelconques ;

– le bloc détecteur, composé d'un volume de détection cylindrique de composition quelconque et d'un collimateur qui fait également office de blindage. Ce dernier est facultatif, et l'orifice de collimation peut être circulaire ou rectangulaire tout comme celui de la source.

Figure 4 : MACALU : Dispositif de mesure simulé

Source et détecteur peuvent être placés soit de part et d'autre des écrans, soit du même coté. Leur inclinaison par rapport à un axe normal aux écrans peut être quelconque et leur

Source γ ou X

Bloc détecteur Écrans

F

Cristal

Blindage

Trajectoire d'un photon

Source γ ou X

Bloc détecteur Écrans

F

Cristal

Blindage

Trajectoire d'un photon

125

position est définie par la distance au point focal F. Cette configuration peut être visualisée graphiquement et modifiée après avoir rentré les données.

Principe de calcul

On trouvera une description détaillée des méthodes Monte-Carlo appliquées au transport de photons et de neutrons dans [3]. Nous nous sommes largement inspirés de [4] pour les calculs concernant la génération aléatoire des paramètres du photon. Un recueil des principaux générateurs aléatoires est donné dans [5] et [6].

Nous avons retenu celui proposé par D.H. LEHNER, réputé comme sûr, qui est basé sur une loi congruentielle linéaire de la forme :

z a z mn n+ = ∗1 mod , avec a = 75 et m = −2 131 .

Le code prend en compte les diffusions multiples des photons dans les différents écrans, dans le détecteur à scintillation et dans le bloc collimateur. On suppose qu'un photon émis par la source ne peut atteindre le bloc détecteur après avoir traversé ou interagi avec les écrans. On autorise le retour éventuel d'un photon vers l'écran précédent ainsi que du détecteur vers le bloc collimateur et vice versa.

Par contre, un photon quittant le bloc détecteur sera considéré comme perdu quelque soit sa direction de départ, la probabilité de retour, tout comme l'énergie associée au photon résultant étant très faibles. Un tel parcours est illustré sur la Figure 1.

Les interactions rayonnement-matière suivantes sont considérées : – effet photoélectrique, – effet Rayleigh, ou diffusion cohérente, – effet Compton, ou diffusion incohérente.

Les particules et le rayonnement d'annihilation produits lors de la création de paire ne

sont pas pris en compte par la simulation. Dans ce cas, on suppose que toute l'énergie du photon est cédée au milieu.

La Figure 2 donne l'organigramme du module Monte-Carlo pour la simulation du transport d'un photon γ ou X.

La Figure 3, celle de la trace du photon dans un milieu quelconque (écran, détecteur ou collimateur).

126

Monte-Carlo

Incrémentation du nombre de photons émis

Emission d'un photon

Passage au repère (foyer,écran)

Sélection du premier écran

Sélection de l'écran suivant

le photon n'est pas absorbé etl'indice de l'écran existe et

il y a impact du photon sur l'écranO N

Le photon est absorbé ON

Passage au repére (foyer,détecteur)

O N

Le photon a cédé de l'énergie auO N

Incrémentation du nombre de photons détectés

Incrémentation des effectifs de la classe

Trace du photon dans le cristal

La touche Esc a été activée ON

Fin

1

2

détecteur

associée à l'énergie cédée au détecteur

par l'utilisateur

Trace du photon dans l'écran

Trace du photon jusqu'au cristal3

Le photon parvient au cristal

Mise à jour du nombre de photons

4et dans le collimateur

Figure 5 : Algorithme de la simulation Monte-Carlo

127

Récupération de pour l' énergie du photon

O N

Trace

Génération du libre parcours

O Nle photon n'est ni absorbé ni

sorti etson énergie est supérieure au seuil

Le photon sort du milieu par l'une de ses faces

le photon sort latéralement ou

Fin

Calcul de la nouvelle position du photon

Incrémentation du nombre d' interactions

le maximum d'interactions est atteint

Géneration aléatoire du type dl'interaction

O NEffet Rayleigh

O NEffet Compton

et calcul de la nouvelle énergie et direction du photon

et calcul de la nouvelle direction du photon

Absorption du photon

Effet Photoélectrique oucréation de paire

µ

NO

( Klein-Nishina )

et calcul du libre parcours moyen

Génération aléatoire des déviations angulaires

Génération aléatoire des déviations angulaires

Figure 6 : Algorithme de la trace du photon dans un milieu quelconque (écran, détecteur, bloc collimateur)

Le principe du calcul repose sur le fait qu’entre sa naissance et sa disparition, un photon

subit une succession de processus aléatoires : - Après avoir généré aléatoirement la direction d'émission et l'énergie du photon, on

détermine ses coordonnées d'impact sur le premier écran, puis son libre parcours moyen. - On tire ensuite un libre parcours et on détermine la nouvelle position du photon. - Si le photon n'a pas quitté l'écran, on tire aléatoirement le type d'interaction.

128

- Si le photon n'a pas été absorbé, suivant le type d'interaction, on tire aléatoirement ses déviations angulaires et on en déduit sa nouvelle direction et énergie.

- Tant que l'énergie et le nombre d'interactions sont inférieurs à un seuil défini par l'utilisateur, la simulation se poursuit comme ci-dessus par le tirage au hasard d'un nouveau libre parcours.

- Ce processus itératif se répète dans les différents écrans, et, s'il y a lieu, dans le milieu collimateur puis le détecteur, pendant toute la durée de vie du photon.

Le seuil en nombre d'interactions et en énergie ne s'appliquent qu'aux écrans. Lorsque le photon interagi avec le milieu détecteur en y laissant de l'énergie, on

incrémente d'une unité la classe correspondant à l'énergie totale cédée.

Résultats fournis Le logiciel fournit les résultats suivants : – le spectre énergétique des photons incidents au détecteur, – le spectre énergétique des photons détectés, – le spectre physiquement détecté par la chaîne de mesure :

La résolution en énergie du spectre simulé est définie par la fenêtre, exprimée en keV par canal, fixée par l'utilisateur. En pratique, pour des photons mono-énergétiques déposant la totalité de leur énergie E dans le cristal, l'amplitude du signal résultant sera distribuée suivant une loi Gaussienne autour d'une amplitude moyenne Ho, avec un écart-type σ proportionnel à √E. Ces fluctuations sont dues aux variations statistiques du nombre d'électrons parvenant à l'anode du photomultiplicateur.

Les différentes causes de ces fluctuations sont décrites dans [7] et [8]. La résolution en énergie du détecteur étant définie par :

Res = FWHM / Ho,

où FWHM = 2.35 σ désigne la largeur à mi-hauteur de la Gaussienne, on aura :

Res = K / √E.

Pour un type de scintillateur donné, l'efficacité lumineuse (et par conséquent la résolution en énergie) dépend de facteurs tels que : la pureté de la matrice, ses dimensions, la température, le convertisseur (photomultiplicateur, photodiode). A notre connaissance, il n'existe aucune formule générale qui en tienne compte.

Le logiciel fournit également :

– le taux de comptage total : Si Ne est le nombre de photons émis au cours de la simulation, Nd, le nombre de photons détectés et No le nombre total de photons émis par la source par unité de temps, le taux de comptage sera : R = No * Nd / Ne. Un photon est détecté lorsqu'il laisse une partie ou la totalité de son énergie dans le milieu détecteur,

– le rendement du détecteur (rapport du nombre de photons détectés et incidents), – le facteur de recouvrement en nombre (rapport du nombre total de photons incidents au

détecteur et du nombre de photons directs), – le pourcentage de photons ayant été détectés via le blindage, – le taux de comptage partiel dans une fenêtre d'énergie.

129

L'affichage des interactions avec les différents milieux et de l'énergie cédée au cristal par les photons peut être activé ou désactivé par l'utilisateur.

On peut à tout moment arrêter la simulation afin de visualiser les résultats puis la reprendre ensuite.

Les données d'entrée, les résultats de la simulation, et la graine aléatoire courante peuvent être sauvegardés sur fichier.

Ceci facilite leur utilisation ultérieure, soit pour :

– visualiser le spectre, – poursuivre la simulation, ou – modifier les résultats et réinitialiser les calculs.

Référence [1] F. TOLA

MACALU – Code Monte-Carlo de simulation du transport de photons γ ou X Note technique

[2] F. TOLA JANU - Système d'aide à la conception de jauges nucléaires par transmission d'un rayonnementβ, γ ou X Note technique

[3] I. LUX, L. KOBLINGER Monte-Carlo particle transport methods : Neutron and photon calculations CRC Press, 1991

[4] M. TERRISSOL Méthode de simulation du transport d'électrons d'énergies comprises entre 10 eV et 30 keV Thèse de docteur es-sciences physiques (Université Paul Sabatier de Toulouse), Octobre 1978

[5] W.H. PRESS, S.A. TEUKOLSKY, W.T. VETTERLING, B.P. FLANNERY Numerical recipes in C. Second Edition Cambridge University Press

[6] S.K. PARK, K.W. MILLER Random number generators: Good ones are hard to find

[7] GLENN F. KNOLL Radiation Detection and Measurement John Wiley & Sons, Second edition

[8] CRISMATEC Catalogue des Détecteurs à Scintillation, 1995

Documents

L’addition de traceurs dans les polymères : Une nouvelle voie ......M. Alain MILLE 69622 VILLEURBANNE Cedex M. Alain MILLE Université Claude Bernard Lyon 1 LIRIS - INFOMATHS Bâtiment