Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
METHOD AND APPARATUS FOR THE DISCRIMINATION AND IDENTIFICATION OF PLASTIC SAMPLES
Document Type and Number:
WIPO Patent Application WO/2018/025069
Kind Code:
A1
Abstract:
A method for discriminating different types of plastic materials, in particular dark non-reflecting plastics, comprising the steps of: (a) providing a substrate (10) with a composition comprising a plurality of metallic elements, each suitable for producing a respective line of fluorescence emission when subjected to excitation by means of an X-ray beam; (b) providing a plastic sample (P) to discriminate interposed between a source of said X-ray beam (2) and said substrate (10); (c) irradiating said sample (P) by means of an X-ray beam with continuous spectrum S(E) of energy E produced by said source (2) and, through said sample (P), said substrate (10) underlying the sample (P); (d) detecting a resulting radiation S 3 (E i ) including: - a fluorescence emission radiation I M F (E i ) at energy E i of each element (i) of said substrate (10) produced as a consequence of said step (c) of irradiating without said sample (P); (e) detecting a resulting radiation S 4 (E i ) including: - a fluorescence emission radiation I M D(E i ) at energy E i of each element (i) of said substrate (10) produced as a consequence of said step (c) of irradiating and attenuated through said sample (P), and - a Compton scattering radiation S 2 (E) produced by said sample (P) 25 produced as a consequence of said step (c) of irradiating; (f) calculating, for each element (i), a parameter qualifying the plastic sample (P) as a function of such fluorescence emission radiation I M F (E i ), I M D(E i ) and Compton scattering radiation S 2 (E): αM (Ei), Ri P/C.

Inventors:
PACELLA, Danilo (l'energia e lo sviluppo economico sostenibile Centro Ricerche Frascati,Laboratorio di Fisica della Fusione a Confinamento Magnetico, Via Enrico Fermi 45 Frascati RM, 00044, IT)
CAUSA, Federica (l'energia e lo sviluppo economico sostenibile Centro Ricerche Frascati,Laboratorio di Fisica della Fusione a Confinamento Magnetico, Via Enrico Fermi 45 Frascati RM, 00044, IT)
CLAPS, Gerardo (l'energia e lo sviluppo economico sostenibile Centro Ricerche Frascati,Laboratorio di Fisica della Fusione a Confinamento Magnetico, Via Enrico Fermi 45 Frascati RM, 00044, IT)
GABELLIERI, Lori (l'energia e lo sviluppo economico sostenibile Centro Ricerche Frascati,Laboratorio di Fisica della Fusione a Confinamento Magnetico, Via Enrico Fermi 45 Frascati RM, 00044, IT)
ROMANO, Afra (l'energia e lo sviluppo economico sostenibile Centro Ricerche Frascati,Laboratorio di Fisica della Fusione a Confinamento Magnetico, Via Enrico Fermi 45 Frascati RM, 00044, IT)
TUCCILLO, Angelo Antonio (l'energia e lo sviluppo economico sostenibile Centro Ricerche Frascati,Laboratorio di Fisica della Fusione a Confinamento Magnetico, Via Enrico Fermi 45 Frascati RM, 00044, IT)
Application Number:
IB2016/054737
Publication Date:
February 08, 2018
Filing Date:
August 05, 2016
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
AGENZIA NAZIONALE PER LE NUOVE TECNOLOGIE, L'ENERGIA E LO SVILUPPO ECONOMICO SOSTENIBILE (ENEA) (Lungotevere Thaon di Revel 76, Roma RM, 00196, IT)
B.M.C.R. S.R.L. (Viale Roma 66, Civitella di Romagna FC, 47012, IT)
International Classes:
G01N23/22; B29B17/02
Attorney, Agent or Firm:
PAPA, Elisabetta et al. (Piazza di Pietra 39, Rome, 00186, IT)
Download PDF:
Claims:
CLAIMS

1. A method for discriminating different types of plastic materials, in particular black, dark non-reflecting plastics, comprising the steps of:

(a) providing a substrate (10) with a composition comprising a plurality of elements, preferably metallic elements, each suitable for producing a respective line of fluorescence emission when subjected to excitation by means of an X-ray beam;

(b) providing a plastic sample (P) to discriminate interposed between a source of said X-ray beam (2) and said substrate (10);

(c) irradiating said sample (P) by means of an X-ray beam with a continuous spectrum S(E) of energy E produced by said source (2) and, through said sample, said substrate (10) underlying the sample (P);

(d) detecting a resulting radiation S^E;) including:

- a fluorescence emission radiation !MD(E;) of energy E; of each element (/) of said substrate (10) produced as a consequence of said step (c) of irradiating and attenuated through said sampie (P), and

- a Compton scattering radiation S2(E) produced by said sampie (P) as a consequence of said step (c) of irradiating;

(e) irradiating said substrate (10) and detecting a base fluorescence emission radiation lF of each element (/) of said substrate in absence of said sample (P) between said substrate (10) and said source of X-rays;

(f) detecting a resulting radiation S3(Ei) including a fluorescence emission radiation IMF(EI) of energy Ε,· of each element (/) of said substrate (10) produced as a consequence of said step (e);

(g) calculating, for each element ( ), at least one parameter qualifying the plastic sample (P) as a function of the ratio between such fluorescence emission radiation lMo(Ei) and iMF(Ej) or as ratio between ΙΜο(Εΐ) and the Compton scattering radiation 82(E).

2. The method according to claim 1 , wherein a collimator is put at the exit of the X-ray tube to provide a spot (order of millimeters) on said sample (P) smaller than the sample.

3. The method according to claim 1 or 2, comprising a further step of measuring a thickness (f) of said sample in an emission direction (/3), in particular prior, or subsequent, to said step (c) of irradiating said sample.

4. The method according to any of the preceding claims, wherein said excitation X-ray beam is a continuous beam, in particular a beam with energy emitted in a range of about 1 -35 KeV, in particular about 1 -25 KeV. 5. The method according to any one of the preceding claims, wherein said excitation X-ray beam is incident (direction l2) on said substrate (10) with an angle with respect to the direction !3 of the detector, included in a range 45-90 degrees.

6. The method according to any one of the preceding claims, wherein said source (2) of X-ray beam is positioned about 10 cm above said sample (P).

7. The method according to any of the preceding claims, wherein said substrate (10) has a composition comprising one or more elements selected from a group comprising: Iron, Lead, Molybdenum, Copper, Mn, Ni, Zn, Ag, Au and/or other materials characterized by emission lines in an energy range of about 2-30 keV. 8. The method according to any of the preceding claims, wherein said substrate (10) has a composition at least comprising two elements having emission fluorescence energy upon X-ray irradiation in a ratio of ½.

9. The method according to any one of the preceding claims, wherein said sample (P) is in contact with, and in particular resting on, said substrate (10). 10. The method according to any one of claims 1 to 8, wherein said sample (P) is spaced from said substrate (10) and without direct contact with it.

11. The method according to any one of the preceding claims, wherein said sample (P) has a substantially planar external surface p, within the X-ray spot, facing said source (2). 12. The method according to any one of the preceding claims, wherein said sample has a thickness in a range of about 0.5-10 mm, in particular about 1 -5 mm.

13. The method according to any one of the preceding claims, wherein an angle (δ) defined between a detection axis (/3) and a radiation axis (l2) of said source (2) is comprised in a range of about 0-25 degrees. 14. The method according to any one of the preceding claims, wherein a detection axis (/3) is arranged substantially perpendicular to said substrate (10).

15. The method according to any one of the preceding claims, which uses a Silicon Drift Detector.

16. The method according to any one of the preceding claims, wherein said parameter is an effective linear calculated as:

wherein

wherein t is the material sample thickness, I F,i is the is the experimental value of the intensity of the i-th fluorescence line of the substrate (without plastic material sample) calculated using the detected spectrum Ss(E) obtained without sample; (Eu, ERi) is a region of interest centred at the peak energy E,- of the i-th fluorescence line; lMDj is the experimental value of the intensity of the i-th fluorescence line calculated using the detected spectrum S4(E) obtained with the sample; B is the background intensity underneath the Line Ej , using the detected spectrum S4(E) obtained with the sample. 17. The method according to any one of the preceding claims, wherein said parameter is a peak-to-continuum level ratio defined as:

wherein lMD i is defined in claim 16 and

wherein C, is the continuum level calculated over a region of interest (ELc, ERC) which is sufficiently broad to include a large portion of Compton scattered spectrum, but without including discrete lines arising either from the metallic back or from additives in the plastic material sample

18. The method according to any one of the preceding claims, which is a method of discriminating dark, i.e. non-reflecting, plastics.

19. The method according to any one of the preceding claims, which is a method of selection of plastics, for quality control in the industrial process, or for differentiated recycling in case of waste plastics.

20. Apparatus (1 ) configured for performing the method according to any of the preceding claims, comprising:

- an X-ray source (2) adapted to emit an X-ray beam with a continuous spectrum, preferably with energy emitted in a range 1 -35 keV;

- a collimator to be applied at the detector to adjust the spot size as necessary; - a substrate (10) with a composition comprising a plurality of elements, in particular a plurality of metal elements, each capable of producing a respective line of the emission fluorescence when subjected to excitation by means of an X-ray beam;

- a spectroscopic detector (3) configured to detect said fluorescence emission lines and continuous spectrum of scattered photons;

- a processing unit (50), wherein the detector is configured to detect the following resulting radiation SsfE/ including: - a fluorescence emission radiation I F (Ej) at energy E/ of each element ( ) of said substrate (10) produced as a consequence of irradiation by the X-ray beam without said sample (P), and and, S4(Ei) including:

- a fluorescence emission radiation // ?a,fE/) at energy E,- of each element ( ) of said substrate (10) produced as a consequence of irradiation by the X-ray beam and attenuated through said sample (P), and

- a Compton scattering radiation S2(E) produced by said sample (P) produced as a consequence of said step (c) of radiating; and wherein said processing unit (50) is configured to calculate, for each element (/'), two parameters qualifying the plastic sample (P), as a function of such fluorescence emission radiation lMDji and Compton scattering radiation S2(E) : a" (Ej) , R' p/c

Description:
METHOD AND APPARATUS FOR THE DISCRIMINATION AND

IDENTIFICATION OF PLASTIC SAMPLES

DESCRIPTION

FIELD OF THE INVENTION

The present invention relates to a method for the discrimination and identification of plastic samples and to an apparatus for performing such method.

BACKGROUND OF THE INVENTION

Given the wide diffusion of objects made of plastics, the problem of recycling this type of materials is particularly critical. Such problem is closely related to that of automatically identifying the type of plastic for quality control for production or a proper delivery in the recycling lines.

Several methods have been proposed in the state of the art for such identification. However, the known methods do not prove truly effective in discriminating black, heavy coloured or non-reflective plastic materials (polymers).

X-ray techniques are not typically used for identification of plastics. In fact, plastic materials are formed by polymeric chains of low-Z elements, including, primarily, carbon (C), hydrogen (H) and oxygen (O). Therefore, under X-ray irradiation, plastic materials do not fluoresce, and for this reason plastic materials are typically referred to as 'dark matrices'. Consequently, X-ray Fluorescence (XRF) cannot be utilised to identify and discriminate plastic materials because such materials do not fluoresce in the X-ray region (Energy > 1 KeV).

Similarly, X-ray Transmission (XRT) is not effective in identifying and discriminating polymers because the differences between polymers - e.g. Polypropylene (PP), Polyethylene (PE), Polystyrene (PS), Polycarbonates (PC) - in terms of mass attenuation coefficients and densities are limited, typically within a 5-10% range, which is comparable with the experimental error. The above limits of the known art are particularly significant in case of black / non-reflecting, plastics. In Midgley et al ("Measurements of the X-ray linear attenuation coefficient for low atomic number materials at energies 32-66 and 140keV", Radiation Physics And Chemistry, Pergamon, Amsterdam, NL, vol. 72, no. 4, 1 March 2005, pages 525-535 ), the scheme described in the paper is a typical configuration of X-ray Transmission (XRT), used to measure the linear attenuation coefficient μ, for a few materials. In this technique a monocromatic X-ray source is used, with the detector on the other side of the sample, exploiting the attenuation of the X- ray beam. Conversely, the technique proposed in the present disclosure (X- SETA) uses a continuous broad X-ray spectrum, the detector is placed on the same side of the source and it exploits not only the attenuation of the X-ray source but also the attenuation of the X-ray lines produced by the substrate, whose excitation rates depends in turn on the absorption in the sample, and it exploits also the Compton scattered photons.

Due to these conceptual and practical differences, X-SETA is extremely more sensitive that XRT to the plastic materials. The measured values of mass attenuation coefficients of PE and PA reported in the paper differ of about less than 4%, while the so called effective mass attenuation coefficients measured with the technique X-SETA differ of almost 40%

Midgley ("Materials analysis using X-ray linear attenuation coefficient measurements at four photon energies; X-ray absorptiometry using measurements at four energies", Physics in Medicine And Biology, Institute of Physics Publishing, Bristol, GB, vol. 50, no. 17, 7 September 2005, pages 4139- 4157) provides a theoretical paper concerning the linear attenuation coefficient μ in case of a mixture of many elements. In general μ is a complicated function of the different atomic species Z, , via the individual cross sections of the physical processes, as function of photon energy E. The paper proposes a factorization, whose dependence on Z is given by a polynomial base expansion, whose coefficients depend on E. This description allows the reduction to a matricial form and consequently application of standard techniques like SVD. Sitko et al ("Quantitative X-ray fluorescence analysis of samples of less than 'infinite thickness': Difficulties and possibilities" , Spectrochimica Acta, Part B: Atomic Spectroscopy, New York, US, vol. 64, no. 1 1 -12 , 1 November 2009, pages 1 161 -1 172) reviews different techniques, based on X-ray fluorescence (XRF), for quantitative analysis of atoms present in a material. While recognition of an atomic specie is easy, assessment of its concentration in the material is more difficult, due to many effects: finite thickness, absorption and propagation of the stimulating X-ray spectrum, double fluorescence in case of other atomic species and so on. The X-SETA method does not exploit X-ray fluorescence for the plastic discrimination, because polymers are made of light Z elements and they do not emit in X-ray band (1 -35 keV). X-SETA uses fluorescence only to assess the additives, in case they are added to the polymer. US6018562 concerns a simplified tomographic reconstruction of objects, based on 1 -D radiographic scans in different X-ray energy intervals, for security analysis of luggage. Due to the energy discrimination, the system allows the coarse recognition of different materials (metals, organic, rubber, tissues and so on). Since this patent it is based on XRT, like the technique discussed in Midgley et al, its sensitivity is not enough to recognize the nature of polymers, like X-SETA does.

SUMMARY OF THE INVENTION

The technical problem underlying the present invention is therefore to overcome the drawbacks mentioned above with reference to the state of the art.

The above problem is solved by a method according to claim 1 and by an apparatus according to claim 20.

Preferred features of the invention are object of the dependent claims.

The method of the invention is an X-ray Spectral Emission technique combining Transmission and Absorption of continuous and line spectra (X-SETA). The X- SETA method is based on low-energy X-ray irradiation of the plastic materials to be identified and on the detection of a re-emitted and/or scattered radiation.

Differently from XRT techniques, in the apparatus and method of the invention an X-ray generator (in particular an X-ray tube) and a detector are on the same side of the plastic sample. Behind the sample, at the opposite part with respect to the detector and generator placement, a metallic substrate, or back, is arranged.

The collected radiation provides information, particularly via a series of measurements and data elaboration, on the type of polymer and, preferably, the type of filler(s) and/or additive(s) present in the polymer matrix.

The method and apparatus of the invention allow including and considering, in the detected radiation, the contribution of Compton scattering and to amplify the small differences in density and response to photoelectric effect of (dark) plastic materials.

The X-ray range of generated energy is typically, but not exclusively, from 1 keV up to 35 keV.

Therefore, the method and the apparatus of the invention allow an effective discrimination of different types of plastic materials, including the identification of dark (i.e. non-reflecting) plastics.

Other advantages, features and use modes of the present invention will result evident from the following detailed description of some embodiments, provided by way of example and not with limitative purpose.

BRIEF DESCRIPTION OF THE DRAWINGS

Reference will be made to the figures of the annexed drawings, wherein:

- Figure 1 shows a diagram, known per se, relating to Carbon mass attenuation coefficient plotted as a function of photon energy [REF:

Photon-based medical imagery, John Wileyhttp://wiley-vch.e- bookshelf.de/products/reading-epub/product-id/754443/title/P hoton- based%2BMedical%2Blmagery.html];

- Figure 2 shows a schematic representation of a preferred embodiment of the apparatus of the invention implementing the X-SETA method, which apparatus and method use, in the example shown, a two-part composite back or substrate;

Figure 3 shows an exemplary experimental X-ray spectra, arriving on the substrate, obtained by the apparatus and method of Figure 2, with and without a sample (the former being, in this representation, a plate of PE of about 3 mm thickness);

Figure 4 shows exemplary excitation rates of Cu K a line of the substrate, obtained by the apparatus and method of Figure 2, with and without sample (the former being a plate of PE of about 3 mm thickness);

Figure 5 shows an exemplary experimental spectrum obtained by the apparatus and method of Figure 2, without sample and with a sample of PE (of about 2.6 mm thickness) with a Cu-Mo metallic back;

Figure 6 shows an enlarged portion of the exemplary experimental spectrum of Figure 5, wherein contribution of Compton scattering is highlighted;

Figure 7 shows exemplary diagrams relating to selected parameters calculated by the apparatus and method of the invention in one specific application, namely: (a) a(Cu), (b) a(Mo), (c) R P/C (Cu) and (d) R P/ c(Mo), as functions of sample thickness (t) for the following pure polymers without additives: polypropylene (PP), polyethylene (PE), polystyrene (PS), acrylonitrile butadiene styrene (ABS), polycarbonate (PC), polyamide (PA) and for a PC (70%)-ABS (30%) blend;

Figure 8 shows the apparatus of Figure 2, emphasizing how the X-SETA arrangement is able to detect also X-ray fluorescence produced by additives or fillers included in a plastic material sample;

Figure 9 shows exemplary diagrams relating to experimental data, representing in particular lines of fluorescence for plastics with different filler and/or additives;

Figure 10 shows exemplary diagrams relating to selected parameters calculated by the apparatus and method of the invention in one specific application, namely a(Mo) curves obtained when analysing plastics with filler or dense plastics;

Figure 1 1 shows exemplary a diagram relating to selected parameters calculated by the apparatus and method of the invention in one specific application, in particular an effect of corrections on a(Cu) parameters obtained for plastics with additives and normalized to the value of the corresponding polymers without additives, such as for example colorants, with and without corrections. DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS OF THE INVENTION

X-ray interaction with plastic materials

X-ray radiation is absorbed in the material principally via photoelectric effect involving core electrons and via Compton and Rayleigh (elastic) scattering. The probability of photoelectric absorption, r, depends on the atomic charge Z of the element and the energy E of the photon: τ = const Z n E ~3 , where const is a constant, 3<n<4 and E is greater than the binding energy of the absorbing electron. The probability of Compton scattering is almost independent of Z and decreases as the photon energy increases. The energy E' of Compton scattered photons is

1

E' = E -

1 + * (1 - cos fl) where E is the energy of the incident X-ray photon, m 0 is the electron mass, c is the speed of light, such that m 0 c 2 = 0.5 MeV, and Θ is the scattering angle. Therefore, the energy lost in the process is

¾1 - cos 0)

ΔΕ = E - l + -¾l - cos 0)

Typically, the energy lost in the Compton scattering of 15 keV photons at 6* = 45 degrees is approximately 1 %, corresponding to 130 eV, while at larger angles it increases, for example if 0 = 180 degrees it is approximately 6%, corresponding to 500 eV. Compton scattering occurs at all angles.

The other relevant type of interaction of X-rays with matter is Rayleigh scattering, mainly in the forward direction, which represents a (minor) loss term in the aforementioned configuration. Differently from Compton scattering, in Rayleigh scattering only the direction of the scattered photon is altered, but not its energy (elastic scattering). In Figure 1 the linear attenuation coefficient for Carbon, one of the major elements in plastics, is plotted as a function of the photon energy, where the contribution of the individual processes are also indicated, namely, photoelectric effect (p.e.), Compton scattering, Rayleigh (elastic) scattering, pair production (significant for photon energies >10MeV).

Preferred embodiments of the invention

Figure 2 shows a schematic representation of an apparatus, or experimental set-up, according to a preferred embodiment of the invention, which apparatus is globally denoted by 1 .

Apparatus 1 comprises, in a basic arrangement:

- an X-ray generator 2, in particular an X-ray tube;

- a spectroscopic detector 3, for example a so-called Silicon Drift Detector (SDD);

- a collimator (which is not represented and it is included in generator 2);

- a substrate or supporting metallic material 10 emitting fluorescence upon X-ray irradiation, which may also be referred to as 'back'.

In the present application, the back 10 is a slab, in particular a composite metallic slab. Preferably, the latter is made of portions of different metals, or composed of different metal particles or nano-particles embedded in a suitable matrix. In the example of Figure 2, a substrate 10 made of a first and a second part, 101 and 102 respectively, made of different metals and arranged side by side or integral one to the other, is represented. Important is that the X-ray spot on the substrate could excite all the metallic species forming the substrate

Preferably, substrate 10 comprises at least two elements having emission fluorescence energy upon X-ray irradiation in a ratio of ½.

Still in Figure 2, the sample plastic material to be identified is denoted by P.

As shown, sample P is interposed between generator 2 and substrate 10, so that sample P can be irradiated by means of an X-ray beam S(E) of energy E produced by generator 2 and, through the sample P, the substrate 10 underlying it is irradiated as well.

X-ray tube requires a collimator to define a small focal spot, which has to be smaller than the sample, and the spot has to fall inside the sample. The size of the spot (order of millimeters) can be adjusted to ensure that the irradiated surface of the sample is flat (thickness "t" constant) within the spot size.

Axis l 2 is defined by the line connecting the output window of the X-ray tube and the center of the X-ray spot on the substrate. Axis l 3 is defined by the line connecting the entrance window of the spectroscopic detector and the center of the X-ray spot on the substrate. Angle δ, defined between the detection axis l 3 and the radiation axis l 2 , is preferably comprised in a range of 0-25 degrees.

Preferably, detection axis i 3 is arranged perpendicularly to the surface of the substrate

Apparatus 1 further includes, or is in communication with, a processing unit 50 for calculating parameters allowing the identification, or discrimination, of plastic materials based upon measurements of detector 3. Such parameters are discussed below.

The quantities involved and indicated in Figure 2 are defined in the following Table 1.

Table 1

Symbol Description

S(E) Incident X-ray spectrum from the X-ray tube carrying energy E.

Si(E) Spectrum of X-ray radiation reaching the lower surface of the

plastic material sample, attenuated by the plastic material sample itself and arriving on the substrate.

S 2 (E) X-ray spectrum scattered by Compton effect from incident

continuum spectrum S(E).

IF X-ray fluorescence line emitted by the back excited by S(E),

without sample (not shown in Figure 2).

IE X-ray fluorescence line emitted by the back excited by S^E), in the presence of the plastic material sample.

ID X-ray fluorescence line detected after attenuation through the

sample.

S 3 (E) X-ray spectrum reaching the detector, without sample (not shown in Figure 2) including fluorescence lines (IF)-

S 4 (E) X-ray spectrum reaching the detector and including both Compton scattering and fluorescence lines (/ D ) attenuated by the plastic material sample.

In the present preferred configuration of the apparatus, the detected spectrum S 4 (E) is composed of all the X-ray lines emitted by the back and Compton scattered photons, with moderate energy shift (continuous spectrum).

With reference to the representation of Figure 1 , in the range of interest (photon energy < 35 kV) the pair production contribution is zero, Rayleigh scattering is negligible (less than 1 % of p.e.) and Compton scattering is smaller than p.e. but not negligible.

Referring to Figure 2, the continuum spectrum emitted by generator 2 is denoted by S(E). The spectrum reaching the lower surface of the sample and impinging on the substrate is S(E) attenuated by the first pass in the plastic material sample, due to p.e. absorption and Compton scattering, according to: S 1 = S E)e wherein t is the plastic material sample thickness (typically measured in mm), δ is the inclination angle of the source with respect to the vertical direction, and μ(Ε) is the linear attenuation coefficient of the plastic material sample, equal to the mass attenuation coefficient (usually indicated in literature as μ/ρ), multiplied by the plastic material sample density p.

Figure 3 shows typical spectrum S(E) emitted by the X-ray tube (without sample) and a corresponding modified spectrum attenuated by the first pass in a plastic material sample S-i(E) (in the case represented, the material is PE with thickness of about 3 mm).

In the chosen configuration, spectrum Si(E) excites the fluorescence of the (composite) metallic back, thus producing a spectrum of discrete lines characteristic of the metallic back material(s). The excitation rate of a fluorescence line of energy E,- is null for photon energies below the ionization energy Ε,·, and cE ~3 for energies above the ionization energy, where c is a material-dependent constant and E is the energy of the incident X-ray photon. As an example, the excitation rates of the K a line of Copper (Cu) obtained upon irradiation with the incident spectra of Figure 3 are plotted in Figure 4 for the cases of excitation without sample and with the aforementioned material plastic sample respectively. From a theoretical point of view, any of the discrete fluorescence lines l F of the bare metallic back (without plastic material sample) can be calculated as follows where C,- is a normalisation constant, Ε,· is the ionisation energy of element / (shell K of the metals forming the back) and E M is the maximum energy of the spectrum S(E).

In the presence of a plastic material sample between the X-ray source and the metallic back, the intensity of any fluorescence line is different because the excitation spectrum is different, in this case the excitation spectrum is Si(E) instead of S(E). The intensity of the i-th fluorescence line, as emitted by the substrate, is calculated as follows:

In this case the intensity l E of the i-th line carries information on the nature of the plastic sample via μ(Ε) and takes into account all relevant (photoelectric effect and Compton scattering) effects. The ratio of the intensities of the metallic back fluorescence lines obtained with and without the plastic material sample, IE /IF , depends on the effects of the first pass through the plastic material sample.

With the plastic material sample placed on the metallic back, the emitted line l E ,i is attenuated by the sample (second pass through the sample) and the detected line can be calculated as

D,i A suitable parameter combining these variables, in analogy with the linear attenuation coefficient, could b

This parameter summarizes the effects of first and second pass through the sample. In reality these variables l D:i and l F cannot be calculated because the function μ(Ε) is unknown. Since our goal is not the measurement of μ(Ε) but simply the discrimination of different kind of plastics, we will define appropriate experimental parameters for this purpose, in analogy to the given theoretical description.

These measured parameters, defined below, will be indicated with an upper index M ("measured")

With this aim in mind, a region of interest (E Li , E Ri ) is defined centred at the peak energy of the i-th fluorescence line, Ε,·. Then, the experimental value of the intensity of the i-th fluorescence line of the bare metallic back (without plastic material sample) is:

(1 ) where, in this case, S 3 (E) is the spectrum detected by the detector (SDD) after irradiation of the bare metallic back (without plastic material sample) with the incident spectrum from the source S(E).

When the plastic material sample is placed on the metallic back, the spectrum S 4 (E) reaching the detector includes the discrete fluorescence line(s) correspondingly attenuated by the plastic material sample, as well as the continuous Compton scattering spectrum due to first and second pass through the plastic material sample. It is then necessary to evaluate experimentally the intensity of the i-th fluorescence line of the back modified by the presence of the plastic material sample. The measurement is taken over the same region of interest (E u , E Ri ) defined above, and is defined as:

(2) where S 4 (E) is the spectrum detected by the SDD after irradiation of the plastic material sample placed on the metallic back with the incident spectrum from the source S(E) and B is the background below the line at Ej, estimated as

In the presence of the plastic material sample, B cannot be neglected, as in the case without sample, because of the background due to Compton scattering (see Figures 5 and 6). I D,i(E) carries information on the polymer via all the physical processes described above that occur in the first and second pass through the plastic material sample.

Then, the first set of experimental parameters used to discriminate plastic materials is defined as:

(3) at all energies Ε,· of the substrate, where I F>j and / ¾ / are readily measured experimentally as defined above in equations (1 ) and (2), before and after placing the plastic material sample on the metallic back, respectively, and the plastic material thickness t is measured using, for example, a caliper before placing the plastic material sample on the back.

The parameter defined in equation (3) above is referred to here as effective linear attenuation coefficient because its definition is similar (same dimensions) to that of the linear attenuation coefficient of any material and it is measured in cm ~

The use of the effective linear attenuation coefficient makes it possible to eliminate the dependence on the X-ray tube parameters (current, voltage, collimator size) parameters of the detector (active area, measurement time, detection efficiency), geometrical parameters (distances of tube and detector from the sample) and absorption in air. By using the ratio (l M Fj/l M D,i), the instrument is always self-calibrated. The calibration measurement l M F j has to be done only when the parameters of the measurements are changed.

In conclusion, the ^ parameter defined in equation (3), contains information on the absorption due to photoelectric effect and Compton scattering occurring in the first and second pass in the material, and the fluorescence excitation of the metallic back. Differently from a traditional XRT experimental arrangement, where the detected beam is on the opposite side of the sample with respect to the X-ray source and then Compton scattered photons are mainly lost, in the proposed experimental arrangement the detector is not collimated and placed in the same side of the source, thus also Compton back scattered photons contribute to the spectrum, producing the background between X-ray lines. Compton scattering occurs mainly on the electron molecular orbitals and therefore is more sensitive to the nature of polymer. For example the chemical formula of Polypropylene (C2H 4 ) and Polyethylene (C3H6) is the same, so from the point of view of photoelectric effect they are identical, being also the density the same by 2-3 %. The only difference is in the spatial distribution of the polymeric chain and in the molecular orbitals. Compton scattering is sensitive to such differences. The second set of parameters used to identify and discriminate plastic materials is the peak-to-continuum level ratio defined as follows. Given the i-th discrete fluorescence line, the peak level is given by l M D j like in formula (2).

A further region of interest (E L c, E RC ) is defined to calculate the continuum level as

Q iH RC S 4 {E)dE- (4) i¾c~¾c) **L€ · where (E LC , E RC ) is sufficiently broad to include a large portion of Compton scattered spectrum (see for example Ccu and CM O in Figure 6), but without including discrete lines arising either from the metallic back or from additives in the plastic material sample (discussed later). The peak-to-continuum level ratio is defined as: (5)

I D depends on the X-ray absorption in the plastic via photoelectric effect and

Compton scattering in the first and second pass; C,- is due only to the Compton scattering S 2 (E) of the X-ray spectrum S(E) of the tube during first and second pass into the plastic material sample. The ratio R'p / c also is a self-calibrated parameter, once the geometry is fixed, that depends strongly on the Compton effect.

The proposed experimental configuration and the discrimination parameters , and R'p / c are more sensitive to the nature of the polymer than the traditional linear attenuation coefficient obtained using the XRT configuration. Further, the experimental configuration is simple and cost-effective. In fact, even if the fluorescence efficiency of metals is low (10 _3 -10 -4 ), the compactness of the experimental apparatus with reduced distances (order of centimeters) between material sample, source and detector and the back just behind the sample, makes it possible to excite the fluorescence lines of the back with sufficient intensity to be detectable even in the presence of the plastic material sample interposed between source and back. A plastic material sample of thickness typically < 10 mm, does not totally absorb the fluorescence lines of the back, on the contrary in the X-SETA experimental arrangement the resulting spectrum contains the signature of the polymer via the different processes that occur in the first and second pass through the plastic material. This system X-SETA is able to recognize plastics that have been classified before. Therefore it has to be trained, producing a data base of reference parameters of,- , R'P/C with suitable well known samples. Practical example of application (i)

Pure polymers - discrimination and identification

To show the effectiveness of the X-SETA method, plastic material samples of thickness 1 mm < t < 10 mm, are considered here. In this case, two fluorescence lines from the metallic back are sufficient for the analysis, therefore, a composite back made of Cu and Mo was chosen. Typical operational parameters are: a portable X-ray tube (50KV and 200 μΑ max) was operated at 25 kV, 90 μΑ, equipped with a collimator to produce a spot of 8 mm diameter on the substrate. A spectroscopic SDD detector was used with 1024 channels, low gain, live time acquisition 5 s and 3 following acquisitions repeated to reduce the effect of fluctuations.

The best choice of metallic backs, for identifying black plastics normally used in appliances, is Cu-Mo: one (Cu) more sensitive to plastics without filler, the other (Mo) to thick plastics or with filler. In the case of thin plastics (< 1 .5 mm) Scandium-Copper is preferable, while for thicker plastics or rubbers (> 8 mm) Lead-Palladium is a more suitable choice. In this latter case the Supply Voltage of the X-ray tube should be raised to 35 keV, to excite 21 .2 keV-K a line of Pd.

Typical fluorescence lines obtained with metallic backs include the energies listed in the following Table 2.

The spectrum of the back obtained without sample is shown in Figure 5 (red curve with circle). The only significant features appearing in the spectrum are the metallic back (Cu and Mo, in this example) K a and K p fluorescence lines. The spectrum detected with the sample (Polyethylene 2.6 mm thickness), shown in Figure 5 (blue curve with diamond), presents the Mo line slightly reduced and the Cu line strongly absorbed (reduced to 20% of the original intensity value) and the clearly visible continuum of the Compton scattering. The Compton scattering is shown also in Figure 6, same case of Figure 5, where the vertical scale has been expanded. Since the background is changing, depending on Compton scattering, the K a lines are always calculated with the background (B) subtraction, equation (2). The horizontal axis shows the channel numbers of the SDD spectrometer, equivalent to X-ray energy (the linearity between the two was previously verified) ranging from 1 .5 keV (channel 50) to 30 keV (channel 1024).

Following the definitions of equations (1 ) to (5), and assuming the use of a Cu- Mo composite metallic back, four experimental parameters are defined for the analysis of plastic material samples: a M (Cu), a M (Mo), R P c (Cu), R P/C (Mo). These four experimental parameters are plotted as function of thickness of the plastic sample, respectively in Figures 7(a-d) for the following pure polymers (without additives): polypropylene (PP), polyethylene (PE), polystyrene (PS), acrylonitrile butadiene styrene (ABS), polycarbonate (PC), Polyamide (PA) and for a blend PC (70%)-ABS (30%). Plastics with additives will be discussed in Section 6. From the data it is deduced that all pure polymers of interest can be discriminated using the a M (Cu) parameter, with a relative change between PP and PC of about 100%. The other parameter, a M (Mo), is less sensitive (total change of 50%) with this selection of materials. The parameters R P c (Cu) and Rp /C (Mo) are strong function of the thickness t, with a few degenerations of polymers, and, therefore, not critical in this case

However, in general, the abovementioned four parameters represent a set of redundant measurements, which improve the confidence in the plastic recognition process, where experimental error, fluctuations and the possible effect of additives are important, as discussed in Section 6.

The identification of an unknown piece of plastic requires first the measurement of its thickness t and the 4 values a M (Cu), a M (Mo), R P c (Cu), R P/C (Mo) and then the comparison with the reference curves of these parameters (see figures 7) for the classified families, minimizing the distance between the measured value and the reference one. Weight factors can be defined to quantify the different sensitivity and confidence of each of these 4 parameters, based on the training data set with certified samples. As an example, the analysis of the spectra of Figures 5 and 6, relative to a sample of PE (2.6 mm) shows that the intensity of the Cu line is reduced to 22% of the reference intensity (without sample), obtaining a value of a M (Cu) = 5.77 cm -1 , Mo line is reduced to 78%, giving a M (Mo) = 0.95 cm -1 . The peak to continuum level are R P C (Cu) = 4.9 and R P C (Mo) = 0.5 ; all these four values fall very close to the curves of the family PE, and then the plastic sample is identified as PE.

From Figure 7(a) it appears that the a M (Cu) curves corresponding to PE and PA are well separated, with many other curves identifying other polymers, in between. The range between these two values is broad: a M PE (Cu)=5.5 cm "1 and a M PA (Cu)=9.5 cm "1 .

The "effective mass attenuation coefficient', dependent only on the nature of the material, is obtained by dividing a M by the polymer density p: (a M /p) PE (Cu) = 5.94 cm 2 /g and (a M /p ) PA (Cu) = 8.28 cm 2 /g. The a/p parameter varies by 39%, well beyond the experimental error of 4%, making it possible to discriminate also of the plastic categories in between.

With a standard XRT technique, values commonly discussed in the literature indicate that the corresponding range of mass attenuation coefficients, is only 3.8 %, which is lower than the experimental error. This is the reason why the XRT technique is not effective in discriminating polymers.

The technique X-SETA has in addition the advantage to work efficiently with low energy X-ray photons, more sensitive to plastic materials.

Practical example of application (ii)

Plastic materials with additives (colouring, fluidifiers, fillers, flame- retardants, ...)

If the plastic material sample is composed of a mixture of pure polymer and additives (including, for example, colouring, fluidifiers, fillers and flame retardants) then the detected spectrum includes also the X-ray fluorescence of the medium-high Z atoms forming the additive or filler. The concept is presented in the schematic of Figure 8.

Such additives or fillers can be elements or chemicals used as catalysts for the polymerisation process, or to modify the mechanical and plastic characteristics of the material, such as, for example, calcium carbonate or glass fibre or talc, or as flame-retardants or colouring.

For any additive it is sufficient identify at least one atom emitting in the X-ray range as a 'marker' of the presence of that additive. For example, Zn for a fluidifier (Zn0 2 ), Ca for calcium carbonate (CaCOs), Si for fibre glass (S1O 2 ), Br for brominate flame retardants or P for phosphate flame retardants, Ti and Fe for example as colourings (ΤΊΟ 2 , FeO 2 ) and so on. Example of fluorescence spectra are shown in Figure 9, where the fluorescence peaks from the individual markers are clearly visible, together with the Cu and Mo lines of the substrate. On the other hand additives containing only low Z materials, not emitting in X- rays, like for example organic flame retardants, do not affect significantly the measured 4 parameters and therefore can be neglected.

It is important to note that the choice of the back slab should be restricted to materials that are not contained in additives used in plastics production, to avoid ambiguities in the resulting spectra, as the case of Cu and Mo.

In general the linear attenuation coefficient μ χ οί a plastic X made of polymer P and containing an additive A is given by

(6) In a similar way the presence in plastic material samples of such elements and composites "A" , containing medium-high X elements, affects also the effective linear attenuation coefficient a defined in equation (3) and the peak to continuum ratio R P/C. However, in practice it is not readily possible to extract the information related to individual additives from the measured parameters. On the contrary the presence of such additives has to be assessed independently by means of their fluorescence and then the parameters will be consequently corrected.

Two categories of additives are considered here.

The first including all cases in which the additive is a filler. Such materials are considered as categories of plastics per se, and are, therefore, characterised by specific values of .

The second category includes all other cases. In this latter case corrections are applied to the measured value of a in order to identify the type of polymer. The corrections are calculated as functions of the concentration of the specific additive(s) included in the plastic sample. The corrective factors are proportional to the measured intensities of the corresponding fluorescence line, and are applied to the and R P C parameters, to derive the contribution of the pure polymer only, in order to identify it. These corrective factors are derived, in a heuristic way, starting from samples with additives and/or fillers of known density.

The intensity l A of the X-ray line of energy E A emitted by the atoms A of an additive is given by:

JrEmax ft

dE άχΞ(Ε) βχν [(-μ Ρ ρ Ρ - μ Α ρ Α )χ] μ Α ΙΙ(Ε, χ) βχν [-μρ(Ε ί )ρρχ] Ε,. . JO where

{S(E)exp [(-μ Ρ ρ Ρ - μ Α ΡΑ)χ]}/Ε 2 - 75

R(E, x) = -

/ Emax dE

E k ,A E 2 - 75

l 0 is the total intensity of the spectrum S(E) of photons emitted by the X-ray tube and arriving on the sample (Figure 3), J, g, ω are parameters related to the specific X-ray transition of the given atom A, μ Ρ: p P are linear attenuation coefficient and density of the pure polymer, μ Α and p the same for the atom A of the additive. Integrations are carried out over the energy from the ionization energy E k,A of the shell K of the element A and the integral over the thickness t of the sample. We can see that intensity l A is function of t also. For low and medium Z elements (up to Zinc), the dependence from t saturates at about thickness of 1 .5 mm, so in these case the dependence by t can be neglected. Moreover the intensity l A depends explicitly over impurity density p A , but also through the integrals. Simulations and laboratory tests confirmed that for impurity concentration p A /pp up to about 1 %, the linear dependence l A =const * p A is valid. For high Z elements, such as Bromine, for example, the dependence on t also has to be considered and for concentrations higher than 1 % a parabolic dependence on p A should be used.

With sample having well known amount of additives (p A ), the constant can be derived by the measurement of l A . It allows therefore, for any sample, the estimation of p A from the measurement of the fluorescence line of the atom representative of a given additive, with the limitations discussed above.

The constant of the formula l A =const * p A varies significantly for the different markers, depending on the energy of the X-ray line, due to the fluorescence efficiency, the air absorption and the detection efficiency of the spectrometer. To detect lines of a few keV, such as Silicon and Phosphor, the detector should be placed close to the sample (by about 3 cm of distance) to minimize air absorption and collect more photons, due to the poor sensitivity at these low energies.

In analogy with formula 6, the measured effective linear attenuation of of a plastic formed by a polymer P and an additive A, in the linear regime l A =const*p A above described, can be written as: where k is a constant, to be assessed by means of certified polymers P with well known concentration of additive A. Once the values of k are known for the additives of interest, the measured value can be corrected to determine the corresponding polymer P. Same corrections can be applied to R P/C , even if these parameters are less sensitive to additives than a.

Practical example of application (iii)

Discrimination and identification of plastic materials: results based upon exemplary experimental data

The measurement of the fluorescence lines / D ,;from the back ranges from 98% to 2% of the corresponding line l M F without sample, providing a dynamic range for the lines of Cu and Mo of almost two orders of magnitude. The uncertainties, expressed as relative standard deviation, associated to the measurement of I M D arise due to three contributions:

1) fluctuations of the X-ray flux generated by the tube, estimated experimentally as: atube = 1 .3 %

2) fluctuations relative to the sampling of the measurement, estimated experimentally as: a sa mpiing = 3 %

3) fluctuations of density and composition of the sample, estimated experimentally as: a sam pie = 2 %

The lines of Cu and Mo are acquired for a few seconds without sample, resulting typically in 10000-30000 counts, with relative (poissonian) error < 1 %.

When the fluorescence line of the substrate is absorbed by the sample, the counts drops and the relative error increases. In this case, a sam piing = 3 % in the worst case, when the lines is strongly absorbed (4%). The combination of these three contributions provides a total standard deviation of <^TOT - 4%, which is smaller that the separation between adjacent plastic categories (~10%), obtained in the application described here.

Plastic materials analysed with the X-SETA method include certified samples and real plastic materials samples obtained from black RAEE wastes. The categories of plastics were:

1 ) certified samples:

• PP, PE, PS, ABS, PC-ABS, PA, PC

• PP with Calcium Carbonate, or with Talc, or with Fibre Glass

· Polymers with Fibre Glass: PP, ABS, PC-ABS, PA

• Brominated plastics

2) RAEE wastes:

• PP+EPDM , PMMA, POM

• PS-HIC, HIPS

· PS FR40, PS-PPE FR40, PC-ABS FR40

• PVC

In particular, 1 18 pieces were available of the second category of plastics. All were analysed with the FTIR method first, obtaining 102 pieces uniquely identified.

These samples were then analysed with the X-SETA method, achieving the success rates of the following Table 3.

Table 3

Polymer Number of % success rate

pieces

PP 4 100

PS 25 92

ABS 23 91

PC-ABS FR40 12 100

PS FR40 10 100

PPE+HIPS FR 40 10 90

PS-HIC 10 100

HIPS 15 100 The average success rate is approximately 95%.

In the case of plastics with filler (such as CaC0 3 , fibre glass, talc) the Cu line is totally absorbed. Therefore, in this case it is necessary to use only the Mo line, and the corresponding parameters a(Mo) and R P/c (Mo).

As an example, measured c^(Mo) curves are shown for plastics with filler (Fiber glass 20% or 30%) and for more dense and opaque polymers (such as PPE, POM, PP+EDPM) showing how sensitive this parameter is, when analysing this type of plastics.

The effect of corrections on the a M (Cu) parameter for additives is demonstrated in Figure 1 1. The experimental value of each polymer with additive is normalized to the value of the same polymers without additive. This is done for many plastic samples of different thickness and polymer families. The blue diamond markers represent the normalized a M (Cu) parameters, the red squares the corresponding corrected parameters. The corrections reduce the initial spread (60%) to an acceptable value (10%). This is comparable with the separation between close plastic families, but the redundancy of the four parameters makes it possible to identify plastics with a high confidence.

The present invention has been described so far with reference to preferred embodiments. It is intended that there may be other embodiments which refer to the same inventive concept, that may fall within the scope of the appended claims.