X-ray apparatus and relevant operating method for the analysis of plastics and other heterogeneous materials

Abstract

An X-ray apparatus for the analysis and recognition of black and colored plastics and heterogeneous materials includes an X-ray source (1) that produces a beam (3) of low-energy radiation that passes through a sample (6) to be analyzed and strikes an emitting element (4) made of a material with an emission line K or L in the range of 6 to 25keV, an X-ray spectrometer (2) is placed next to the X-ray source (1) and these components are close enough to each other and to the sample (6) to be able to perform a fluorescence radiation analysis (8), an analysis of the back-scattering radiation (7) by both Rayleigh and Compton effects, and an analysis of the transmission of the monochromatic radiation (5') emitted by the emitting element (4), using all these phenomena to identify plastics of all types and colors, with and without additives, and other heterogeneous materials such as wood, cardboard, glass, concrete, ceramics, chipboard, etc.

Description

[0001] X-RAY APPARATUS AND RELEVANT OPERATING METHOD FOR THE ANALYSIS OF PLASTICS AND OTHER HETEROGENEOUS MATERIALS

[0002] The present invention relates to an X-ray apparatus for the analysis of black and colored plastics and other heterogeneous materials, with particular application to the characterization and recovery of plastic materials from scrap by separation from other non-plastic materials.

[0003] As is well known, at the end of their lives, cars, and also other industrial and household products of essentially metallic composition, are ground with large hammer mills (so-called car shredders) that reduce them into pieces smaller than 150 mm in size so as to obtain scrap metal. On leaving these mills, the mixed ground material undergoes a deferrization action using large electromagnetic drums for the purpose of recovering and cleaning up the ferromagnetic steel. What is discarded by such electromagnetic drums consists mainly of plastics, rubbers, polyurethane foams, glass, aluminum, copper, zinc, zamak, lead, stainless steel, electrical wires, stone residues, iron oxides, and some ferromagnetic steel parts lost during the deferrization action.

[0004] Such a mixture of heterogeneous materials can also result from municipal solid waste and waste electrical / electronic equipment (WEEE), so there is a need to provide for the separation of unwanted materials within a given type of materials. In addition, the plastic materials to be separated may be of various types, so there is also a need to perform accurate characterization of the plastic piece (polymer, filler, additives, dyes...) so that the various types of plastics can be recycled separately, and this while separating them from any other type of material (e.g., leathers, organic material, wood).

[0005] It should be noted that the apparatus and method described here typically find application in the dynamic separation of pieces up to speeds compatible with industrial processes, with individual piece measurement times of 10-30 ms to achieve an identified material flow of several tons / hour. However, another possible application is in a laboratory instrument configuration, to make compact benchtop and portable instruments for static analysis of plastics with high accuracy in a measurement time on the order of several seconds and with an evaluation of the absolute concentration of fillers and additives. In the following, specific reference will be made to dynamic separation, but the explanations given are also applicable to static analysis.

[0006] Several techniques are known in the recycling field for separating materials from a scrap flow, and different types of separators use one or more techniques depending on the type of materials to be separated.

[0007] A first type of separator uses an X-ray technique, referred to as the X-Ray Fluorescence=XRF technique, which allows information on the chemical nature of a material to be gathered through a surface investigation, since the signal comes from a few hundred micrometers deep from the surface. Metals are selected using their characteristic fluorescence lines, but the distance between the analyzed sample and the X-ray source, which is at least 125 mm, precludes the detection of both low-energy lines and a back- scattering signal (so-called BS=B ack-scattering). XRF separators cannot reliably identify light metals such as aluminum and magnesium due to the speed of analysis that the application under consideration imposes (a few milliseconds per piece), since the fluorescence phenomenon is very weak for such light metals, while it is significantly more intense for heavier chemical elements (e.g., Ti and up).

[0008] Similarly, because plastics consist of polymer chains of low atomic weight Z elements, such as carbon, hydrogen, and oxygen, they do not emit fluorescence when subjected to X-rays, and thus are referred to as "dark matrices." The only case in which fluorescence radiation (XRF) of the analyzed plastic sample is obtained is if it contains elements (additives, dyes, etc.) that emit in the X-ray energy range and reach the detector. However, fluorescence alone indicates the presence of a chemical element but does not provide information about the type of material, so the XRF method by itself cannot be used to identify and discriminate polymeric plastics.

[0009] In the same way, the transmission method (XRT=X-Ray Transmission), which uses continuous X-ray spectra to be sent over the material to measure its attenuation, is not effective by itself in identifying and discriminating plastics because the differences in density and mass attenuation coefficient between polymers are typically less than 10%.

[0010] WO 2018 / 025069 describes a method and equipment for identifying plastics by a combination of fluorescence (XRF) and transmission (XRT) measurements of radiation reflected from a sample holder made of at least two metals. However, such a solution requires measurement times on the order of 5 to 20 seconds, thus being totally unsuitable for an industrial application for recycling plastic waste, being useful only as laboratory equipment.

[0011] Another technique used for alumina sorting is the Laser-Induced Breakdown Spectroscopy (LIBS) technique, which is an optical technique of analyzing recombination or fluorescence lines in the visible / near infrared emitted after surface ablation with a high- power laser. LIBS is a local analysis technique, i.e., it provides the composition of the sample only at the point of analysis, and it is micro-destructive because it alters (in the area of analysis) the material to be selected. LIBS has the disadvantage of being very expensive, cannot be used effectively with transparent materials because the laser beam passes through them without being absorbed, and also does not allow the analysis of many plastic materials because ablation triggers chemical oxidation reactions that irreparably compromise the information on the type of plastic.

[0012] A similar technique is that used in the olfactory method, in which the part is burned to recognize the polymer from the odors given off. However, this technique depends on the operator's skills and therefore does not guarantee accurate results, nor can it detect the presence of additives and pollutants. Simple manual sorting done by trained personnel, using visual and tactile recognition, also suffers from the same lack of reliability and is slow and inadequate for high volumes of material to be processed.

[0013] There are also wet or dry densimetric separation processes that allow the separation particularly of light polymers, such as PE and PP, from heavy plastics such as PA, PC, ABS, and PC-ABS and those with fillers, which are often not recycled because they are not separately recognized, but account for up to 40% of the total volume.

[0014] Alternatively, for quick identification on belt, the NIR (Near InfraRed) method is used, which is effective for colored plastics but does not work with black (non-reflective) plastics. This method is used to recognize the polymer, but it generally does not reveal the presence of additives (reinforcers or dyes) and pollutants, such as bromine.

[0015] It is therefore the purpose of the present invention to provide an apparatus and method of X-ray analysis that overcome the above drawbacks. This purpose is achieved by means of an apparatus and method that combine three different physical processes in the range of 2-50keV of low-energy X-ray radiation (so-called soft X-rays): X-ray back- scattering (XRBS), selective emission of fluorescence spectral lines (XRF) in case of impurities or additives present in the sample, and finally transmission and subsequent attenuation (XRT) of a monoenergetic line appropriately generated in the apparatus.

[0016] More specifically, the apparatus implementing the present method called FIXS (Fast Identification using X-ray Scattering) includes a low-energy X-ray source to irradiate the samples to be identified, a passive emitting element (called a top) placed on the opposite side of the sample from said X-ray source, and a spectrometer placed on the same side of the X-ray source with respect to the sample. The FIXS method consists of a spectral analysis part, a definition of the measured quantities part, and a decision part for sample recognition. Further advantages and features of this apparatus and its method are given in the dependent claims.

[0017] The fundamental advantage of the present apparatus and method is that it is applicable for fast and automatic analysis of both black and colored plastics, without requiring any pre-selection or pretreatment of the analyzed material, and simultaneously identifying possible heterogeneous materials (wood, cardboard, glass, concrete, ceramics, chipboard, ...) that may be found mixed together with plastics.

[0018] An additional advantage is the ability to derive from data processing information not only on the type of polymer, but also on the presence and type of fillers and / or additives present in the plastic matrix, with a quantitative assessment of the concentration.

[0019] Another important advantage of the above apparatus is its structural simplicity and low cost, since it does not require analyzing components (source, sensors) that are technically different from those used in conventional apparatuses, and costs substantially the same as they do while providing better performance.

[0020] Further advantages and features of the apparatus and method according to the present invention will be apparent to those skilled in the art from the following detailed description of an embodiment thereof with reference to the attached drawings, in which:

[0021] Fig.l is a view showing schematically the analysis section of an apparatus according to the invention;

[0022] Fig, 2 is a view showing schematically the energy bands of back-scattering radiation (XRBS) as a function of material depth;

[0023] Fig, 3 is a diagram of the X spectrum produced by air, i.e., in the absence of a sample; Fig, 4 is a diagram of the X spectrum produced by a 3.5-mm-thick ABS sample;

[0024] Fig, 5 is a detail of Fig.4 expanded in the vertical scale, with the energy integration bands shown with dashed vertical lines;

[0025] Fig, 6 is a diagram in arbitrary units of the intensity of the germanium line versus the integral of the P2 energy band, for some types of pure polymers;

[0026] Fig, 7 is a diagram similar to Fig.6, for the value of the integral of the P3 energy band with respect to that of P4;

[0027] Fig, 8 is a diagram in arbitrary units of the intensity of the germanium line versus the integral of the P4 energy band, for different categories of materials; and

[0028] Fig, 9 is a diagram similar to Fig.8, for the value of the integral of the P3 energy band with respect to that of P4.

[0029] Referring to Fig.1, it can be seen that a FIXS analysis apparatus according to the present invention can be made in a configuration with an X-ray source 1 with an anodecathode voltage between 20 and 60kV, preferably 20-40kV, and power between 2 and 50W, preferably between 5 and 10W, with a collimator producing a cone 3 of X-rays with an opening angle between 10° and 30°, preferably 15°. An X-ray spectrometer 2, typically an SDD (Silicon Drift Detector), with an active area between 10 and 50 mm2, preferably about 25 mm2(effective diameter about 5 mm), and a resolution in energy of at least 10%, preferably around 3-4%, is arranged adjacent to X-ray source 1 with an angle 9 between the longitudinal axis A of source 1 and the longitudinal axis A' of spectrometer 2 between 10 and 40°, preferably not more than 30° and more preferably not more than 20°.

[0030] An emitting element 4 (top) of a single suitable material with a diameter of 6 to 15 mm, preferably 8 mm, is placed on the opposite side of a sample 6 from the X-ray source 1, located on the A’ axis of detector 2 and preferably perpendicular to it. The top 4 is placed on a support, for example at the lower end of an aluminum rod (not shown in the figure) preferably less than 6 mm in diameter. Note that the combination of the aperture of the collimator and the diameter of top 4 must be such that beam 3 completely irradiates top 4, i.e., the surface of top 4 is included in the base of cone 3.

[0031] The material of top 4, typically a metal, has an emission line K or L in the range of 6 to 25keV; more specifically, its monochromatic line must be chosen so that its attenuation is not less than about 5% in the case of samples with the minimum thickness and not greater than about 95% in the case of maximum thicknesses. These limits are valid for fast automatic recognition whose measurement takes place in a few tens of milliseconds, while in the case of a benchtop instrument, whose measurement times are some seconds, these limits can be 2% and 98%, respectively. Otherwise, if the top 4 line were not attenuated for thin samples or were totally absorbed for thick samples, information would be lost and the range of evaluable material thickness would be narrowed.

[0032] In the case of recycled thermoplastics (WEEE, automotive), thicknesses vary from 1 to 5 mm; therefore, a line with energy around lOkeV meets the requirements. In this case, it is suggested to use germanium, which has an emission line Ka at 9.8keV and is a metal that easily lends itself to being attached to the support rod to make up top 4, and may also be good for wood, cardboard, glass, as long as they are not too thick. For thermoset rubbers, which have greater opacity than thermoplastic materials, as well as for thicker wood and glass, it is advisable to use cadmium (energy Ka « 23keV) as the material for top 4. In case of less opaque materials, metals with lower energy K or L emissions, such as Fe, Cu, Zn, S, Pb, can be chosen.

[0033] The distance between top 4 and the entrance window of detector 2 is between 20 mm and 80 mm, preferably about 30-40 mm in case the material to be analyzed is WEEE or automotive plastics, or 30-60 mm in case of mixed materials. In fact, by arranging the sliding or support plane of material 6 to be analyzed at 5-10 mm from the window of detector 2, top 4 must be at a distance from this plane greater than the thickness of material 6 to be analyzed. In the typical dynamic analysis application, material 6 is fed through a conveyor belt (not shown) followed by a chute provided with an opening, preferably about 20 mm long in the longitudinal direction of feeding and about 10 mm wide in the transverse direction, so as to use this opening for the passage of the X-ray beam 3 emitted by source 1 and of radiations 5’, 7, 8 resulting from top 4 and sample 6, as described in detail below.

[0034] Note that the chute is positioned no more than 10 mm from detector 2 in order not to lose the low-energy component of the back-scattering by the Rayleigh effect, and that the entire apparatus can be oriented horizontally, instead of vertically, in the case of vertically falling pieces. Furthermore, in the case of a laboratory instrument, it can be made either as in Figure 1 or in the inverted mode, that is, with top 4 at the bottom and source 1 and detector 2 at the top.

[0035] The operation of the apparatus according to the invention is easily understood in the light of the previous description, and the individual components XRT, XRBS, and possibly XRF, used in the FIXS method are determined and detected as follows, taking as a reference the X spectrum in the absence of sample shown in Fig.3, in which the emission line of germanium is visible while the contribution due to back-scattering produced by air is quite negligible.

[0036] XRT: The top 4, made of germanium in the example, is hit by the polychromatic radiation 3, emitted by source 1, and emits monochromatic radiation at energy EG6= 9.8keV in all directions, of which the portion of interest is the quasi-cylindrical beam of monochromatic radiation 5 that passes through sample 6 and reaches the active area of detector 2 below, which determines the diameter of beam 5 at the level of sample 6 that is about 6 mm, much smaller than the minimum dimension of sample 6, which is typically 30 to 120 mm or more. However, for samples with a smaller minimum dimension the active area of detector 2 and hence the diameter of beam 5 has to be reduced so that it does not exceed one third of said minimum dimension, thereby reducing the emitted X- ray flux, with the result that measurement times have to be proportionally extended.

[0037] The monochromatic radiation 5' that reaches detector 2 after passing through sample 6 makes it possible to determine the attenuation of the germanium line. The magnitude of this attenuation depends on the thickness of sample 6, which is not known and changes from piece to piece, and on the mass attenuation coefficient of the material of sample 6 with respect to the radiation 3 of varying energy E (continuous spectrum) emitted by the polychromatic source 1 and with respect to the monochromatic radiation 5 of energy E' coming from top 4. Attenuation is then defined as the ratio I / Io'. where:

[0038] - Io is the intensity of radiation emitted from top 4 in the absence of sample 6, as in Fig.3, as a function of the energy spectrum E of radiation 3; and

[0039] - / is the intensity of radiation 5’ emitted by top 4 after it has passed through sample 6; and this intensity / is a function of the following parameters:

[0040] - p which is the mass attenuation coefficient of sample 6 based on the energy spectrum E of radiation 3;

[0041] - p' which is the mass attenuation coefficient of sample 6 at the energy E’ of the K or L line of the top 4 material (e.g., germanium); and

[0042] - 1 which is the thickness of the piece.

[0043] The attenuation thus results in a small part from attenuation due to the effect of sample 6 material due to the passage of polychromatic radiation 3 through it, described by the coefficient p(E), which slightly reduces the emissivity of monochromatic radiation 5 from top 4, and from a larger contribution due to the attenuation of this radiation 5 as it passes through sample 6 material and described by the coefficient p'(E’). The mass attenuation coefficient p, which is generally a function of the energy E of the X-radiation, depends on the chemical composition of the polymer, its density and the possible presence in the plastic of elements with atomic number Z>12 that absorb and emit in the region of soft X-rays.

[0044] XRF: The fluorescence lines are emitted by the elements with atomic number Z>12 present in sample 6, so in the case of plastics it is only the elements that constitute fillers, additives, dyes and flame retardants that emit these lines. Therefore, fluorescence radiation 8 is present only if the plastic contains these types of elements, or in the case of samples of other materials of appropriate atomic number.

[0045] XRBS: Back-scattering radiation 7 is scattered from sample 6 through the two distinct processes of Rayleigh scattering (coherent) and Compton scattering (incoherent): the former is an elastic process in which the scattered photon retains the same energy as the incident one, changing only the direction of emission, while the latter is an inelastic process in which the scattered photon loses energy relative to the incident one, with the energy decrease increasing with the energy of the photon itself. The energy loss is about 0.4% at 2keV and about 6% at 15keV, but since the incident spectrum is continuous, this small shift in energy of Compton back-scattering radiation is not appreciable. In general, the Rayleigh back-scattering component is dominant at low energy while the Compton component is dominant at higher energy. The energy value for which the two components are equivalent depends on the material, being about lOkeV for plastics, wood, paper, cardboard, and increasing for heavier materials (about 20keV for glass or higher for bricks and stones).

[0046] As mentioned earlier, in the configuration described above, the angle 9 between the A-axis of source 1 and the A'-axis of detector 2 is between 10° and 40°, preferably 20°, so the scattering angle for which the scattered photons arrive on detector 2 is between 140° and 170°, preferably 160°. The XRBS radiation arriving on detector 2 is between about 2 and 25keV, but it undergoes significant attenuation between about 2 and 5keV, due to air absorption, and gradually decreases to zero between 15 and 25keV; thus, the XRBS radiation measured between 5 and 15keV is much more intense than both the low- energy and higher-energy radiation.

[0047] Note that Compton scattering occurs in all directions, but the maximum occurs for q = 0 (forward-scattering) or q = 180 degrees (back-scattering), where q is the angle between the direction of X-ray propagation and the direction of the photons undergoing scattering. Therefore, the configuration that maximizes the back-scattering signal is one in which source and sensor form an angle of 0°, but in practice this is not possible because source and sensor should coincide. However, due to the innovative use of a low-voltage, low-power source, both source 1 and detector 2 are small so that they can be positioned to form a small angle 9 between them, typically 20° or less.

[0048] The method, therefore, uses the intensity of the back-scattering radiation in the different energy bands, as described below, in which the contribution of the Rayleigh and Compton components is different and the penetration depth is very different.

[0049] Due to the above configuration, the present apparatus is able to perform XRT, XRF and XRBS radiation analysis simultaneously, using both back-scattering phenomena caused by the beam of soft X-rays. In fact, a thorough study of all these types of radiation under the measurement conditions described here, conducted by the inventors with different geometries and materials of various thicknesses, has revealed how it is possible to derive information about the nature of sample 6 and its thickness, which is particularly important for the recognition and separation of a variety of heterogeneous materials, including all types of plastics.

[0050] Since the X-rays of the incident continuous spectrum 3 penetrate sample 6 to a gradually increasing extent as their energy increases, their relative back-scattering radiation also emerges from sample 6 from gradually increasing depths as the photon energy increases, as shown schematically in Figure 2.

[0051] As for the present example of analysis apparatus with top 4 made of germanium, the continuous XRBS back-scattering radiation acquired by detector 2 is preferably divided into the following energy bands over which the intensity of the acquired spectrum is calculated:

[0052] - PO = integral in the range (2keV- 2.5keV)

[0053] - Pl = integral in the range (2.5keV- 3keV)

[0054] - P2 = integral in the range (3 keV - 3.5keV)

[0055] - P3 = integral in the range (3.5keV - 6keV)

[0056] - P4 = integral in the range (11.5keV - 18.5keV)

[0057] - P5 = integral in the range (18.5keV - 21.7keV)

[0058] - P6 = integral in the interval (21.7keV - 25keV)

[0059] Since in these intervals there may also be K or L emission lines of elements present in sample 6, these must be subtracted from the respective values of the P parameters. In the case of plastics, for example, the lines that might most commonly appear are those of P, S, Ca, Ti, Cr, Fe, Br, Sb.

[0060] An example of a spectrum for a 3.5-mm-thick ABS plastic sample, devoid of any emission lines other than germanium, is shown in Fig.4 and is expanded in the vertical scale in Fig.5 to show the back-scattering continuum (XRBS) with the aforementioned P0-P6 energy integration bands shown with dashed vertical lines. If a 1% titanium oxide dye was also present in the sample, the Ka line of titanium fluorescence at about 4.7keV would also be seen in the diagram of Fig.4. The subdivision of the back-scattering spectrum into the seven proposed P0-P6 bands is not the only one possible, as a different number but not less than four can also be defined, including two at low energy to account for the Rayleigh component and two at medium / high energy to account for the Compton component. The important thing is that the bands span the entire range between 2 and 25keV, even if they do not totally cover it.

[0061] It should be noted that the XRT component due to the attenuation of the K or L emission line of the top 4 material (Ge in our example) is absolutely critical to include, since the XRBS component and the XRF component, if any, are not sufficient in general for material classification. Therefore, the bands in which the K or L emission lines are subtracted should not cover the top 4 line; in fact, in the example above the Ge line at 9.8keV is not in any band, being between P3 and P4.

[0062] The P ranges listed above, in addition to having different contributions from the Rayleigh and Compton components, are related to different sampling depths of the piece under analysis. For lighter materials such as plastic, wood, etc., the resulting depth is from a few micrometers for PO up to half a millimeter for P3. From P4 onward, the depth of penetration of radiation into plastics is on the order of a centimeter, but since most plastics are between 1 and 4 mm thick, the P4-P6 radiation escapes from the sample. For P4-P6 intervals, therefore, the back-scattering comes from the entire thickness of the part and, thus, the intensity of the measured radiation is proportional to the thickness of the piece. In any case, the more energetic P4-P6 parameters still have sufficient penetration (at least hundreds of microns) to be unaffected by any surface treatment (coating, varnish, dust, etc.) and thus are useful, together with the other P0-P3 parameters, for material identification.

[0063] In general, parameters P0-P6 are sensitive to the material of the sample, for example, the polymer in the case of plastics, the presence of elements with Z>12 in it, and the different constitution of the surface part. Parameters PO to P3 are related to the low-energy part, where the XRBS intensity is lower and therefore, in the case of measurement with moving pieces, their values have a greater fluctuation (measurement error). In addition, small errors can be made in subtracting the K or L emission lines from the various P intervals, especially for small values of these lines. The redundancy of the P0-P6 parameters compensates for any errors due to measurement fluctuations or subtraction of lines.

[0064] Thus, the present FIXS method of material identification involves the use of the measurement of the attenuation of the top 4 monoenergetic line (XRT), i.e., Ge line intensity (IGC) in the present example, the measurement of P0-P6 integrals (XRBS) and, if present, fluorescence lines (XRF1, XRF2...). To combine all this information together, machine learning techniques are used based on a set of samples representing all the materials of interest, each with different thicknesses ranging from the minimum to the maximum possible thickness for the method. For each of these samples, measurements of all the parameters (IGC, PO, Pl, P2, P3, P4, P5, P6, XRF1, XRF2...) are taken to create a database on which to train the neural network, whose identification accuracy can be verified with a set of measurements other than the training measurements. In the case of creating the database for fast recognition of flowing materials, as in the case of recycling, sample measurements must be repeated several times to also account for measurement fluctuations due to the sample flowing process with respect to the analysis section.

[0065] In the case of moving pieces sliding in a horizontal plane, as in Fig. l, their speed can be between 0.5 and 3 m / s, preferably between 1.5 and 2 m / s. The spectra are acquired continuously and without dead time, integrating them in time intervals ranging from 1 to 10 ms, preferably between 3 and 5 ms.

[0066] In particular, the spectra shown in Figures 3-5 were obtained with acquisition times of about 30 ms, which is the typical time for measuring pieces of plastic or mixed materials flowing on a belt at a speed of about 2 m / s. Each spectrum acquired by spectrometer 2 will contain a continuous spectrum given by back-scattering (XRBS), the line of germanium or other top 4 element used to evaluate transmission (XRT), and possibly fluorescence (XRF) lines of elements present in sample 6.

[0067] The passage of each piece at spectrometer 2 is recorded in a time sequence of spectra, which are recorded preferably every 3 ms, without dead time, transferred to a memory buffer and then analyzed. The beginning and end of the piece are recognized through a system of thresholds, such that the presence of three or four "empty" spectra indicates that these are two different pieces. The physical distance between the pieces to avoid overlapping therefore depends on the speed at which the material is fed on the conveyor belt, as the distance must correspond to the minimum number of empty spectra needed to distinguish the pieces. For example, when the belt speed is 3 m / s, a minimum distance of about 3 cm must be ensured between two pieces, corresponding to a time of 10 ms in which three or four "empty" spectra are detected. In this way, a zeroed signal is obtained between the two pieces sufficient to be able to identify the end of the first piece and the beginning of the second.

[0068] The geometry of the apparatus, as described in Fig.l, the type of material to be analyzed and the characteristics of source 1 in terms of voltage, current, and collimator diameter determine the flux of photons reaching detector 2. In the typical case of dynamic analysis illustrated above, the optimum operating point is to have a rate of counts on detector 2 of about 500.000 to 1.000.000 photons per second, so as to have the highest possible statistics. Higher fluxes would lead to excessive dead time, which in any case must be corrected statistically, while lower fluxes result in a degradation of statistics and thus of readout accuracy. Therefore, by acting on the voltage that defines the spectrum, the current that defines the number of photons, and the diameter of the collimator that adjusts the beam width, one adjusts these source 1 parameters according to the material so as to fall within this range of photons per second. In the case of the geometry of Fig.1 and thermoplastic or heterogeneous materials described below, the voltage range is 35 to 40 kV and the current range is between 100 and 150 microamperes.

[0069] Figures 6 through 9 are graphs showing, by way of example only, the variability of some of the quantities used by the present method in the two very different cases of thermoplastics and heterogeneous materials. It is important to point out that with the FIXS method and apparatus described here, unlike other X-ray analysis apparatuses used in present separators, the decision on the type of material is not binary, but provides for all categories of interest (multi-sorting). This means that the separator must provide multiple exit routes, with at least one route for each category of interest, plus one route for unidentified materials.

[0070] As mentioned earlier, in the present FIXS method, the parameters listed above are used to achieve the identification of various types of plastics, including with the recognition of fillers (glass fiber, calcium carbonate, talc), dyes (titanium oxide, iron oxide, etc.) and flame retardants (bromine, phosphorus).

[0071] The identification of plastics is done using all parameters but, by way of example, Figures 6 and 7 show graphs (scatter plots) for some of the most important and popular types of pure polymers using the intensity of the germanium line IGC versus P2, and P3 versus P4 as parameters, respectively. These two graphs show how these variables respond differently to different types of plastics. Polymers with fillers and additives are identified by also including fluorescence lines (XRF1, XRF2, ...) as additional parameters.

[0072] The number of parameters in this example defines a space with at least 8 dimensions (the intensity of the germanium line IGC and the seven P0-P6 energy bands), which requires the use of machine learning techniques to classify the various types of materials. It is precisely the large number of measured parameters, although partly redundant, that makes it possible to remove the partial overlaps observed when only two-dimensional planes are considered, as are the graphs in Figures 6 and 7.

[0073] Note that since the minimum number of bands is 4, as mentioned above, the space could have as few as 5 dimensions in the absence of fluorescence lines; conversely, in the case of several fluorescence lines and a greater number of energy bands the dimensions could be many more than 8.

[0074] Similarly, the present FIXS method and apparatus can also be used for the identification of heterogeneous materials such as wood, glass, paper, cardboard, brick, rubbers, metals, ceramics, chipboard, etc. As with the separation of plastics, heterogeneous materials are mapped using the above parameters, for example, Fig.8 shows the graph of IGC as a function of P4, and Fig.9 shows the graph of P4 as a function of P3.

[0075] Thus, the general method of FIXS analysis, static or dynamic, according to the invention can be summarized in the following steps: a) generation of a low-energy polychromatic X-ray radiation beam by means of a source 1 having an anode-cathode voltage in the range of 20-60 kV, preferably 20-40 kV, a power in the range of 2-50 W, preferably 5-10 W, and a collimator producing a cone 3 of X-rays of aperture between 10° and 30°, preferably 15°, such as to irradiate the entire surface of an emitting element 4 placed on the opposite side of a sample 6 to be analyzed from the source 1, said emitting element 4 being made of a single material with the K or L emission line in the range of 6 to 25keV; b) sending of said X-radiation beam on sample 6 along an emission axis A; c) detection of the fluorescence radiation 7 and back-scattering radiation 8 emitted by sample 6, and of the monochromatic radiation 5’ emitted by the emitting element 4 and passed through sample 6, such detection being carried out by means of an X-ray spectrometer 2 with a spectral resolution of at least 10%, preferably about 3-4%, placed on the same side of source 1 with respect to sample 6 and arranged at a distance of not more than 10 mm from the chute or support where sample 6 is placed, said emission axis A forming with the detection axis A' of said X-ray spectrometer 2 an angle 9 between 10° and 40°, preferably not more than 30° and more preferably not more than 20°; d) calculation of an XRT transmission parameter measured as the attenuation of said monochromatic radiation 5’ of the emitting element 4 with respect to the condition of absence of sample 6; e) calculation of one or more XRF fluorescence parameters in case of elements in sample 6 that generate K or L emission lines different from the K or L emission line of the material of the emitting element 4; f) calculation of a plurality of XRBS back-scattering parameters measured as intensity integrals in at least four bands of the continuum, including at least two low- energy bands (to account for the Rayleigh component) and at least two mid / high-energy bands (to account for the Compton component), said bands extending over the interval between 2 and 25keV without totally covering it and without including the monochromatic radiation line 5’ of the emitting element 4, while if other K or L emission lines of elements present in sample 6 fall in said bands they are subtracted from the respective values of the integrals; g) use of said XRT, XRF, XRBS parameters to characterize sample 6 using machine learning techniques based on a database containing measured values for all said parameters on a set of samples representing all materials of interest, each with different thicknesses ranging from the minimum to the maximum possible thickness, said values being measured multiple times in the case of databases for fast recognition of samples flowing with respect to spectrometer 2.

[0076] In the typical case of the application of the above method to the separation of materials in an industrial separator as described above, there is also a step aa), preceding step a), in which a flow of material containing a plurality of samples 6 is fed to the analysis system at a speed, preferably between 0.5 and 3 m / s, selected so that the measurement time is at least on the order of 10-20 ms, and providing for a distance between two consecutive samples 6 such that at said selected speed said distance corresponds to a sufficient time to detect at least three empty spectra. In addition, steps c) to g) are repeated in a time sequence of spectra, which are recorded preferably every 3 ms, with no dead time, transferred to a memory buffer and then analyzed. Finally, in step c) the characteristics of source 1 in terms of voltage, current and collimator diameter are preferably adjusted so that the rate of counts on detector 2 is approximately between 500.000 and 1.000.000 photons per second.

Claims

CLAIMS1. X-ray analysis apparatus comprising:- an X-ray source (1) configured to emit a polychromatic radiation beam along an emission axis A towards a sample (6) to be analyzed placed on a chute or support, said X-ray source (1) having an anode-cathode voltage in the range 20-60 kV, preferably 20- 40 kV, and a power in the range 2-50 W, preferably 5-10 W;- an X-ray spectrometer (2) with a spectral resolution of at least 10%, preferably about 3- 4%, positioned on the same side of said sample (6) with respect to said X-ray source (1) and suitable to detect radiations along a detection axis A’, the angle (9) between said emission axis A of the X-ray source (1) and said detection axis A’ of said X-ray spectrometer (2) being between 10° and 40°, preferably not more than 30° and more preferably not more than 20°;- an emitting element (4) positioned on the opposite side of the sample (6) with respect to the X-ray source (1) and made from a material having a K or L emission line in the range between 6 and 25 keV; and- a control unit configured to receive readout data from the X-ray spectrometer (2); characterized in that the X-ray source (1) includes a collimator that generates a cone (3) of X-rays with an aperture angle between 10° and 30°, preferably 15°; and in that the X-ray spectrometer (2) is located at a distance along the detection axis A’ from said chute or support for the sample (6) not greater than 10 mm, the chute or support for the sample (6) being provided with an opening for the passage of said cone (3) of X- rays, of the fluorescence (7) and back-scattering (8) radiations emitted by the sample (6), and of a monochromatic radiation (5’) emitted by said emitting element (4) and passed through the sample (6); and in that the emitting element (4) has the following characteristics:- it has a diameter comprised between 6 and 15 mm, preferably 8 mm, and the combination between the opening of said collimator and said diameter is such that the cone (3) of X-rays irradiates the whole surface of the emitting element (4);- it is made of a single material;- it is placed on the detection axis A’ of the X-ray spectrometer (2), preferablyperpendicularly thereto; and in that said control unit is configured for:(a) calculating an XRT transmission parameter measured as the attenuation of said monochromatic radiation (5’) emitted by the emitting element (4) with respect to the condition of absence of the sample (6);(b) calculating one or more XRF fluorescence parameters in case of elements present in the sample (6) generating K or L emission lines different from the K or L emission line of the material of the emitting element (4);(c) calculating a plurality of XRBS back-scattering parameters measured as integrals of the intensity in at least four bands of the continuous back-scattering spectrum, including at least two low-energy bands and at least two medium / high-energy bands, said bands extending over the interval between 2 and 25keV without totally covering it and without including the monochromatic radiation line (5’) of the emitting element (4), while if other K or L emission lines of elements present in the sample (6) fall in said bands they are subtracted from the respective values of the integrals;(d) utilizing said XRT, XRF, XRBS parameters to characterize the sample (6) using machine learning techniques based on a database containing the measured values for all said parameters on a set of samples representing all materials of interest, each with different thicknesses ranging from the minimum to the maximum possible thickness, said values being measured several times in the case of databases for fast recognition of samples flowing with respect to the X-ray spectrometer (2).

2. Apparatus according to claim 1, characterized in that the X-ray spectrometer (2) consists of a sensor of the SDD type with an active area between 10 and 50 mm2, preferably about 25 mm2.

3. Apparatus according to claim 1 or 2, characterized in that the emitting element (4) is secured at the bottom end of an aluminum rod with a diameter smaller than 6 mm.

4. Apparatus according to any of the preceding claims, characterized in that the material of the emitting element (4) has a K or L emission line at a value such that its attenuation is not less than about 5% in the case of samples with the minimum thickness and not greater than about 95% in the case of samples with the maximum thickness, whenmeasurements are made in a few tens of milliseconds, these limits being able to be 2% and 98%, respectively, when measurement times are a few seconds.

5. Apparatus according to any of the previous claims, characterized in that the distance between the emitting element (4) and the X-ray spectrometer (2) is between 20 mm and 80 mm, preferably about 30-40 mm in the case that the material to be analyzed are plastics, or 30-60 mm in the case of heterogeneous materials.

6. Apparatus according to any of the preceding claims, characterized in that the opening in the chute or support for the sample (6) is about 20 mm long in a longitudinal feed direction and about 10 mm wide in a transverse direction.

7. An industrial separator comprising:- a feed device for a flow of ground material containing plastic materials or other types of material;- an X-ray analysis section of said ground material; and- a control unit configured to receive readout data from said X-ray analysis section and to control means for separating the analyzed material from said flow of ground material; characterized in that the X-ray analysis section comprises an apparatus according to any of the preceding claims, and in that it includes multiple outlets of the analyzed material, with at least one outlet for each category of material, plus an outlet for the unidentified materials.

8. X-ray analysis method, comprising the steps of: a) generating a polychromatic X-ray radiation beam by means of a source (1) having an anode-cathode voltage in the range 20-60 kV, preferably 20-40 kV, a power in the range 2-50 W, preferably 5-10 W, and a collimator that generates a cone (3) of X-rays with an aperture angle between 10° and 30°, preferably 15°, so as to irradiate the whole surface of an emitting element (4) positioned on the opposite side of a sample (6) with respect to said X-ray source (1), said emitting element (4) being made from a single material having a K or L emission line in the range between 6 and 25 kV; b) directing said X-ray radiation beam at said sample (6) along an emission axis A; c) detecting the fluorescence (7) and back-scattering (8) radiations emitted by the sample (6), and the monochromatic radiation (5’) emitted by the emitting element (4) and passed through the sample (6), this detection being carried out by means of an X-rayspectrometer (2) with a spectral resolution of at least 10%, preferably about 3-4%, positioned on the same side of the X-ray source (1) as the sample (6) and arranged at a distance of not more than 10 mm from a chute or support where the sample (6) is placed, said emission axis A forming with the detection axis A' of said X-ray spectrometer (2) an angle (9) between 10° and 40°, preferably not more than 30° and more preferably not more than 20°; d) calculating an XRT transmission parameter measured as the attenuation of said monochromatic radiation (5’) emitted by the emitting element (4) with respect to the condition of absence of the sample (6); e) calculating one or more XRF fluorescence parameters in case of elements present in the sample (6) generating K or L emission lines different from the K or L emission line of the material of the emitting element (4); f) calculating a plurality of XRBS back-scattering parameters measured as integrals of the intensity in at least four bands of the continuous back-scattering spectrum, including at least two low-energy bands and at least two medium / high-energy bands, said bands extending over the interval between 2 and 25keV without totally covering it and without including the monochromatic radiation line (5’) of the emitting element (4), while if other K or L emission lines of elements present in the sample (6) fall in said bands they are subtracted from the respective values of the integrals; g) utilizing said XRT, XRF, XRBS parameters to characterize the sample (6) using machine learning techniques based on a database containing the measured values for all said parameters on a set of samples representing all materials of interest, each with different thicknesses ranging from the minimum to the maximum possible thickness, said values being measured several times in the case of databases for fast recognition of samples flowing with respect to the X-ray spectrometer (2).

9. Method according to claim 8, characterized in that step a) is preceded by a step aa) in which a flow of material containing a plurality of samples (6) is individually fed to the analysis system at a speed, preferably between 0.5 and 3 m / s, selected so that the measurement time is at least in the order of 10-20 ms, and providing for a distance between two consecutive samples (6) such that at said selected speed said distance corresponds to a time sufficient to detect at least three blank spectra, and by the fact thatsteps c) to g) are repeated in a time sequence of spectra, which are preferably recorded every 3 ms, with no dead time, transferred to a memory buffer and then analyzed.

10. Method according to claim 8 or 9, characterized in that in step c) the characteristics of the X-ray source (1) in terms of voltage, current and collimator diameter are adjusted so as to have a count rate on the X-ray spectrometer (2) comprised between about 500.000 and 1.000.000 photons per second.

Images