Method and device for characterising the elementary composition of a sample, by neutron activation

Abstract

The invention relates to a calibration method for characterising the elementary composition of a sample by neutron activation, the method comprising the following steps: receiving (201, 202), for different samples each consisting of one type of pure chemical element, an N-dimensional energy diagram, N being an integer strictly greater than 2, containing the number of occurrences of the detection, by a neutron activation device, of N coincident gamma rays, as a function of the energies of the N gamma rays, the energy diagram being determined by measurement or by simulation; for each sample, selecting (203) at least one set of points of interest in the energy diagram, on the basis of the knowledge of an energy level diagram of the pure chemical element; for each sample, concatenating (204), in order to obtain a characteristic vector of the chemical element, the projections, on one axis of the energy diagram, of each set of points of interest identified for all the samples; and defining (205) a calibration base by the set of characteristic vectors associated with each chemical element.

Description

DESCRIPTION Title of the invention: Method and device for characterizing the elemental composition of a sample, by neutron activation

[0001] The invention relates to the field of characterization of materials via spectroscopic analysis using scintillation detectors.

[0002] The invention applies in particular to the characterization of conventional waste with a view to its recycling. The proposed method aims more precisely to determine the contents of pure chemical elements in a sample that one wishes to characterize.

[0003] The recycling of technological waste generally responds to two challenges.

[0004] A first challenge concerns reducing the environmental impact of end-of-life waste. Some components can, in fact, prove dangerous, both for the environment and for humans, when their end-of-life is not managed properly.

[0005] A second issue concerns the reuse of certain materials where possible. For example, cobalt and rare earths have highly unstable prices, which makes importing these materials potentially complicated for industry. Recycling materials already present in the territories could represent a substantial stock of raw materials, less dependent on international markets.

[0006] The rapid growth in the number of chemical elements used in industry, linked to the technological revolution (particularly in the field of electronics, batteries, or permanent magnets, for example), is leading to the parallel development of suitable recycling methods in order to recover a wide range of elements, some of which can be recovered (or even of strategic importance) in very variable proportions.

[0007] Thus, the characterization of the chemical elements contained in end-of-life products is a crucial step to be carried out before their recycling, in order to size a recycling process appropriate to each product family. Indeed, the material content of technological waste determines the way which will then be recycled. The composition of Li-ion batteries, for example, has evolved significantly over the last decade. The ratios of Nickel / Manganese / Cobalt used in battery cathodes have changed significantly, ranging from (1:1:1) to (8:1:1). Assessing these ratios, without first dismantling the battery, is an important step in its recycling.

[0008] Thus, there is a need to determine, in a non-intrusive manner, the content of pure chemical elements in a sample, for example a battery or any other object or material to be characterized.

[0009] Waste characterization and recycling is a rapidly evolving sector.

[0010] Currently, the techniques applied for waste sorting, including X-ray fluorescence analyses of X-ray lines L (or FXL), allow elemental analysis, however limited to a very low depth (< 1 mm) in the samples.

[0011] Other surface characterization techniques exist, but most do not allow elemental analyses to be performed. Examples include color sorting, eddy current sorting, and LIBS (Laser Induced Breakdown Spectroscopy). Various density measurements, using the buoyancy of materials in different liquids, are also available.

[0012] Among the techniques for characterizing large samples in their entire volume, such as batteries, without having to disassemble or crush them beforehand, there is neutron activation, which allows, thanks to the penetrating power of neutrons, except in special cases (elements not emitting neutron activation gamma radiation, materials absorbing thermal neutrons), to probe the matter over several tens of cm. The analysis of gamma radiation induced by neutrons, of an energy much higher than that of X-rays and therefore also much more penetrating (several tens of cm compared to a few cm for X-rays K and a few mm for X-rays L ), allows larger volumes of waste to be characterized.

[0013] Traditionally, neutron activation techniques have used semiconductor detectors, hyperpure germanium (HPGe), for their very good energy resolution, as presented in patent application US20020175288. However, these come with several drawbacks: their high price, the cooling required for their operation, but above all their small sensitive volume (some 100 cm 3 ), which can result in long measurement times. In addition, the use of GeHP in a neutron activation cell can be problematic due to the degradation of energy resolution under a fast neutron flux.

[0014] To compensate for the shortcomings of the GeHP detector, it can be replaced by NaI scintillation detectors which are characterized by much larger sensitive volumes (up to several thousand cm 3per detector as described in reference [1]), which allows to acquire a greater number of data per unit of time.

[0015] However, the use of Nal(TI) scintillation detectors results in degraded energy resolution (gamma lines are approximately 30 times wider than with a GeHP detector), thus greatly complicating elemental analysis. Indeed, the gamma spectrum includes many lines close in energy which merge. Furthermore, since the peaks are more spread out, the ratio between the signal (gamma lines) and the continuous background noise of the spectrum (Compton continuum) is much lower.

[0016] Therefore, the techniques applicable in gamma spectrometry, using for example the net areas under characteristic peaks, are difficult to apply with precision.

[0017] Peak overlaps are common with Nal(TI) scintillation detectors. They are caused by the Gaussian spread of the energy peaks. Differentiating some peaks can therefore be complicated if the gamma radiation energies are energetically too close.

[0018] Some works (for example in the American patent US6438189) propose a temporal differentiation of gamma radiation by first separating the prompt gamma radiation emitted by fast and slow neutrons, then observing the delayed gamma radiation in a second time. However, the significant background noise observed in the prompt gamma radiation zone is limiting in relation to the sensitivity of the measurement. As for the signal of the delayed gamma radiation, although benefiting from a signal to noise ratio much better, it is however of too low intensity to allow rapid analysis.

[0019] There is therefore a need for a method allowing the characterization, at the elementary level, of large samples in their entire volume (typically of the order of magnitude of 5 to 20 dm 3 or even greater than 20 dm 3 ), in a non-intrusive manner. More specifically, there is a need to characterize samples with Nal(TI) scintillation detectors with improved performance and / or easier implementation compared to prior art techniques.

[0020] The proposed solution to the above-mentioned problem lies in the simultaneous implementation of several Nal(TI) scintillation detectors and in an improved processing of the data measured by these detectors. In particular, the invention is based on the coincidence spectrometry technique which is presented in particular in references [1], [2], [3] and [4] This technique is based on the use of Nal(TI) scintillation detectors. It consists of measuring in coincidence certain gamma rays when they are emitted by a de-excitation cascade involving several gamma rays, and thus creating coincidence spectra. These coincidence measurements thus provide additional information to simple spectrometry which is limited to the study of the energies of the gamma rays emitted individually.

[0021] The subject of the invention is a calibration method for characterizing the elemental composition of a sample by neutron activation, the method comprising the following steps: To receive, for different samples each composed of a type of pure chemical element, an N-dimensional energy diagram, N being an integer strictly greater than 2, containing the number of occurrences of the detection, by a neutron activation device, of N coincident gamma rays, as a function of the energies of the N gamma rays, the energy diagram being determined by measurement or by simulation, For each sample, select at least one set of points of interest in the energy diagram, based on knowledge of an energy level diagram of the pure chemical element, For each sample, concatenate, to obtain a characteristic vector of the chemical element, the projections, on an axis of the energy diagram, of each set of points of interest identified for all the samples, Define a calibration base by the set of characteristic vectors associated with each chemical element.

[0022] According to a particular aspect of the invention, N is equal to 2 and each set of points of interest is defined by a straight line which intercepts the two axes of the diagram at an energy value equal to the highest energy level in the energy level diagram of the chemical element or to an intermediate energy level associated with a transient state.

[0023] According to a particular aspect of the invention, the calibration base is defined by a calibration matrix composed of the characteristic vectors associated with each chemical element.

[0024] The invention also relates to a method for characterizing the elemental composition of a sample by neutron activation comprising the following steps: Determine for the sample, by means of a neutron activation device, a measurement of an N-dimensional energy diagram, N being an integer strictly greater than 2, containing the number of occurrences of the detection, by the device, of N coincident radiations, as a function of the energies of the N radiations, Receiving a calibration base defined by a set of characteristic vectors for a set of pure chemical elements, said base being obtained by means of the calibration method according to the invention, Determine a characteristic vector of the sample by concatenating the projections, on an axis of the measured energy diagram, of each set of points of interest identified for the calibration base, Determine a vector A whose components define the contents of each of the chemical elements of the calibration base in the sample by comparing the characteristic vector of the sample to the characteristic vectors which define the calibration base.

[0025] According to a particular aspect of the invention, the determination of the vector A is carried out by searching for the coefficients of a linear combination of the characteristic vectors which define the calibration base, which minimize an error between the characteristic vector of the sample and said linear combination.

[0026] According to a particular aspect of the invention, the determination of the vector A is carried out by searching for the vector A such that the product of said vector A with the calibration matrix defining the calibration base is equal to the characteristic vector of the sample.

[0027] According to a particular aspect of the invention, the determination of the vector A is carried out by: Calculating the pseudo inverse matrix of the calibration matrix, Calculating the matrix product between the sample characteristic vector and the pseudo inverse matrix.

[0028] According to a particular aspect of the invention, an N-dimensional energy diagram is obtained by means of coincidence gamma spectrometry measurements.

[0029] The invention also relates to a device for characterizing the elemental composition of a sample comprising a neutron generator, at least N gamma radiation detectors, with N an integer greater than or equal to 2, an area capable of receiving a sample, an information acquisition unit characterizing the gamma radiation perceived by the detectors and a processing unit capable of receiving said information and configured to execute the steps of the method according to the invention.

[0030] According to a particular aspect of the invention, the gamma radiation detectors are Nal(TI) scintillation detectors.

[0031] The invention also relates to a computer program comprising instructions which cause the device according to the invention to execute the steps of the method according to the invention.

[0032] Other features and advantages of the present invention will become more apparent upon reading the following description in relation to the following appended drawings.

[0033] [Fig. 1 a] represents a diagram of a coincidence spectrometry measuring device according to a first view,

[0034] [Fig. 1 b] represents a diagram of a coincidence spectrometry measuring device according to a second view,

[0035] [Fig. 1 c] represents a sectional view of the device described in Figures 1 a and 1 b

[0036] [Fig. 2] represents a flowchart detailing the steps of implementing a calibration method according to an embodiment of the invention,

[0037] [Fig. 3a] represents an example of a 2D spectrum of coincidence measurements for an example of a pure chemical element: sulfur,

[0038] [Fig. 3b] represents an example of projection of points of interest of the spectrum of Figure 3a,

[0039] [Fig. 4] shows an example of a sulfur level diagram,

[0040] [Fig. 5a] shows another example of a 2D spectrum for a sulfur sample,

[0041] [Fig. 5b] represents an example of a 2D spectrum for a titanium sample,

[0042] [Fig. 5c] shows an example of a 2D spectrum for a chromium sample,

[0043] [Fig. 6] represents the 2D spectrum of sulfur on which all the projection energies of the three elements are identified: sulfur, titanium, chromium,

[0044] [Fig. 7] represents an example of a characteristic vector of sulfur,

[0045] [Fig. 8] represents a flowchart detailing the steps of a method for characterizing the composition of a sample according to one embodiment of the invention.

[0046] Figures 1 a, 1 b and 1 c represent a diagram of the same coincidence spectrometry measuring device according to two different views and a sectional view.

[0047] The measuring device 100 is suitable for carrying out a coincidence spectrometry measurement on a sample 101.

[0048] The measuring device 100 comprises several Nal(TI) scintillation detectors 102 arranged so as to surround the sample 101. In the example of the figures 1 a, 1 b, 1 c, the number of detectors is equal to eight but this number can be different. Generally it is at least equal to two and the exact number is optimized with respect to the sample volumes 101 to be analyzed and with respect to the processing method described below. The Nal(TI) scintillation detectors 102 offer much larger sensitive volumes than the GeHP-based detectors in order to carry out faster measurements. Furthermore, the degradation of energy resolution of the Nal(TI) scintillators compared to the GeHP scintillators is compensated by the measurement processing method proposed according to the invention.

[0049] The measuring device 100 operates using a neutron activation technique which consists of bombarding the sample 101 with neutrons so as to achieve radiative neutron capture. The nuclei of the matter capture the neutrons essentially when they have become thermal (i.e. of low energy, close to 25 meV). They then find themselves in an excited state then in a deexcited state with the emission of gamma radiation. The energy of the emitted gamma radiation is characteristic of the nuclei concerned. Thus, the analysis of this energy makes it possible to characterize the elements.

[0050] The measuring device 100 thus comprises a neutron generator, for example, which emits neutrons in a pulsed manner. The pulsed mode makes it possible to separate in time the fast neutrons or those undergoing thermalization (which exist only during the pulse and the few tens of ps which follow it), and the thermal neutrons which exist between two pulses. This makes it possible to avoid gamma radiation induced by the fast neutrons (during the pulses, the neutrons being emitted at an energy of 14 MeV, for example, with a deuterium-tritium generator) or during the thermalization of the neutrons (for a few tens to hundreds of ps after each pulse).

[0051] A polyethylene 106 cell allows the moderation of neutrons which are slowed down by diffusion on the hydrogen nuclei of polyethylene (CH2) n. The polyethylene cell 106 is further filled with graphite 107 to create a column of thermal neutrons below the sample 101.

[0052] Cell 106 and the graphite column 107 it contains make it possible to maximize the process of radiative capture of neutrons by the nuclei of sample 101 to be characterized.

[0053] This moderation of neutrons is accompanied by the emission of numerous parasitic gamma radiations (for example, gamma radiation of 4.44 MeV following the inelastic scattering reaction 12 C(n,n')), which are not specific to the analyzed sample 101 placed in a cavity at the top of the thermal neutron column.

[0054] Sample 101 is placed in an area delimited by the elastoboro mat 104 (mixture of boron carbide B4C and elastomer), as close as possible to the detectors. This area is optimized in relation to the field of action of the detectors, in other words the solid angle of detection and the cloud of thermal neutrons.

[0055] The scintillation detectors 102 are coupled to a photomultiplier in order to transform the visible light created during the interaction of gamma radiation into an electrical signal. The scintillation detectors 102 are positioned as close as possible to the sample 101 to be analyzed in order to maximize the detection efficiency. This positioning depends on the volume of the sample considered. In the example of Figure 1 a, the detectors 102 are contiguous but they can be further apart or superimposed depending on the geometry of the sample 101. In order to limit the crosstalk of the gamma rays, that is to say the parasitic coincidences due to gamma scattering from one detector to another, it is possible to place lead plates between the detectors.

[0056] The device 100 also includes a lead brick 103 placed under the detectors 102 which allows the shielding of the detectors against the parasitic gamma radiation emitted in the cell during the neutron emission pulses, then during their moderation, in particular by inelastic diffusion, and finally during their radiative captures, once they have reached the thermal energy.

[0057] The device 100 also comprises two elastomeric (B4C) mats 104, 105 which shield the detectors against stray thermal neutrons. The first mat 105 is located under the lead brick 103 and the second lines the vertical walls of the detectors, on the side of the sample 101 to be analyzed.

[0058] The device also includes acquisition electronics (not shown in Figures 1 a, 1 b, 1 c) to digitize the energy perceived by the detectors as well as the time of interaction. In other words, the essential information of each detection, such as the arrival time and the energy of a detected gamma radiation are grouped into a list. Coincidences between gamma rays are then sorted in the acquisition files according to the arrival time associated with each detection.

[0059] The acquisition electronics are configured to acquire measurements in a time range during which the neutrons have been completely slowed down (by the moderator cells), thus eliminating the components of gamma radiation linked to the interactions of fast and slowing neutrons.

[0060] Finally, a computer or any other equivalent processing unit is used to process the data digitized by the acquisition electronics and implement the sample characterization method according to the invention.

[0061] The invention relates more specifically to a new method for analyzing coincidence spectrometry measurements carried out by the device of Figure 1 a. This method is broken down into two main phases: a first phase consisting of a calibration of the method so as to characterize pure chemical elements to construct a calibration base, a second phase consisting of an exploitation of the calibration base to determine the content of pure chemical elements of a new sample.

[0062] Figure 2 describes, on a flowchart, the steps of implementing a calibration method according to one embodiment of the invention.

[0063] The calibration method aims to characterize several samples consisting of pure chemical elements such as sulfur, titanium, chromium or any other chemical element.

[0064] The method begins at step 201 with a first set of coincidence spectrometry measurements carried out for a first sample consisting of a single pure chemical element. The objective is to spectrally characterize this chemical element in order to then construct a calibration matrix which defines a calibration basis for different pure chemical elements.

[0065] The measurements carried out in step 201 are obtained using the device of figures 1 a, 1 b, 1 b and consist of a list of information characterizing each detection of energy of a gamma radiation. More precisely, this list allows to define coincidences between gamma rays, that is, radiation which is detected by at least two different detectors at very close times.

[0066] For each detection measured in step 201 by one of the detectors, information is received on its time of arrival and the energy of the radiation deposited in the scintillator.

[0067] In step 202, an N-dimensional spectrum is then constructed, N being an integer at least equal to 2, by combining the measurements obtained using the different scintillation detectors 102. The construction of these spectra is based on the simultaneous detection by at least N detectors of several gamma radiations from the same de-excitation cascade.

[0068] In the remainder of the description, the invention is described for N = 2, i.e. for two-dimensional coincidence spectra. Without departing from the scope of the invention, N-dimensional spectra, with N greater than 2, may be envisaged provided that the measuring device comprises at least N separate scintillation detectors.

[0069] An example of a 2D spectrum is given in Figure 3a for the case of sulfur. Such a 2D spectrum or energy diagram represents the energies measured simultaneously by two different detectors, on the x and y axes. The z axis is represented by a scale on the right of Figure 3a which specifies the intensity of each pixel of the diagram which corresponds to a number of hits or occurrences, that is to say the number of simultaneous detections of the energy pair whose values ​​are the x and y coordinates.

[0070] In step 203, one or more sets of points of interest are then determined on the 2D spectrum of Figure 3a. The points of interest are selected because they correspond to specific transitions and energy levels of the atomic nuclei of the sample considered.

[0071] For example, in Figure 3a, the selected points of interest are those located between the two lines D1,D2.

[0072] The expression "points of interest" should not be understood as being limited to a point on the diagram in the primary sense of the term. Indeed, in practice, detectors may have an energy resolution that is not perfect and the points of interest are spread across the diagram in the form of tasks also called clusters. In the rest of the document the expression "points of interest" is used to more generally designate clusters or groups of neighboring points, the spread of the cluster depending on the energy resolution of the detectors.

[0073] Generally speaking, the sets of points of interest are concentrated in bands defined by two parallel lines with slopes equal to -1 and which are parallel to the diagonal of the 2D spectrum. The thickness of these bands is defined so as to encompass the points of interest and takes into account in particular the resolution of the detectors. Each line is defined by an energy value with zero abscissa or ordinate (the 2D spectrum being symmetrical). This energy value is characteristic of the isotope (or one of the isotopes if there are several) of the pure chemical element and is determined from a diagram of the nuclear levels of these isotopes.

[0074] More precisely, the energy values ​​sought are those which correspond to the most probable gamma radiation emissions. The energy values ​​sought correspond to the transitions between the energy levels of the isotope(s) of the chemical element.

[0075] An example of an isotope energy level diagram 32 S of sulfur is given in Figure 4.

[0076] For sulfur, the element's excitation level 33 S, compound nucleus formed during neutron capture by the 32 S, corresponds to an energy of 8642 keV. Thus, a first set of points of interest, for sulfur, is defined on the spectrum of Figure 3a by points which are located approximately on a straight line such that the sum of the energies detected by the two detectors is equal to the maximum energy of 8642 keV.

[0077] In the example of Figure 3a, line D1 is defined by a sum of energies equal to 8700 keV and line D2 by a sum of energies equal to 8600 keV.

[0078] In other words, the points of interest of a 2D spectrum are defined by a straight line x+y=E where E corresponds to an energy level characteristic of the chemical element considered.

[0079] The selected points of interest are then projected onto one of the axes (x or y) to obtain the diagram in Figure 3b which gives the intensity of the detected radiation as a function of its energy.

[0080] In this example, we can identify energy couples (E Y I = 3221 and E Y 2 = 5421 keV) or (E Y r = 841 and E Y2 ' = 7801 keV) whose sum of energies (of each couple) gives the energy of the energy level of the projection (E prOje ction = E Y1 + E Y2 = E Y i' + E Y2 = 8642 keV).

[0081] These different characteristic energies correspond to the cascade gamma ray transitions identified on the sulfur energy level diagram 33 S in Figure 4.

[0082] The level diagram of a compound nucleus formed by neutron capture by an isotope of a chemical element, here sulfur, lists all the possible excitation states for this compound nucleus after its neutron capture. It thus shows all the transitions between the different levels, corresponding to the emission of gamma rays as well as their probability, called branching ratios. The lifetimes associated with these levels are of the order of a picosecond, well below the temporal resolution of scintillation detectors which is of the order of a nanosecond, and we therefore consider the emissions of all gamma rays from a de-excitation cascade as simultaneous events.

[0083] For the same compound nucleus formed by neutron capture of an isotope of a chemical element, several different projections can be defined corresponding to different energy levels. The maximum energy level is always considered, but intermediate energy levels can also be taken into account.

[0084] Indeed, when an energy level de-excites towards the fundamental level via the emission of two gamma rays in series, a set of points of interest appears on the 2D spectrum. If two 'paths', each involving two successive transitions, start from the same excitation level to reach the fundamental level, two distinct sets of points of interest appear on the 2D spectrum, on the same diagonal. This diagonal is identified as having the energy of the sum of the two gamma rays coming from the same cascade of de-excitation. In the example of sulfur, we have two sets of different points (+ their symmetrical points) identified on the diagonal projection at the energy of 8642 keV.

[0085] We therefore have a first set of points of interest which corresponds to the state characterized by the highest energy level obtained by the compound nucleus after neutron capture.

[0086] The other sets of points of interest for the same compound nucleus are characterized by combinations of transition energy between different energy levels. Indeed, we must also consider transitions that lead back to a transient state that is higher in energy than the ground state. Thus, if an intermediate energy level deexcites by means of two cascading gamma rays, this results in a set of points of interest on the 2D spectrum. Reusing the sulfur level diagram in Figure 4, we can see that the 3221 keV energy level can deexcite by the emission of two cascading gamma rays: a 2380 keV gamma ray, then another 841 keV. Another set of points of interest therefore exists on the 2D spectrum at the intersection of these two energies, on a diagonal of energy 841 + 2380 = 3221 keV.

[0087] Figure 5a shows, on the same 2D spectrum corresponding to a sulfur sample, the three diagonals of sets of points corresponding to three energy levels characteristic of sulfur: S,1 E ; S,2 E ; S,3 E .

[0088] Figure 5b shows another example of a 2D spectrum for a titanium sample, also with three characteristic energy levels: Ti, 1 E ; Ti,2 E ; Ti,3 E .

[0089] Figure 5c shows yet another example of a 2D spectrum for a chromium sample, this time with a single characteristic energy level: Cr,1 E .

[0090] Steps 201, 202, 203 are repeated for all pure chemical elements considered and which are likely to be present in future samples to be analyzed. These may be, for example, the elements C, O, N, Na, Mg, Al, Si, Fe, Ni, Mn and Co, this list not being exhaustive.

[0091] At the end of step 203, when all the pure chemical elements have been processed, we have all the projection energies characteristic of each chemical element. For the example in figures 5a, 5b, 5c, we thus have 7 projection energies for three chemical elements.

[0092] In step 204, a characteristic vector is then constructed for each chemical element, which is made up of the concatenation of the projections of the points of interest noted on the 2D spectrum of this chemical element but for all the projection energies of all the chemical elements (not only the projection energies identified as being of interest for this chemical element).

[0093] The objective of this step is to take into account the interferences that a given element may have on the projections of another element.

[0094] To define this characteristic vector, we plot, on each of the 2D spectra corresponding to the pure elements, all the diagonal projections linked to all the characteristic energies previously noted for each pure element.

[0095] By plotting on the 2D spectrum of a given element, the projections linked to the energy levels of the other elements, we can take into account the interference that one element can cause on another.

[0096] Indeed, when several elements are present in a sample, the elements associated with high-energy diagonals can create a background noise on the lower-energy diagonals (due to the possible partial deposition of energy in the detectors following Compton scattering or materialization). This background noise is intrinsically taken into account in the way of plotting the diagonal projections. All the diagonal projections obtained for the 2D spectrum of a given pure element (those that characterize it and those that characterize the other elements) are put end to end (concatenated) in the form of a characteristic vector. It is this vector specific to each element that will be used subsequently to create a calibration matrix.

[0097] At the end of step 204, we therefore obtain a list of characteristic vectors of each of the elements of the basis considered. By default, with the proposed method, all these vectors have the same dimension since they are based on the same set of energies of the characteristic projections.

[0098] Figure 6 represents the 2D spectrum of sulfur on which the seven diagonals corresponding to the seven projection energies for all three elements: sulfur, titanium and chromium have been placed.

[0099] The characteristic vector of sulfur is obtained by concatenating these seven projections on one of the x and y axes.

[0100] An example of the shape of the characteristic vector of sulfur is given in Figure 7.

[0101] This characteristic vector of sulfur is obtained when analyzing on a larger basis than that illustrated in Figure 6, comprising a dozen chemical elements and thirty different projections. The channels of each of the projections were grouped to increase the total count per channel. Indeed, on the 2D spectra, a pixel is 1 keV / 1 keV or 10 keV / 10 keV. To perform diagonal projections, and limit the number of channels with a zero value, the data are grouped by groups of 30 keV, or 50 keV or 100 keV according to the user's choice.

[0102] In the case illustrated in Figure 7, channels 1 to 10 correspond to a given diagonal projection. The energies recorded on this projection were grouped on only 10 channels.

[0103] Channels 10 to 20 correspond to another projection, etc.

[0104] Figure 7 shows, among other things, two projections presented previously:

[0105] Cr,1 Ps is positioned at a higher energy level than the sulfur maximum (8642 keV), and almost no hits are recorded on this projection,

[0106] s,1 Ps is the projection onto the highest sulfur level (8642 keV), also shown in Figure 6.

[0107] In step 205, a calibration base is defined by the set of characteristic vectors constructed for the different pure chemical elements.

[0108] In a particular embodiment, all of the characteristic vectors determined in step 204 can be grouped in the form of a matrix M, called the calibration matrix, the number of rows of which corresponds to the number of chemical elements in the base. The number of columns of the matrix M is equal to the total number of channels of the characteristic vectors of the elements. The calibration base is then defined by the calibration matrix M.

[0109] We can also calculate the pseudo inverse matrix of this matrix M (generalization of the inverse of a matrix to non-invertible cases by removing some of the properties required of inverses, here a non-square matrix), which will be used during the analysis phase.

[0110] The pseudo inverse matrix is ​​given by the relation:

[0111] M + =M t *(M*M t ) -1, where M' denotes the transpose of the calibration matrix M.

[0112] Once the calibration base has been constructed for a given set of pure chemical elements, it is possible to characterize a new sample, of unknown composition, from new measurements on this sample and the calibration base.

[0113] Figure 8 details the steps for implementing a method for characterizing a sample according to one embodiment of the invention.

[0114] The method aims to characterize the composition of a sample, i.e. the content, in this sample, of the pure chemical elements which were used to define the calibration matrix during the calibration phase.

[0115] The method begins at step 801 with the acquisition of measurements identical to those carried out at step 201 of the calibration method but this time for a sample whose composition is unknown and which we wish to characterize.

[0116] In step 802, a 2D coincidence spectrum is determined from the measurements (step identical to step 202).

[0117] In step 803, a characteristic vector V is determined Y of the sample by concatenating all the projections obtained from the projection energies determined for all the pure elements of the calibration phase. This step is identical to step 204 for one element.

[0118] Then, in step 804, the content of pure elements contained in the sample is calculated.

[0119] According to a first variant embodiment of step 804, when the calibration base is defined from a calibration matrix M, step 804 is carried out by solving the following system:

[0120] V Y = A ' M

[0121] M is the calibration matrix and A is a row vector of dimension equal to the number of pure elements used to calculate the calibration matrix. The components of vector A are the pure element contents (percentage) of each of the pure elements in the analyzed sample.

[0122] One possible solution method is to calculate the pseudo inverse matrix of the calibration matrix M and then calculate the vector A as the product of the characteristic vector and the pseudo inverse matrix.

[0123] A= V Y *M t *(M*M t )' 1 = V Y *M +

[0124] According to another alternative embodiment of step 804, when the calibration base is more generally defined by a set of characteristic vectors representative of each pure chemical element, the vector A containing the contents of a new sample in pure elements can be determined by applying a least squares resolution method so as to minimize a difference between the vector V Y and a linear combination of the characteristic vectors defining the calibration base, the coefficients of this combination corresponding to the sought components of vector A.

[0125] In another embodiment, the least squares resolution method can be replaced by a gradient descent algorithm, or a Bayesian inference method or a machine learning algorithm. In general, any numerical resolution method allowing the coefficients of the vector A to be determined from the vector V Y and vectors characterizing the calibration base can be considered as a replacement for the methods listed above.

[0126] Without departing from the scope of the invention, the proposed method can be extended to the measurement of coincident gamma radiation detected on more than two detectors simultaneously.

[0127] For example, 2D spectra can be replaced by three-dimensional spectra, each dimension of which corresponds to the detection of radiation by one of three separate detectors. In general, N-dimensional spectra with N greater than 3 can also be considered.

[0128] In the case where N is strictly greater than 2, the selection of points of interest is no longer carried out by drawing lines but via the construction of planes or hyperplanes.

[0129] Then, the step of projecting the points of interest onto one of the axes of the spectrum is similar to the N=2 case. References

[0130]

[0001] J. Rouit et al., “Prompt Gamma Neutron Activation Analysis with Gamma-Gamma Coincidences for Recycling Waste Characterization.”

[0131] [2] R. P. Gardner, C. W. Mayo, E. S. El-Sayyed, W. A. Metwally, Y. Zheng, and M. Poezart, “A feasibility study of a coincidence counting approach for PGNAA applications,” Applied Radiation and Isotopes, vol. 53, no. 4, pp. 515-526, Nov. 2000, doi: 10.1016 / S0969-8043(00)00206-2.

[0132] [3] W. A. Metwally, C. Mayo, X. Han, and R. Gardner, “Coincidence counting for PGNAA applications: Is it the optimum method?,” Journal of Radioanalytical and Nuclear Chemistry - J RADIOANAL NUCL CHEM, vol. 265, pp. 309-314, juillet 2005, doi: 10.1007 / s10967-005-0826-2.

[0133] [4] N. Markovic, P. Roos, and S. P. Nielsen, “COINCIDENCE GAMMARAY SPECTROMETRY,” p. 6.

Claims

CLAIMS 1. Calibration method for characterizing the elemental composition of a sample by neutron activation, the method comprising the following steps: Receive (201, 202), for different samples each composed of a type of pure chemical element, an N-dimensional energy diagram, N being an integer strictly greater than 2, containing the number of occurrences of the detection, by a neutron activation device, of N coincident gamma rays, as a function of the energies of the N gamma rays, the energy diagram being determined by measurement or by simulation, For each sample, select (203) at least one set of points of interest in the energy diagram, from the knowledge of an energy level diagram of the pure chemical element, For each sample, concatenate (204), to obtain a characteristic vector of the chemical element, the projections, on an axis of the energy diagram, of each set of points of interest identified for all the samples, Define (205) a calibration base by the set of characteristic vectors associated with each chemical element.

2. Calibration method according to claim 1 in which N is equal to 2 and each set of points of interest is defined by a straight line which intercepts the two axes of the diagram at an energy value equal to the highest energy level in the energy level diagram of the chemical element or to an intermediate energy level associated with a transient state.

3. Calibration method according to any one of claims 1 or 2 in which the calibration base is defined (205) by a calibration matrix composed of the characteristic vectors associated with each chemical element.

4. Method for characterizing the elemental composition of a sample by neutron activation comprising the following steps: Determine (801,802) for the sample, by means of a neutron activation device, a measurement of an N-dimensional energy diagram, N being a integer strictly greater than 2, containing the number of occurrences of the detection, by the device, of N coincident radiations, as a function of the energies of the N radiations, Receiving a calibration base defined by a set of characteristic vectors for a set of pure chemical elements, said base being obtained by means of the calibration method according to any one of claims 1 to 3, Determine (803) a characteristic vector of the sample by concatenating the projections, on an axis of the measured energy diagram, of each set of points of interest identified for the calibration base, Determine (804) a vector A whose components define the contents of each of the chemical elements of the calibration base in the sample by comparing the characteristic vector of the sample to the characteristic vectors which define the calibration base.

5. Method for characterizing the elemental composition of a sample according to claim 4 in which the determination (804) of the vector A is carried out by searching for the coefficients of a linear combination of the characteristic vectors which define the calibration base, which minimize an error between the characteristic vector of the sample and said linear combination.

6. Method for characterizing the elemental composition of a sample according to one of claims 4 or 5 in combination with claim 3 in which the determination (804) of the vector A is carried out by searching for the vector A such that the product of said vector A with the calibration matrix defining the calibration base is equal to the characteristic vector of the sample.

7. Method for characterizing the elemental composition of a sample according to claim 6 in which the determination (804) of the vector A is carried out by: Calculating the pseudo inverse matrix of the calibration matrix, Calculating the matrix product between the sample characteristic vector and the pseudo inverse matrix.

8. A method according to any preceding claim wherein an N-dimensional energy diagram is obtained by means of coincidence gamma spectrometry measurements.

9. Device for characterizing the elemental composition of a sample comprising a neutron generator, at least N gamma radiation detectors, with N an integer greater than or equal to 2, an area capable of receiving a sample, an information acquisition unit characterizing the gamma radiation perceived by the detectors and a processing unit capable of receiving said information and configured to execute the steps of the method according to any one of claims 1 to 8.

10. Characterization device according to claim 9 in which the gamma radiation detectors are Nal(TI) scintillation detectors.

11. Computer program comprising instructions which cause the device according to one of claims 9 or 10 to execute the steps of the method according to any one of claims 1 to 8.

Images