Devices and methods for measuring objects containing beta radioactivity

Abstract

To provide a device and method for measuring objects containing beta radioactivity. [Solution] A method for characterizing an object, wherein the object comprises at least one radionuclide that emits β radiation, and the method comprises the steps of: positioning a detector facing the object; detecting radiation emitted by the object; forming an input spectrum containing a β component from the radiation detected by the detector; and applying an identification algorithm related to the radionuclide to the input spectrum, wherein the identification algorithm determines the presence of the radionuclide related to the object in the object; identifying the radionuclides contained in the object according to step d); and applying a deconvolution algorithm to the input spectrum to estimate the contribution of each radionuclide identified in step e) to the input spectrum.

Description

[Technical Field]

[0001] The technical area of ​​the present invention is the measurement of the β radioactivity of an object by spectroscopic measurement. [Background technology]

[0002] Knowledge of the radioactivity status of processes and equipment in nuclear facilities is necessary to establish robust decommissioning scenarios and to manage waste, particularly to classify waste and determine how to dispose of it. In-situ non-destructive nuclear measurements combined with modeling techniques enable the establishment of radiation inventories of processes, permanent equipment, and civil structures.

[0003] Gamma-ray spectroscopy is one of the most commonly used passive, non-destructive nuclear measurement techniques to obtain qualitative and quantitative information about gamma-ray emitting radionuclides present in an object. Gamma-ray spectroscopy is used to acquire gamma-ray spectra, where characteristic peaks corresponding to signatures that enable the identification of radionuclides can be identified.

[0004] However, certain radionuclides emit little to no gamma rays. One example of these is the so-called pure beta emitter, whose radiological characterization in situ is difficult due to the extremely short free path of electrons in a high-density medium. 90 It is Sr. Pure beta emitters are typically identified and their radioactivity quantified in the laboratory by destructive analysis of samples taken in situ. However, these destructive laboratory measurements have several drawbacks, namely, doubts about the representativeness of the collected samples, as well as the cost and time of analysis.

[0005] The publication Vetter K, "In-situ quantification of gamma-ray and beta-only emitting radionuclides," arXiv, April 9, 2023, http: / / arxiv.org / abs / 2304.07632, hereafter referred to as [Vetter], describes a spectroscopic measurement that acquires spectra using a compact CdZnTe semiconductor detector. The device is 137 Gamma rays generated by Cs, 137 Internal conversion electrons emitted by Cs, and 90 To detect the beta rays emitted by Sr, the detector is positioned close enough to the object being characterized. From the spectrum, components are extracted: one representing interactions occurring in the surface region of the detector, including most of the electron-derived interactions (internal conversion electrons, beta rays), and another representing interactions occurring deep within the detector, including interactions involving gamma photons. The spectrum is acquired by placing a screen between the detector and the object to determine which surface interactions are due to gamma-ray photons. The electron and photon spectra are then deconvoluted using a maximum likelihood algorithm to quantify the radioactivity of the radionuclides.

[0006] The advantage of the method described in the publication [Vetter] is that it allows for the estimation of the radioactivity of pure beta emitters in an object without taking a sample and without destructive analysis. However, this method requires determining the depth of interaction in the detector, which is relatively complex. Separating the spectrum into components representing electron-derived interactions and components representing photon-derived interactions can be cumbersome, especially when done in situ, i.e., not under laboratory conditions. Moreover, interpreting the spectrum is relatively complex.

[0007] Another drawback is the fact that a CdZnTe detector is used. This type of detector has a volume of several centimeters. 3It is based on crystals limited to this. Finally, another constraint in implementation is that two measurements must be performed: one with a screen and one without.

[0008] The inventors have developed an alternative method having the same objective as the method described in the publication [Vetter]. The inventors' method does not require consideration of the depth of interaction in the detector. Furthermore, it is not limited to the use of semiconductor detectors and can be advantageously implemented with scintillator detectors. [Prior art documents] [Non-patent literature]

[0009] [Non-Patent Document 1] Vetter K, “In-situ quantification of gamma-ray and beta-only emitting radionuclides,” arXiv, April 9, 2023, http: / / arxiv.org / abs / 2304.07632 [Non-Patent Document 2] Venara J. et al., “Design and development of a portable β-spectrometer for 90Sr activity measurements in contaminated matrices,” Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, vol. 953, p. 163081, February 2020. [Overview of the Initiative] [Means for solving the problem]

[0010] The first subject of the present invention is a method for characterizing an object, wherein the object comprises at least one radionuclide that emits beta radiation, and the method is: -a) A step of positioning a detector facing an object, wherein the detector is configured to acquire a spectrum representing the distribution of energy emitted within the detector by radiation emitted by the object, -b) A step of detecting radiation emitted by an object during the acquisition period using a detector and obtaining the spectrum of the detected radiation, -c) A step of forming an input spectrum from the spectrum of detected radiation that includes a β component corresponding to the energy distribution emitted by β radiation within the detector, -d) A step of applying an identification algorithm related to a radionuclide to the input spectrum, wherein the identification algorithm is configured to determine the presence of the relevant radionuclide within the object, and step d) is repeated for various radionuclides by performing different identification algorithms. -e) A step to identify each radionuclide contained in the object, according to step d), -f) A step of applying an input spectrum deconvolution algorithm to estimate the contribution of each radionuclide identified in step e) to the input spectrum, -g) For each radionuclide identified in step e), estimate the radioactivity and / or the depth to which the radionuclide extends within the object from the contribution of the radionuclide to the input spectrum estimated in step f). Includes.

[0011] Steps d), f), and g) are performed by the processing unit based on the input spectrum. Step e) may also be performed by the processing unit.

[0012] According to one possibility, in step d), each identification algorithm is an artificial intelligence identification algorithm associated with each radionuclide, and at least two different radionuclides are associated with two different respective identification algorithms.

[0013] Each identification algorithm can be a neural network.

[0014] According to one possibility, the deconvolution algorithm is based on a deconvolution database containing at least one spectrum representing each radionuclide identified in step e).

[0015] According to one possibility, - The deconvolution database contains various spectra representing different distributions of radionuclides within an object for a given radionuclide. -Step g) includes the step of determining the distribution of radionuclides within the object.

[0016] According to one possibility, - The deconvolution database contains various spectra representing different depths of the radionuclide within an object, from the surface of the object facing the detector, for a given radionuclide. -Step g) includes determining the depth to which the radioactive nuclide extends within the object.

[0017] At least one radionuclide associated with the identification algorithm may be a pure beta emitter.

[0018] The detector may include an organic scintillator material for detecting radiation emitted by an object.

[0019] The detector may include a volume of inorganic scintillator or semiconductor less than 10 mm thick, positioned facing the object, with the thickness measured perpendicular to the object.

[0020] According to one possibility, the object has natural radioactivity, and step c) - A step to estimate the spectrum of the object's natural radioactivity, - In order to form the input spectrum, the step is to subtract the spectrum of the object's natural radioactivity from the spectrum obtained in step b). Includes.

[0021] The detector may include a movable screen configured to be inserted between the detector and the object, and this method is - A step of acquiring the background spectrum with the screen inserted between the detector and the object, - The background spectrum is subtracted in step c) from the spectrum obtained in step b) to form the input spectrum. Includes.

[0022] A second subject of the present invention is a detection device comprising a detector configured to acquire a spectrum of beta radiation emitted by an object, the spectrum representing the distribution of energy emitted within the detector during the interaction of ionizing radiation within the detector, and the device comprising a processing unit configured to carry out steps d) to f) of the method described in the first subject of the present invention.

[0023] The detector may include an organic scintillator material for detecting radiation emitted by an object.

[0024] The detector may include a volume of inorganic scintillator or semiconductor with a thickness of less than 10 mm.

[0025] The detector may include a movable screen configured to be inserted between the detector and the object.

[0026] The present invention will be better understood by reading the disclosure of exemplary embodiments presented in the remainder of this specification with reference to the figures listed below. [Brief explanation of the drawing]

[0027] [Figure 1] This figure schematically shows a measuring device that enables the implementation of the present invention. [Figure 2A]This figure shows the spectrum of 137Cs measured in the laboratory. Unless otherwise specified, for each spectrum described in this patent application, the x-axis corresponds to energy (in MeV) and the y-axis corresponds to the number of interactions detected. [Figure 2B] This figure shows the gamma spectrum of 137Cs measured in the laboratory. [Figure 2C] This figure shows the spectrum of 90Sr measured in the laboratory. [Figure 3] This figure shows the βγ spectrum of an object containing 137Cs and 90Sr. [Figure 4] This is a diagram illustrating a modeled configuration. [Figure 5A] This figure shows the modeled spectrum, as well as the contributions of artificial radionuclides (137Cs and 90Sr) and natural radionuclides in two different configurations. [Figure 5B] This figure shows the modeled spectrum, as well as the contributions of artificial radionuclides (137Cs and 90Sr) and natural radionuclides in two different configurations. [Figure 6] This figure schematically illustrates the main steps of the method according to the present invention. [Figure 7A] This is a diagram illustrating another modeled configuration. [Figure 7B] This figure shows an example of a training spectrum. [Figure 8A] This figure shows a modeled spectrum of an object containing 90Sr. [Figure 8B] This figure shows the probability of the presence of various radionuclides in the spectrum of Figure 8A. This is the output of the identification algorithm. [Figure 9A] This figure shows the modeled spectrum of an object containing 14C and 36Cl. [Figure 9B] This figure shows the probability of the presence of various radionuclides in the spectrum of Figure 9A. [Figure 10A]This figure shows a spectrum modeled to have an exponential radioactivity gradient for a given 90Sr radioactivity distributed in the depth direction. [Figure 10B] This figure shows the modeled spectrum of Figure 10A, normalized by each integral. [Figure 10C] This figure shows the ratio between each spectrum in Figure 10B and the spectrum representing the uniform 90Sr radioactivity normalized by integration. [Figure 10D] This figure corresponds to Figure 10C, showing the distribution of 137Cs radioactivity in the depth direction. [Figure 11] This figure compares the modeled spectrum of 90Sr radioactivity distributed with an exponential gradient to a depth of 40 mm with the spectrum of uniform 90Sr radioactivity. [Figure 12A] This figure shows the stepwise fitting of the spectrum to the measured spectrum, based on the iteration of the deconvolution algorithm. [Figure 12B] This figure shows the stepwise fitting of the spectrum to the measured spectrum, based on the iteration of the deconvolution algorithm. [Figure 12C] This figure shows the stepwise fitting of the spectrum to the measured spectrum, based on the iteration of the deconvolution algorithm. [Figure 12D] This figure shows the stepwise fitting of the spectrum to the measured spectrum, based on the iteration of the deconvolution algorithm. [Figure 13] This diagram schematically shows a sample taken laterally from a channel in a graphite-moderated reactor. [Figure 14A] This figure shows the probabilities of the presence of 137Cs and 90Sr in graphite samples, with each probability established by running various discriminant neural networks 100 times for each sample. [Figure 14B]This figure shows the probabilities of the presence of 137Cs and 90Sr in graphite samples, with each probability established by running various discriminant neural networks 100 times for each sample. [Figure 14C] This figure shows the probabilities of the presence of 137Cs and 90Sr in graphite samples, with each probability established by running various discriminant neural networks 100 times for each sample. [Figure 14D] This figure shows the probabilities of the presence of 137Cs and 90Sr in graphite samples, with each probability established by running various discriminant neural networks 100 times for each sample. [Figure 15] This figure shows one example of an embodiment of spectral deconvolution for determining the contribution of various radionuclides to the measured spectrum. In Figure 15, the x-axis corresponds to each channel. [Figure 16A] This figure shows the spectrum measured for a graphite sample. [Figure 16B] This figure shows the spectrum measured with a screen placed between the detector and the graphite sample. [Figure 16C] This figure shows the spectrum obtained by subtracting the spectra shown in Figures 16A and 16B. [Modes for carrying out the invention]

[0028] Figure 1 shows a measuring device that enables the measurement of radioactivity of object 2. The device is described in the publication "Design and development of a portable β-spectrometer for" by Venara J. et al. 90As described in “Sr activity measurements in contaminated matrices,” Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, Vol. 953, p. 163081, February 2020, the system includes a scintillator detector 10 comprising a scintillator material 11, preferably an organic scintillator material, preferably a scintillator material based on polyvinyltoluene (PVT). Under the influence of the interaction between ionizing radiation and the scintillator material, optical pulses are formed. These optical pulses are converted into electrical pulses by one or more photodetectors 12. The electrical pulses are then processed by a spectroscopic measurement circuit 13. The spectroscopic measurement circuit 13 is configured to form an amplitude histogram of the pulses detected by the organic scintillator during the acquisition period.

[0029] The relationship between pulse amplitude and energy value has traditionally been determined through an energy calibration function derived from energy calibration. When ionizing particles deposit all their energy within the scintillator, the amplitude of the resulting pulse corresponds to the energy of the particle before interaction within the detector material. The detector's energy calibration, i.e., the relationship between pulse amplitude and energy, is determined by the emission of electrons at discrete energy values ​​via internal conversion. 207 Bi and / or 137 This is performed using a Cs source. If the source used for calibration emits gamma rays, it is possible to utilize specific characteristic energies, such as the energy corresponding to the Compton edge or the energy of the photoelectron peak.

[0030] Organic scintillator detectors are suitable for spectroscopic measurements of β-type charged particles. Organic scintillators are also sensitive to X-ray or γ-ray type ionizing photons. However, the constituent materials of organic scintillators have low atomic numbers, which makes them less susceptible to photoelectric interactions compared to inorganic scintillators or semiconductor detectors.

[0031] The scintillator detector is covered with a thin, optically opaque jacket 14, for example, an 18 μm thick aluminum-coated polyethylene terephthalate (PET) film, to ensure opacity to ambient light. The thin thickness minimizes the probability of β-ray interaction. For the remainder of this text, the term β particle refers to β-particle.

[0032] The thickness e of the scintillator material is, for example, 4 mm. The diameter of the scintillator material in this example is 76 mm. Organic scintillator materials have the advantage of being relatively less sensitive to gamma rays due to their low atomic number. In addition, this type of scintillator limits the backscattering of β particles. Another advantage is the constant stability of the scintillator material's response to thermal fluctuations. The response with respect to the generated light intensity is stable from 0°C to 50°C when exposed to a given radiation, which meets the conditions for field use.

[0033] The measurement device includes a processing unit 20 configured to perform the spectral processing steps described below. The processing unit 20 is programmed to execute instructions stored in memory connected to the processing unit by a wired or wireless link. The processing unit 20 may, in particular, include a microprocessor.

[0034] According to one modification, the detector may include a semiconductor suitable for β-spectroscopy measurements. This could be, for example, silicon, specifically planar silicon.

[0035] The advantage of organic scintillator materials is that they can be manufactured in various shapes and sizes. When the object is a sample, the shape can be adapted to the shape of the sample.

[0036] According to one possibility, the detector 10 includes a movable screen 15, which functions as a shutter, - in a closed position between the detector 10 and the object 2 to be measured, - or in an open position that releases the space between the detector 10 and the object 2 as shown in FIG. 1 and is configured to be arranged.

[0037] The presence of the movable screen 15 is not essential.

[0038] In this example, the screen 15 is movable in a plane parallel to the detector material 11. The screen is made of, for example, aluminum, and its thickness is, for example, equal to 4 mm.

[0039] The present invention is based on the detection by the detector 10 of β - particles and optionally γ - particles emitted by the object 2. The object 2 137 is a βγ emitter such as Cs, or 14 C, 90 Sr, or 36 is an inspection object having a bulk or surface radioactivity due to a radionuclide that is a pure β - emitter such as Cl.

[0040] To have sufficient sensitivity to β - particles, the detector is preferably arranged at a short distance d from the object to be characterized. The distance d is preferably non - zero and is between a few millimeters and a few centimeters, for example 1 cm. This enables the formation of a βγ spectrum, and the small distance between the detector and the object increases the contribution of the β - particles to the spectrum.

[0041] According to one possibility, the contribution of gamma radiation to the βγ spectrum is limited by inserting a screen 15 between the detector 10 and the object 2. This makes it possible to obtain the gamma spectrum. The gamma spectrum is then subtracted from the βγ spectrum, thereby making it possible to form a β spectrum that is thought to represent only the β radiation emitted by the object.

[0042] Therefore, the present invention is implemented in either a βγ spectrum or a β spectrum. Generally, the present invention is implemented in a spectrum in which the contribution of β radiation is greater than 10%, or more specifically, 20% or 30% or more. Contribution means the number of interactions by β radiation that are taken into account when forming the spectrum.

[0043] In the example described below, the object is a concrete wall. A further difficulty with materials such as concrete is the presence of natural radioactivity. Natural radioactivity, though low, can complicate the interpretation of measurements, especially when dealing with low levels of artificial radioactivity, particularly below about 1 Bq / g. Among the naturally occurring radioactive elements potentially present in an object, for example... 40 K and 232 Th and 238 Examples of uranium decay products can be found. Such radioactive elements exist, for example, in objects made of concrete.

[0044] Another well-known difficulty in the field of nuclear measurements is the potential presence of background gamma noise, which originates from artificial radioactivity present in the detector environment outside the object being analyzed.

[0045] Figures 2A to 2C show known in the laboratory. 137 Cs or 90 This is the measured spectrum when an Sr point source is placed in front of the detector at a distance d of 10 mm. Figure 2A shows the spectrum. 137 The spectrum of a Cs source is shown, and that spectrum is, 137m In equilibrium with Ba 137 It is formed from beta and gamma radiation emitted by Cs. Figure 2B shows137 Figure 2B shows the spectrum of gamma radiation from Cs. Figure 2B shows the screen used to absorb the beta radiation emitted by the source, as previously described. 137 The result was obtained by placing the detector between the Cs source and the detector. Comparing Figure 2A and Figure 2B, it can be seen that a significant share of the spectrum is due to the contribution of beta radiation. This is because the distance between the detector and the source is small.

[0046] Figure 2C shows, 90 Sr source ( 90 The spectrum shown is that of a β particle in equilibrium with Y, and this spectrum is entirely attributable to the interaction of β particles within the detector. The radioactivity of the standard source was 9 kBq, and the acquisition time was 10 minutes.

[0047] Figure 3 shows a radioactive material with approximately 9 kBq. 137 Cs standard point source and 90 The spectra measured in the laboratory using both Sr standard point sources are shown.

[0048] The method described below is based on a spectrum like the one shown in Figure 3. - Identifying beta- or beta-gamma emitters present within the object being analyzed. - Estimating the distribution of each identified radionuclide within an object, i.e., its surface or bulk distribution, and possibly the thickness from the surface where each radionuclide exists. -and / or quantify the radioactivity of each identified radionuclide. This is the purpose.

[0049] The present invention is particularly advantageous when it is suspected that at least one pure β-emitting material is present in the object being analyzed. The inventors have observed that, due to the short free path of electrons in the detector material, in this case an organic scintillator, the obtained spectrum differs considerably depending on whether the radioactivity is distributed near the surface or deep within the object being analyzed. By using a low-density organic scintillator material, it may be possible to obtain different spectra depending on the depth to which the radioactivity is distributed within the object, especially in the case of radionuclides that emit high-energy β particles. Therefore, although inorganic scintillator materials, such as NaI, CsI, LaBr3, or semiconductors (Si, Ge), can be used, the use of organic scintillator materials allows for better differentiation between spectra corresponding to radioactivity distributed at various depths in the object being analyzed. The use of organic scintillators makes it possible to obtain a higher ratio between the β contribution and the γ contribution to the measured spectrum. In addition, organic scintillators can be sized to expose a larger area to the object being examined compared to scintillators made of inorganic materials. Furthermore, organic scintillators can be obtained in a wide variety of planar and non-planar shapes, thereby enabling good conformance to the shape of the object being inspected.

[0050] The use of organic scintillators is preferred but not essential for the implementation of the present invention. For example, a semiconductor detector or inorganic scintillator detector thin enough to limit sensitivity to gamma radiation can be used. For example, a semiconductor detector or inorganic scintillator based on a Si crystal may be used.

[0051] Effects of natural radioactivity One of the intended applications of this invention is low-level inspection of concrete civil engineering structures, aimed at checking whether the radioactivity level meets a predetermined target. This includes, for example, the so-called clearance level. 137 Cs or 90Sr can achieve radioactivity levels of 1 Bq / g). At such levels, the natural radioactivity of certain materials, such as concrete, can complicate the interpretation of the spectrum, as mentioned above.

[0052] The inventors modeled the effect of natural radioactivity on the spectrum measured by the device described with reference to Figure 1. Figure 4 shows one modeled shape, in which the detector is confined within a metal jacket 16. In Figure 4, a uniform radioactivity extends 1 cm thick from the surface of the object. 90 Sr and 137 The radioactivity of Cs was modeled, with the object facing the detector. The modeling was performed using the MCNP transport code (MCNP stands for Monte Carlo N particle). Figure 5A shows 1 Bq / g 90 Sr and 137 Cs radioactivity and 0.5 Bq / g, 0.03 Bq / g, and 0.02 Bq / g respectively 40 K, 232 Th, and 238 The modeled β spectrum, taking into account 3U radioactivity, is shown. The β spectrum was modeled considering only the interaction of β particles within the detector. Figure 5A shows the contribution of each radionuclide. The share of native radioactivity is 26% of the total β spectrum (the "totalnatβ" spectrum in Figure 5A). The value of 26% corresponds to the integral of the totalnat spectrum relative to the integral of the total spectrum.

[0053] Figure 5B shows 1 Bq / g 137 Cs radioactivity and 0.5 Bq / g, 0.03 Bq / g, and 0.02 Bq / g, respectively. 40 K, 232 Th, and 238 The modeled gamma spectrum, taking into account u-radioactivity, is shown. The gamma spectrum was modeled considering only the interactions of gamma particles within the detector. The share of natural radioactivity is 18% of the total gamma spectrum ("totalnatγ" spectrum in Figure 5B).

[0054] Table 1 shows,137 Cs radioactivity and 90 Under the condition that Sr radioactivity is equal and uniformly distributed over a concrete thickness of 1 cm, various specific 137 Cs and 90 This shows the share of natural radioactivity in the β and γ spectra relative to Sr radioactivity.

[0055] [Table 1]

[0056] The results shown in Table 1 indicate the share of natural radioactivity in the measured spectrum. If this share is considered too large, for example, for low levels of artificial radioactivity, the acquired spectrum may be corrected to remove the contribution of natural radioactivity. This can be done by estimating the contribution of the object's natural radioactivity to the measured spectrum (βγ spectrum or β spectrum). Natural radioactivity, assumed to be uniform throughout the object, - Analysis of a sample taken from an object, or analysis of a sample taken from another object considered representative. -or measurement, for example, measurement of an object or another object considered representative by high-resolution (e.g., germanium) gamma-ray spectroscopy. This could be the cause.

[0057] Next, in order to obtain a βγ spectrum or β spectrum in which the contribution of natural radioactivity can be ignored, the contribution of natural radioactivity to the spectrum is subtracted.

[0058] Figure 6 is a schematic diagram illustrating the main steps of the method according to the present invention.

[0059] Step 100: Place device 10 facing the object to be inspected and obtain the spectrum of radiation emitted by the object. This could be a βγ spectrum if the object contains gamma emitters, or a β spectrum if the object contains only pure β emitters.

[0060] Step 110: Obtain the γ spectrum and correct the obtained spectrum for the γ contribution.

[0061] In the optional step 110, a screen 15 is placed between the object and the detector. This makes it possible to obtain a spectrum representing the γ component of the βγ spectrum acquired in step 110. The β spectrum is formed by subtraction, as described below with reference to Figures 16A to 16C.

[0062] Step 110 is optional. It is performed if the contribution of gamma radiation to the acquired spectrum is too large.

[0063] Step 120: Correct for natural radioactivity.

[0064] In step 120, the contribution of spontaneous radioactivity to the β or βγ spectrum derived from step 110 or step 100 is estimated. This contribution is subtracted from the spectrum obtained in step 100 or from the spectrum derived from step 110. Step 120 is optional. It is performed if the contribution of spontaneous radioactivity to the obtained spectrum or the spectrum derived from step 110 is too large.

[0065] Following steps 100-120, the input spectrum Sp is either the spectrum acquired in step 100 or the spectrum formed after the optional corrections described in relation to steps 110-120. in The input spectrum Sp is obtained. in This forms the input data for the algorithm described in steps 130 and 140, which are performed by the processing unit 20.

[0066] Step 130: Identify the radioactive nuclide.

[0067] One important aspect of the present invention is the combination of two consecutive steps of input spectrum analysis, firstly, an identification step to identify radionuclides present in an object based on the input spectrum, without quantification. The objective is to identify the present radionuclides within a predetermined list. Following this first step, a second step is performed, based on the identifications made, with the aim of estimating the contribution of the identified radionuclides to the spectrum.

[0068] The identification step is performed using an identification algorithm, which is an artificial intelligence supervised learning algorithm. Each identification algorithm is intended to identify the presence of one radionuclide i in the input spectrum. Each identification algorithm is associated with a neural network, such as a convolutional neural network (CNN), which is associated with one radionuclide i. i For example, this could be a Bayesian convolutional network. Index i refers to the radionuclide to which the neural network is related. Such neural networks are generally used to process structured data such as images or histograms. A series of convolutional layers makes it possible to extract features from the input spectrum. The convolutional layers are fed into a multilayer perceptron type layer, which forms a convolutional neural network (CNN). i This makes it possible to determine the probability of the presence of a related radionuclide from features extracted by the convolutional layer.

[0069] The output of each identification algorithm is the probability that the radionuclide associated with the identification algorithm is present in the object. Therefore, there are as many identification algorithms as there are radionuclides present in the list that may be present in the object. Input spectrum Sp inThis forms the input data for each of these algorithms. For each radionuclide in the list, the output of each algorithm allows us to consider whether the radionuclide is identified or unidentified based on the input spectrum. Each identification algorithm is pre-trained using training spectra modeled or acquired in the presence and absence of the radionuclide to which each identification algorithm pertains, in an object or an equivalent object.

[0070] Each identification algorithm CNN i The output is the probability P that radioactive nuclide i is present in the object being examined. i Therefore, if the value exceeds a certain threshold, for example 0.5, it is considered that radioactive nuclide i is present in the object. Conversely, if the value is below the threshold, radioactive nuclide i is not present in the object.

[0071] Preferably, the convolutional neural network implements Monte Carlo dropout, which corresponds to deactivating specific neurons according to a probability law or randomly during the training phase and use of the neural network. Therefore, the network may derive different results each time it is applied to the same input data. This makes it possible to obtain measurement statistics.

[0072] Figure 7A schematically shows the detector model, and using this model, various training spectra were obtained using the MCNP code for various radionuclides, in this case. 14 C, 36 Cl, 90 Sr, 137 The data for Cs was established. The detector was assumed to be 10 mm from the standard source 2. Starting with four initial spectra derived from modeling the surface distribution of each radionuclide, the training database was created by combining spectra weighted by various randomly defined parameters, namely, the number of radionuclides in the spectrum, the proportion of each radionuclide, the number of interactions included in the spectrum, the minimum energy, and the maximum energy. In this way, 10 3~10 7 Taking into account the number of interactions (counts) involved, 500,000 training spectra were generated.

[0073] Figure 7B shows the training spectra used to parameterize the identification algorithms associated with various radionuclides. Figure 7B shows the proportion of each radionuclide in each spectrum.

[0074] Figures 8A and 8B show the first example of applying the identification algorithm. Figure 8A is 90 The measured β spectrum of a Sr source is shown. The spectrum shows the radionuclides. 14 C, 36 Cl, 90 Sr, 137 Each Cs was used 100 times by its respective identification algorithm. Figure 8B shows the probability of existence (Y-axis) defined for each radionuclide in the form of a box plot. 90 Sr has been systematically identified, while, 14 C, 36 Cl, and 137 The median probability of the existence of Cs is always less than 0.5.

[0075] Figures 9A and 9B show a second example of the application of the identification algorithm. Figure 9A shows a 5% 14 C and 95% 36 The β spectrum of a mixture containing Cl is shown. The spectrum shows the radioactive nuclide. 14 C, 36 Cl, 90 Sr, 137 Each of the identification algorithms defined for each Cs was used 100 times. Figure 9B shows the probability of existence (Y-axis) defined for each radionuclide in the form of a box plot. 14 C and 36 Cl has been systematically identified, on the other hand, 90 Sr and 137 The probability of Cs being present is always less than 0.5.

[0076] The performance of the identification algorithm is, 14C, 36 Cl, 90 Sr, and 137 The evaluation was performed using test spectra obtained by combining four experimentally measured spectra using different Cs sources. 10,000 test spectra were formed by combining the four measured spectra and varying the following characteristics: relative ratio, radioactivity, number of interactions in the spectrum, minimum energy, and maximum energy of the spectrum. The 10,000 test spectra were processed by an identification algorithm.

[0077] Tables 2, 3, 4, and 5 are confusion matrices. The first column represents the true value. The first row represents the result of the identification algorithm. 0 means the radionuclide is not present, and 1 means the radionuclide is present. The matrix values ​​correspond to the detection rates assigned to the radionuclide. The boxes corresponding to row 0 and column 1 correspond to false positives. The boxes corresponding to row 1 and column 0 correspond to false negatives. The boxes corresponding to row 1 and column 1, and row 0 and column 0, correspond to correct detections, and the values ​​determined by the identification algorithm correspond to the true value.

[0078] [Table 2]

[0079] [Table 3]

[0080] [Table 4]

[0081] [Table 5]

[0082] The results presented with reference to Figures 8B and 9B, as well as the confusion matrices shown in Tables 2 to 5, demonstrate the reliability of the identification achieved by subjecting the β or βγ spectrum to various algorithms, each algorithm being adapted to a single radionuclide to identify the radionuclide within the spectrum.

[0083] Depending on the output of each identification algorithm, it is determined whether or not each radionuclide associated with the identification algorithm is present in the object.

[0084] Step 140 Deconvolution

[0085] In this step, the spectrum is subjected to a deconvolution algorithm to extract the respective components associated with each previously identified radionuclide. One important aspect of this step is that the deconvolution is not performed blindly, but based on prior knowledge derived from the identification step.

[0086] The deconvolution algorithm is based on a deconvolution database that includes at least one detector response, i.e., one spectrum for each identified radionuclide modeled against known radioactivity and known distribution of the radionuclide within an object.

[0087] According to one possibility, the distribution of radionuclides in an object is known, and the distribution can be considered uniform, for example, if the object is a sample analyzed in a laboratory that has undergone homogenization. If the measurement is performed on an activated object, the distribution of radionuclides can be determined by modeling the neutron flux to which the object is exposed. If the object is made of a non-porous material, such as a metal, the radioactivity can be assumed to be surface radioactivity.

[0088] When an object is made of a porous material, such as concrete, various distributions of radioactivity are possible. For example, radioactivity may have a gradient that decreases from the surface of the object. For example, the gradient may take the form of an exponential function that decreases with depth. This type of profile is typical of contamination transport. If z is the depth within the object from the surface of the object, the distribution of radioactivity A(z) as a function of depth can be considered to have the following forms: A(z)=A(0)e -λz (1)

[0089] A(0) is the surface radioactivity, and λ is the shape coefficient of the exponential function. λ, whose unit is the reciprocal of the unit of length, sets the depth zmax to which the radioactivity is distributed within the object. Depth zmax is the depth at which the radioactivity is distributed relative to the surface radioactivity A(0).

[0090]

number

[0091]

number

[0092] The definition of λ or zmax makes it possible to define the volume in which radioactivity is assumed to be concentrated.

[0093] Preferably, the database includes, for each identified radionuclide, various modeled spectra corresponding to different distributions of the radionuclide within the object and the radioactivity of the identified radionuclide. Thus, a database is obtained that contains spectra representing various distributions of radioactivity within the object for various radionuclides. For example, taking into account exponential gradients as described in (1) or (2), the database includes modeled spectra corresponding to various parameters λ and zmax for various nuclides.

[0094] The deconvolution algorithm is performed using representative spectra corresponding to the radionuclides identified in step 130. Prior identification of radionuclides allows for the selection of modeled spectra corresponding to each identified radionuclide in the deconvolution database. Deconvolution can then be performed using a small number of modeled spectra, limited to only the identified radionuclides. This avoids deconvolution errors, particularly false positives, i.e., the conclusion that a radionuclide is present when it is not. Since a limited number of radionuclides are considered, various distribution profiles may be considered for each identified radionuclide; for example, in the case of a gradient with a decreasing exponential function as described in (1), various radioactivity depths zmax are possible. Therefore, deconvolution not only allows for the estimation of the radioactivity of each selected radionuclide, but also allows for the estimation of the depth to which the radionuclide extends within the object. It should be noted that the depth associated with one radionuclide may differ from the depth associated with another radionuclide.

[0095] The advantage of combining the identification step with the deconvolution step is that in the deconvolution step, only spectra representing the identified radionuclides can be selected. This makes it possible to take into account various distributions for each radionuclide. Without the selection of radionuclides, deconvolution that takes into account various distributions would be more dangerous because it would require considering too many spectra.

[0096] A deconvolution database may include spectra modeled for various depths of radioactivity, as well as spectra obtained by interpolation between these modeled spectra, such as spectra obtained by interpolation between two modeled depths of radioactivity.

[0097] To perform deconvolution, the inventors implemented a method based on the definition and maximization of a likelihood function. The estimated distribution of radioactivity for each radionuclide corresponds to the distribution that maximizes the likelihood function. The likelihood function can be maximized using the MLEM algorithm, as described in the prior art. Deconvolution may also be performed by other methods, such as regression, or by supervised learning algorithms, such as neural networks. In this case, the output of the neural network corresponds to the contribution of each radionuclide to the input spectrum.

[0098] When using the MLEM method, the spectrum is deconvolved over multiple iterations, as the iterations progress, by fitting multiple spectra from a deconvolution database corresponding to various identified radionuclides to the input spectrum, and finding the combination that best fits the input spectrum. The iterations continue until a convergence criterion is met, which may be minimizing a cost function that represents the difference between the input spectrum and the spectrum obtained by combining the spectra from the deconvolution database for the identified radionuclides. The cost function can be calculated for all of the spectrum's energy or for a given region of interest. The region of interest is determined in advance, for example, for each radionuclide, depending on the variability of the detector response as a function of depth; see Figures 10C and 10D below.

[0099] The present inventors have found that the radioactivity is 1 Bq / g and the radionuclide is 90 Assuming that the material consists solely of Sr, various spectra corresponding to different radioactivity depths zmax in the range of 0.1 mm to 500 mm were modeled using exponential gradients defined by (1) and (2). The modeled object was a concrete wall.

[0100] Figure 10A shows various modeled spectra. Figure 10B shows the modeled spectra normalized by integration of each spectrum. A spectrum representing a uniform distribution of radioactivity passing through a wall with a thickness of 500 mm was also modeled. The spectrum corresponding to uniform radioactivity was normalized by integration. Figure 10C shows the normalized spectrum of Figure 10B divided by the spectrum corresponding to uniform radioactivity normalized by integration.

[0101] As the depth of radioactivity increases, the detector response tends to approach the response corresponding to a uniform profile. Figure 10C shows two dashed lines corresponding to deviations of +5% or -5% from a uniform profile. This deviation corresponds to the minimum acceptable deviation, i.e., the minimum deviation required to ensure a good assessment of the radioactivity distribution. Up to zmax = 40 mm, the profile shown in Figure 10C deviates from the ±5% band corresponding to uniformly distributed radioactivity. Therefore, 90 In the case of Sr, the shape of the β spectrum allows us to determine the depth of maximum radioactivity in the object, up to a maximum zmax = 40 mm, under the assumption of a predetermined decreasing radioactivity gradient. Figure 11 shows Radioactivity decreases exponentially down to a maximum depth of -40 mm zmax. - Radioactivity uniformly distributed throughout the entire wall A comparison of the modeled spectra, taking this into account, is shown.

[0102] The two spectra overlap, which confirms the conclusion drawn based on Figure 10C.

[0103] Figure 10D is equivalent to Figure 10C, 90 Instead of Sr radioactivity 137 This is a diagram of Cs radioactivity. Figure 10D was obtained by modeling the βγ spectrum. Figure 10D shows two dashed lines corresponding to deviations of +5% or -5% from a uniform profile. Up to zmax = 500 mm, the profile shown in Figure 10D is uniformly distributed. 137Deviates from the ±5% band corresponding to the Cs radioactivity. Therefore, 137 In the case of Cs, the shape of the βγ spectrum makes it possible to determine the depth zmax of the maximum radioactivity within an object equal to at least 500 mm under the assumption of a predetermined decreasing radioactivity gradient.

[0104] Figures 10C and 10D make it possible to define the spectral regions of interest used to calculate the cost function. For example, it is possible to take into account regions where high variability as a function of depth is observed.

[0105] Figures 12A to 12D show the fit after various numbers of iterations of the spectrum (solid line) derived from the MLEM algorithm to the spectrum derived from the measurement ("Mes" - dashed line). The measured spectrum corresponds to the spectrum of the surface distribution consisting of 92% 137 Cs and 8% 90 Sr, and the measured spectrum contains 10,000 counts, i.e., 10,000 detection pulses. Figures 12A to 12D correspond to 1, 10, 100, and 10,000 iterations, respectively. It can be seen that as the iteration progresses, the spectrum reconstructed by the MLEM algorithm approaches the measured spectrum.

[0106] Table 6 contrasts the respective percentages determined from each spectrum reconstructed by MLEM for each number of iterations. The second row shows the actual percentages. It can be seen that the percentages approach the actual values as the iteration progresses.

[0107]

Table 6

[0108] The inventors 90 Sr and 137 For various depths zmax of Cs, assuming a decreasing exponential gradient, 90 Sr radioactivity and 137The MLEM deconvolution algorithm was applied to various ratios of Cs radioactivity. The configurations are shown in Table 7.

[0109] [Table 7]

[0110] The β, γ, and βγ spectra were modeled for each of these configurations using MCNP, and then MLEM deconvolution was applied. The deconvolution results are compared in Table 8 (Configuration 1), Table 9 (Configuration 2), Table 10 (Configuration 3), and Table 11 (Configuration 4).

[0111] In each table, the following are shown: zmax (in mm), the standard deviation related to the determined zmax (mm), and radioactivity A( 137 Cs) or A( 90 The Sr (unit Bq) and the standard deviation associated with the determined radioactivity (unit Bq) are given.

[0112] [Table 8]

[0113] [Table 9]

[0114] [Table 10]

[0115] [Table 11]

[0116] The results shown in Tables 8 to 11, in particular, regarding the βγ spectrum, 137 When taken into consideration for Cs, and the β spectrum 90When Sr is taken into account, the reliability of the algorithm is demonstrated. Configuration 4( 90 For Sr with zmax = 400 mm), as described with reference to Figure 10C, 90 corresponds to a depth of radioactivity exceeding the maximum depth of 40 mm defined for Sr. 137 Taking into account the β spectrum to quantify the depth or radioactivity of Cs radioactivity 90 Sr radioactivity 137 can cause errors when the Sr radioactivity is at least three times greater than the Cs radioactivity, which is true for Configurations 2 and 4.

[0117] Therefore, 90 Sr and 137 In the case of a mixture of Cs, it seems optimal to take into account the βγ spectrum, keeping in mind that these two radionuclides are fission products and are frequently encountered together in spent fuel treatment facilities or in the case of radioactivity contamination related to spent fuel.

[0118] The trials conducted together confirm the ability of the present invention to quantify the radioactivity of the identified radionuclides and estimate the depth of radioactivity of each of these radionuclides.

[0119] Comparison with the case where the identification algorithm is not implemented The spectra shown in Figures 8A and 9A were used as the input spectra for the deconvolution algorithm without prior identification of the radionuclides, i.e., without implementing the identification algorithm. 90 Regarding the spectrum corresponding to Sr radioactivity (see Figure 8A), the implementation of the MLEM algorithm resulted in the following radioactivity percentages, i.e., 0% of 14 C, 4% of 36 Cl, 90% of 90 Sr, 6% of 137 Cs. 14 C and 36 Regarding the spectra corresponding to Cl radioactivity (see Figure 9A), the implementation of the MLEM algorithm resulted in the following radioactivity percentages, i.e., 5% of14 C, 90% 36 Cl, 0% 90 Sr, 5% 137 Cs was brought about.

[0120] experimental trial Steps 110–140 were performed on graphite samples S taken by coring laterally into horizontal channels CH intended for loading and unloading fuel into the core of a gas-cooled graphite-moderated reactor. Each sample extended between a side F1, called the "channel side," adjacent to the fuel channel and facing inward, and the opposite side F2, called the "core side," facing towards the graphite moderator of the reactor. Figure 13 schematically shows the channels CH and the location of the core from which the samples S were extracted.

[0121] Each graphite sample was cylindrical, 15 mm in diameter and 20 mm thick. The detector was positioned 25 mm away from one side of each sample.

[0122] The training spectrum and deconvolution database includes the major radionuclides that are likely to be measured, i.e., -Activation products: 14 C, 36 Cl, 60 Co, 133 Ba, 152 EU, 154 EU -Possible fission products: 137 Cs and 90 Sr It was created with that in mind.

[0123] The objective of the analysis was to confirm the presence of fission products indicating the possibility of cladding rupture. To assess potential contamination by fission products, four identification neural networks were used, each to identify fission products. 137 Cs and 90 The parameters were parameterized to handle distinct levels of minimum radioactivity of Sr. The minimum radioactivity is 0.1 Bq·g, respectively. -1 , 1 Bq·g-1 , 10 Bq·g -1 , and 100 Bq·g -1 It was equivalent to this. Hereafter, these neural networks will be referred to as A, B, C, and D, respectively.

[0124] Four networks were trained using 50,000 modeled spectra, with 50% of the spectra considered to include contributions from fission products and 50% considered not to include contributions from fission products. For spectra that include contributions from fission products, 137 Cs / 90 The Sr ratio was randomly selected to be between 0.25 and 4. The depth of the distribution was also randomly selected so that zmax was between 0.1 mm and 5 mm. The shallow depth was justified by the fact that contamination by fission products was assumed to be dry. The presence of gamma-ray background noise was also modeled, and its level was set based on a reference measurement that was thought to represent the background noise during the measurement. For each training spectrum, the measurement time was defined. To simulate the statistical variation of background noise, a contribution of random background noise was added to the modeled spectrum, taking the measurement time into account. Activation products ( 14 C, 36 Cl, 60 Co, 133 Ba, 152 EU, 154 The radioactivity of Eu was established based on graphite activation calculations.

[0125] 80% of the spectrum was used for training. 10% of the spectrum was used to fit the model for validation. 10% of the spectrum was used for testing.

[0126] Several samples taken along two different fuel channels were analyzed. Each spectrum measured in the graphite sample was analyzed 100 times using four neural networks. For each sample, analysis was performed on both the core side and the channel side.

[0127] Figures 14A–14D show the outputs from four neural networks for various samples, as a function of distance from the central part of the reactor, when spectra measured on either the core side (F2 side) (Figures 14A and 14C) or the channel side (F1 side) (Figures 14B and 14D) are used. The y-axis corresponds to the probability of contamination being present (between 0 and 1), calculated taking into account 100 analyses for each neural network. Figures 14A–14D are derived from 100 outputs of the neural networks that yielded significant probabilities. 137 Cs and 90 This shows the probability of Sr's presence. The arrows identify neural networks where a significant probability was obtained.

[0128] The x-axis corresponds to the sample position (in cm) relative to the origin, which corresponds to the central part of the reactor. Figures 14A and 14B show the probabilities obtained as a function of the outputs of four sample neural networks (used 100 times) for samples taken from the first channel, where the reference was 36-17C. Figures 14C and 14D show the probabilities obtained as a function of the outputs of four neural networks for samples taken from the second channel, where the reference was 19-13C.

[0129] Only one sample taken from channel 36-17C was found to be likely to be low. 137 Cs+ 90 It is determined that it possessed Sr radioactivity. In this channel, neural networks C and D conclude that it is not radioactive.

[0130] In channels 19-13C, the majority of the analysis of the spectrum measured on the channel side is,137 Cs+ 90 We conclude that Sr exists.

[0131] In one sample taken from the channel side at a position 15 cm along channel 36-17C, 137 Cs+ 90 Sr radioactivity was detected by two neural networks (A and B). The spectra were deconvoluted, with and without considering the output of the identification neural network. Table 12 shows the results. 137 Cs and 90 The results obtained by performing the deconvolution algorithm with and without considering the presence of Sr are compared.

[0132] Table 12 shows the evaluation of radioactivity A (in Bq) and associated relative uncertainty (Σ) for various radionuclides when the deconvolution algorithm is performed with and without prior implementation of the identification neural network (I). For this sample, neural networks A and B, activated 100 times, 137 Cs and 90 We concluded that Sr was present in 50% and 70% of the cases, respectively (see Figure 14B). The last column shows the relative difference between the estimated radioactivity in (I) and (II). 137 Cs and 90 By taking Sr into account, it becomes possible to adjust the estimated radioactivity value for the activation products.

[0133] [Table 12]

[0134] The deconvolution algorithm was performed on a sample taken from 120 cm along channels 19-13C on the channel side. For this sample, neural networks A-D, activated 100 times, 137Cs and 90 We concluded that Sr was present in 100% of each case (see Figure 14D).

[0135] The deconvolution algorithm resulted in minimization of the cost function for the following: - Equivalent to 246±46Bq and 296±31Bq, respectively. 90 Sr and 137 Cs radioactivity. - Less than 200 μm and less than 500 μm, respectively. 90 Sr and 137 The depth of Cs radioactivity. This confirms the hypothesis of surface contamination.

[0136] The samples were characterized by high-resolution gamma-ray spectroscopy, which is considered a reference method. 137 The Cs radioactivity was 358 Bq, with a relative uncertainty of 30%.

[0137] Assuming a margin of error, the βγ spectroscopy measurement is consistent with the reference method.

[0138] Figure 15 shows the measured βγ spectrum and 137 Cs and 90 Input spectra of Sr and the activation product (act) Sp in This shows the contributions to the deconvolution algorithm, and these contributions arise from the deconvolution algorithm.

[0139] As previously shown, in the case of strong gamma radiation, the present invention can be implemented in the beta spectrum. To do this, the following is measured: - βγ spectrum of an object, - Gamma spectrum with the screen inserted between the detector and the object.

[0140] The β spectrum can sometimes be obtained by subtracting the γ spectrum from the βγ spectrum, taking into account differences in acquisition periods.

[0141] Figures 16A to 16C illustrate this possibility. Figure 16A shows the measured βγ spectrum. Figure 16B shows the measured γ spectrum. Figure 16C shows the β spectrum calculated by taking the difference between the βγ spectrum and the γ spectrum.

[0142] While we have described spectra obtained by non-destructive methods, this method can be generalized to the analysis of spectra where the β component cannot be ignored. This can then be done using laboratory methods, such as liquid scintillation. In this case, the object is the sample placed in front of the β spectrometer.

Claims

1. A method for characterizing an object, wherein the object contains at least one radionuclide that emits beta radiation, and the method -a) A step of positioning a detector (10) facing the object (2), wherein the detector is configured to acquire a spectrum representing the distribution of energy emitted within the detector by the radiation emitted by the object, -b) The steps of detecting the radiation emitted by the object during the acquisition period using the detector and obtaining the spectrum of the detected radiation, -c) The step of forming an input spectrum from the spectrum of the radiation detected by the detector, which includes a β component corresponding to the energy distribution emitted by the β radiation within the detector, -d) A step of applying an identification algorithm related to a radionuclide to the input spectrum, wherein the identification algorithm is configured to determine the presence of the radionuclide related to the identification algorithm within the object, and step d) is repeated for various radionuclides by performing different identification algorithms. -e) A step of identifying each radionuclide contained in the object in accordance with step d), -f) A step of applying an input spectrum deconvolution algorithm to estimate the contribution of each radionuclide identified in step e) to the input spectrum, -g) For each radionuclide identified in step e), the step of estimating the radioactivity and / or the depth to which the radionuclide extends within the object from the contribution of the radionuclide to the input spectrum estimated in step f); Includes, A method wherein steps d) to g) are performed by a processing unit based on the input spectrum.

2. The method according to claim 1, wherein in step d), each identification algorithm is an artificial intelligence identification algorithm associated with each radionuclide, and at least two different radionuclides are associated with two different respective identification algorithms.

3. The method according to claim 2, wherein each identification algorithm is a neural network.

4. The method according to any one of claims 1 to 3, wherein the deconvolution algorithm is based on a deconvolution database containing at least one spectrum representing each radionuclide identified in step e).

5. - The deconvolution database includes, for a given radionuclide, various spectra representing various distributions of the radionuclide within the object, The method according to claim 4, wherein step g) includes determining the distribution of the radioactive nuclide within the object.

6. - The deconvolution database includes, for a given radionuclide, various spectra representing various depths of the radionuclide within the object from the surface of the object facing the detector, The method according to claim 5, wherein step g) includes determining the depth to which the radioactive nuclide extends within the object.

7. The method according to any one of claims 1 to 3, wherein at least one radionuclide associated with the identification algorithm is a pure beta emitter.

8. The method according to any one of claims 1 to 3, wherein the detector includes an organic scintillator material for detecting the radiation emitted by the object.

9. The method according to any one of claims 1 to 3, wherein the detector comprises a volume of an inorganic scintillator or semiconductor having a thickness of less than 10 mm, positioned facing the object, and the thickness is measured in a direction perpendicular to the object.

10. The object has natural radioactivity, and step c) - A step of estimating the spectrum of the natural radioactivity of the object, - A step of subtracting the spectrum of the natural radioactivity of the object from the spectrum obtained in step b) in order to form the input spectrum. The method according to any one of claims 1 to 3, including

11. The detector includes a movable screen (15) configured to be inserted between the detector and the object, and the method is - A step of acquiring a background spectrum with the screen inserted between the detector and the object, - The background spectrum is subtracted in step c) from the spectrum obtained in step b) to form the input spectrum. The method according to any one of claims 1 to 3, including

12. A detection device (1) comprising a detector (10) configured to acquire a spectrum of beta radiation emitted by an object (2), wherein the spectrum represents the distribution of energy emitted in the detector during the interaction of ionizing radiation in the detector, and the device comprising a processing unit (20) configured to carry out steps d) to f) of the method according to any one of claims 1 to 11.

13. The detection device according to claim 12, wherein the detector includes an organic scintillator material for detecting the radiation emitted by the object.

14. The device according to claim 12, wherein the detector comprises a volume of an inorganic scintillator or semiconductor with a thickness of less than 10 mm.

15. The device according to any one of claims 12 to 14, wherein the detector includes a movable screen (15) configured to be inserted between the detector and the object.

Images