METHOD FOR DETERMINING AN ELEMENT CONCENTRATION OF AN EDS/WDS SPECTRUM OF AN UNKNOWN SAMPLE AND A CORRESPONDING DEVICE

Abstract

The present invention discloses method and device for determining an element concentration of an EDS/WDS spectrum of an unknown sample. The method comprises performing a preliminary quantification of the EDS/WDS spectrum of the unknown sample and identify a plurality of elements in the unknown sample; identify at least one pre-stored standard sample including the plurality of elements; determine, for each element of the plurality of elements, a similarity score for the corresponding element in each identified standard sample; select, for each element of the plurality of elements, the one standard sample among the at least one standard sample by using the determined similarity score and identify the concentration of the corresponding element in the selected standard sample; and perform quantification of the EDS/WDS spectrum of the unknown sample by using, for each element of the plurality of elements, the identified concentration of the respectively selected standard sample.

Claims

1. A method for determining an element concentration of an EDS/WDS spectrum of an unknown sample, comprising the steps of: perform a preliminary quantification (S100) of the EDS/WDS spectrum of the unknown sample and identify a plurality of elements in the unknown sample; identify (S200) at least one pre-stored standard sample including the plurality of elements; determine (S300), for each element of the plurality of elements, a similarity score for the corresponding element in each identified standard sample; select (S400), for each element of the plurality of elements, the one standard sample among the at least one standard sample by using the determined similarity score and identify the concentration of the corresponding element in the selected standard sample; and perform quantification (S500) of the EDS/WDS spectrum of the unknown sample by using, for each element of the plurality of elements, the identified concentration of the respectively selected standard sample.

2. The method for claim 1, wherein the determining the similarity score comprises the steps of: determine (S310) at least one among an atomic number correction factor Z.sub.u,k, an absorption correction factor A.sub.u,k and a fluorescence correction factor F.sub.u,k for each element of the plurality of elements of the preliminarily quantified unknown sample; determine (S320) the same one among the atomic number correction factor Z.sub.s,k, the absorption correction factor A.sub.s,k and the fluorescence correction factor F.sub.s,k, for the corresponding elements of the identified standards; and calculating (S330), for each element of the plurality of elements, a square deviation between the determined correction factor of the element of the unknown sample and the corresponding one of the standard sample.

3. The method for claim 2, wherein determining the similarity score comprises calculating (S340) a sum of square deviations between the respective correction factors Z.sub.u,k, A.sub.u,k and F.sub.u,k of the element of the unknown sample and the respective correction factors Z.sub.s,k, A.sub.s,k and F.sub.s,k of the standard sample.

4. The method for claim 2, wherein the determining of the similarity score further includes using the concentration of the element in the standard sample.

5. The method for claim 4, wherein determining the similarity score further includes calculating (S350) a term log(c.sub.s,k) and subtracting the penalty term from the calculated square deviation, wherein c.sub.s,k corresponds to the concentration of the element in the standard sample.

6. The method for claim 5, wherein determining the similarity score includes setting (S360) a relative weight between the calculated term log(c.sub.s,k) and the calculated square deviation.

7. The method for claim 6, wherein determining the similarity score includes calculating the term:
S=w.sub.1.Math.[(Z.sub.u,k−Z.sub.s,k).sup.2+(A.sub.u,k−A.sub.s,k).sup.2+(F.sub.u,k−F.sub.s,k).sup.2]−w.sub.2.Math.log c.sub.s,k, wherein w.sub.1>0 refers to a preset first weight factor, w.sub.2>0 refers to a preset second weight factor.

8. The method for claim 1, wherein the step of performing the preliminary quantification includes a standardless quantification in which concentrations are initially assumed and iterated to reach the preliminary quantification of the EDS/WDS spectrum of the unknown sample.

9. The method for claim 1, wherein the step of performing the preliminary quantification includes a standard quantification in which concentrations according to a standard sample are selected and iterated to reach the preliminary quantification of the EDS/WDS spectrum of the unknown sample.

10. The method for claim 1, wherein after the step of performing quantification (S500) of the EDS/WDS spectrum of the unknown sample, the method comprises the step of determining the sum of concentrations (S600) of the elements of the quantification result, and accept the quantification result, when the sum of concentrations is within a threshold interval around 100%.

11. The method for claim 10, wherein: the determining the similarity score comprises the steps of: determine (S310) at least one among an atomic number correction factor Z.sub.u,k, an absorption correction factor A.sub.u,k and a fluorescence correction factor F.sub.u,k for each element of the plurality of elements of the preliminarily quantified unknown sample; determine (S320) the same one among the atomic number correction factor Z.sub.s,k, the absorption correction factor A.sub.s,k and the fluorescence correction factor F.sub.s,k, for the corresponding elements of the identified standards; and calculating (S330), for each element of the plurality of elements, a square deviation between the determined correction factor of the element of the unknown sample and the corresponding one of the standard sample; the determining of the similarity score further includes using the concentration of the element in the standard sample; the determining of the similarity score further includes calculating (S350) a term log(c.sub.s,k) and subtracting the penalty term from the calculated square deviation, wherein c.sub.s,k corresponds to the concentration of the element in the standard sample; and the method comprises to repeats the steps of S300, S400 and S500 for adjusted weights, when the sum of concentrations is within a threshold interval around 100% until the sum of concentrations of the elements of the quantification result is within the threshold interval around 100%.

12. A computer program comprising instructions which, when executed by a computer, cause the computer to carry out the method for claim 1.

13. A device (10) for determining an element concentration of an EDS/WDS spectrum of a sample, comprising at least one processor (20) operatively connected to a storage (30), wherein the at least one processor (20) is configured to perform the method for claim 1.

14. A spectroscopic system (100) for obtaining an EDS/WDS spectrum of an unknown sample, the system (100) including a device (10) according to claim 13.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Features will become apparent to those of ordinary skill in the art by describing in detail exemplary embodiments with reference to the attached drawings in which:

[0051] FIG. 1 illustrates a method for determining an element concentration of an EDS/WDS spectrum of an unknown sample;

[0052] FIG. 2 illustrates a method for determining a similarity score according to embodiments of the invention;

[0053] FIG. 3 illustrates a device for determining an element concentration of an EDS/WDS spectrum of an unknown sample; and

[0054] FIG. 4 illustrates a schematic system for generating an EDS/WDS spectrum of an unknown sample according to an embodiment of the invention.

DETAILED DESCRIPTION OF THE INVENTION

[0055] Reference will now be made in detail to embodiments, examples of which are illustrated in the accompanying drawings. Effects and features of the exemplary embodiments, and implementation methods thereof will be described with reference to the accompanying drawings. In the drawings, like reference numerals denote like elements, and redundant descriptions are omitted. As used herein, the term “and/or” includes any and all combinations of one or more of the associated listed items. Further, the use of “may” when describing embodiments of the present invention refers to “one or more embodiments of the present invention.”

[0056] It will be further understood that the terms “include,” “comprise,” “including,” or “comprising” specify a property, a region, a fixed number, a step, a process, an element, a component, and a combination thereof but do not exclude other properties, regions, fixed numbers, steps, processes, elements, components, and combinations thereof.

[0057] FIG. 1 illustrates a method for determining an element concentration of an unknown EDS/WDS spectrum according to an embodiment of the invention.

[0058] A spectrometric system 100, cf. the schematic FIG. 4, may detect energy- or wavelength dispersive secondary particles 130 as for example X-rays emitted from an unknown sample 150. The detected photons can be represented by an EDS/WDS spectrum which includes intensities (i.e. the number of photons) emitted by the unknown sample 150 according to the individual elements therein. The collected EDS/WDS spectrum thus includes information on the concentrations of the elements of the sample 150. The EDS/WDS spectrum may be obtained by a spectroscopic system 100 detecting frequencies/wavelengths emitted by the sample 140 in response to an excitation process through primary particles 120, i.e. for example electrons or X-rays. The EDS/WDS spectrum may also be a pre-stored spectrum in a storage medium to be used for quantification according to the method as explained in the following.

[0059] According to step S100 a preliminary quantification of the EDS/WDS spectrum of the unknown sample 150 is performed. The preliminary quantification also includes to identify a plurality of elements in the EDS/WDS spectrum of the sample. The identification of the elements can be performed by comparing the unknown spectrum with pre-known spectra in terms of the position, i.e. the energies and wavelengths, of the peaks therein and associate the peaks to particular elements in the sample. The comparison and the detection of the plurality of elements can be performed by an element identification algorithm. The number of elements is not restricted to a particular number and depends on the element composition of the unknown sample 150.

[0060] The quantification of the EDS/WDS spectrum includes a preliminary determination of the concentrations of the elements in the unknown sample 150. In an embodiment, the preliminary quantification of the concentrations can be performed by standardless quantification. In a standardless quantification initial concentrations are assumed or guessed and with these numbers an iteration process is executed to preliminarily determine the quantification of the EDS/WDS spectrum of the unknown sample 150. The result is a set of data including for example the preliminary concentrations of the unknown sample 150 and/or correction factors, i.e. the atomic number correction factor Z.sub.u,k the absorption correction factor A.sub.u,k and the fluorescence correction factor F.sub.u,k for each element k in the unknown sample 150. The standardless quantification is fast and reduces computational time.

[0061] The preliminary quantification can also be obtained by using standards, i.e., applying a standard quantification. For example, a storage may be provided in which various standards according to the elements contained in the respective sample are pre-stored. For example, various steel standards including elements of steel may be stored in the storage. Other standards may be referred to minerals, glasses, organic samples or semiconductor materials or particular subclasses thereof. The standard can be selected to reach the preliminary quantification of the EDS/WDS spectrum of the unknown sample. The result is a set of data including for example the preliminary concentrations of the unknown sample 150 and/or correction factors, i.e. the atomic number correction factor Z.sub.u,k, the absorption correction factor A.sub.u,k and the fluorescence correction factor F.sub.u,k for each element k in unknown sample 150. For example, each of the standard samples may be used for preliminary quantification and the best result is taken as the preliminary result. In the way of standard-based quantification, a more reliable and more accurate preliminary quantification can be achieved compared to the standardless preliminary quantification.

[0062] In step S200 at least one pre-stored standard spectrum including the plurality of elements is identified. Thus, from the database, the standards including the plurality of elements are identified. For example, as shown in table 3 below, four different standard samples (steel A, B, C and D) for a steel sample including the same elements are identified. Each standard sample comprises a particular composition of elements having concentrations, i.e. mass fractions of the total mass of the sample, listed in table 3 for each steel standard. The example serves only for illustration and the invention may allow a larger number of standards for a given set of identified elements. For example, the quantification can be improved when more standards are provided for each identified element to improve the accuracy of the selections.

[0063] In step S300, for each element of the plurality of elements, a similarity score for the corresponding element in each identified standard sample is determined. The score is evaluated by a comparison process between the standard sample and the preliminarily quantified unknown sample 150 for each element. The score quantifies the similarity between each corresponding element of the standard sample and the unknown sample 150.

[0064] An evaluation of a particularly developed similarity score is illustrated in the description with respect to FIG. 2 as described below. In an example, cf. table 3 below, for the element Si the similarity score of the same element Si of standard steel B is better than the similarity score for standard steel D. This is indicated by the underlining of the corresponding concentration. As further example in table 3, for the element Mn in the unknown sample, the standard sample steel C has better score than steel B indicated by the underlining. Thus, the score is element-based and depends on the chosen element. For different elements of the unknown sample 150, the scores differ between the standards. The table 3 indicates the concentrations of the elements in each standard and discloses the best score for each element by underlining.

[0065] In step S400, the method includes to select, for each element of the plurality of elements, the one standard among the at least one standard sample using the calculated similarity score. Additionally, the concentration of the corresponding element in the selected standard sample is identified. Therefore, element-wise, a standard of the various standards is determined and the concentration of the corresponding element is selected.

[0066] This is illustrated in table 3 below. For example, as demonstrated in table 3, for element Mo steel A has best similarity score and thus the concentration of 1.4 is identified for the element Mo. In another example, for the elements Si, Cr, and Cu, steel B has best similarity score and the element concentrations 0.4, 16.1 and 2.0 are identified for these elements. For the elements Mn and Ni the standard steel C has the best score and the respective concentrations 3.1 and 9.9 are identified for these elements. For elements Fe and Co steel D has best similarity score and the concentrations of 50.6 and 0.03 are identified for these elements. Thus, for each element, one standard sample is selected among the plurality of standard samples which has the best score and the corresponding pre-known concentration is identified from that selected standard sample.

[0067] In step S500 quantification of the EDS/WDS spectrum of the unknown sample 150 is determined by using, for each element of the plurality of elements, the identified concentration of the respectively selected standard sample for each element. Thus, the concentrations of the best standards according to the element-wise score are used as input for the quantification process.

[0068] The input concentrations are therefore not according to one of the standards but the method selects the best standards (and their concentrations) for each element to be the input for the quantification and to obtain the concentrations of the unknown sample 140. An iteration scheme can then be used, which is initialized by the identified concentration of the selected standard for each element. The result of the element concentrations is obtained, when a termination condition of the iteration process is met.

[0069] Thereby, from a set of identified standards, an optimal standard for each element, i.e. element-wise, is automatically identified by using a determined score value so that optimized concentrations are input for the quantification process. Thus, the accuracy of the quantification results is improved. The algorithm to reach the quantification results, i.e. each of the concentrations, in the last step therefore does not have to rely on a particular standard but selects element-wise the optimal standard according to the determined score.

[0070] In the following tables 1-3, the quantification results are compared between the state of the art selections (as shown in table 1 and 2) using a particular standard and the element-wise selection of standards according to the present invention in table 3.

[0071] In all tables 1-3, the spectrum is a steel sample as an example. The steel samples comprise the elements Si, Cr, Mn, Fe, Co, Ni and Mo. The values in the table refer to concentrations, i.e. mass fraction or mass concentration in % of total mass, of the respective elements. In the present example, four steel standard samples A, B, C and D are identified in the storage. The concentrations of the unknown steel refer to the result of the quantification process, cf. step S500, for the different inputs in table 1, 2 and 3. The concentrations of the target values refer to the “true” values of the sample 150 to compare the accuracy of the predictions of the quantification process.

[0072] In the comparative prior art example according to table 1, element concentrations of the standard steel A are used as input for the quantification process to determine the concentrations of the unknown sample. As can be noticed, the quantification result underestimates the concentrations of the elements. This becomes apparent from the sum of concentrations being below 100% or by comparing with the target values in table 3.

TABLE-US-00001 TABLE 1 (comparative prior art example according to a first embodiment) Spectrum Si Cr Mn Fe Co Ni Cu Mo Sum Steel A 0.3  1.2 0.5 96.4 0.01 0.1 0.1 1.4 99.9 Unknown steel 1.0 17.5 0.9 65.7 0.01 6.4 3.4 0.5 95.4

[0073] In comparative prior art of table 2 the standard sample of steel D is used as input for the quantification. As can be noticed, the quantification result overestimates the concentrations of the elements. This can be derived from the sum of concentrations being above 100% or by comparing with the target values in table 3.

TABLE-US-00002 TABLE 2 (comparative prior art example according to a second embodiment) Spectrum Si Cr Mn Fe Co Ni Cu Mo Sum Steel D 1.3 25.0 1.6 50.6 0.1 20.0 0.1 0.4  99.0 Unknown steel 1.0 17.8 1.2 67.5 0.6  8.8 6.7 0.5 104.2

[0074] In comparison thereto, in table 3, the underlined concentrations refer to the identified concentrations belonging to the respectively selected standard for each element. It can be noticed that the quantification for this optimized input is in particular more accurate when compared with the target values. This can be in particularly seen for the concentration predictions of the elements Co and Cu in comparison with the predictions in tables 1 and 2. Therefore, the element-wise selection of particular standards based on a similarity score of elements improves the input for the quantification process and thus also the quantification result. As such, also the predictions obtained after performing an iteration algorithm are closer to target values and therefore improve the predications quantitatively. Further, it can be noticed that the sum of concentrations is close to 100% thus reducing the problem of under- and/or overestimation.

TABLE-US-00003 TABLE 3 (example according the present invention) Spectrum Si Cr Mn Fe Co Ni Cu Mo Sum Steel A 0.3  1.2  0.5 96.4 0.01  0.1 0.1 1.4 99.9 Steel B 0.4 16.1 13.8 65.7 0.01  1.5 2.0 0.0 99.6 Steel C 0.5 13.7  3.1 72.1 0.01  9.9 0.3 0.2 99.9 Steel D 1.3 25.0  1.6 50.6 0.03 20.0 0.1 0.4 99.0 Unknown steel 1.0 17.8  1.3 67.5 0.03  8.6 3.1 0.5 99.8 Target values 1.0 18.0  1.2 67.6 0.09  8.6 2.8 0.5 99.7

[0075] The method further allows to define an acceptance criterion and a recursive improvement of the quantification result.

[0076] After the step of performing the quantification S500 of the EDS/WDS spectrum of the unknown sample 150 is determined, in step of S600 the sum of concentrations of the elements of the quantification result is calculated. The quantification is accepted, when the sum of concentrations is within a threshold interval, e.g. set to [98, 102] around 100%. Thus, an acceptance criterion is used to ensure the quality of the quantification result. In the above case, the sum is 99.8 which results in acceptance according to above criterion.

[0077] In case the sum of concentrations is not within said threshold interval, the method may recursively adjust weights of the similarity score and repeat the steps of S300, S400 and S500 until the sum of concentrations of the elements of the quantification result lies within the threshold interval around 100%. Changing weight factors in a controlled way may result in a slightly different selection of standards for example for one element so that the sum of concentrations may be improved. The weight factors will be explained below for a particular similarity score and thus allow recurrence to ensure an accurate quantification result of concentrations. The recurrence can be automatically implemented.

[0078] FIG. 2 illustrates a preferred method for determining a similarity score according to embodiments of the invention. This similarity score designed by the applicant is particularly accurate and has turned out to be particularly successful for accurate selection of the concentrations in the various standards resulting in accurate quantification predictions.

[0079] In step 310, the method comprises to determine S310 an atomic number correction factor Z.sub.u,k, an absorption correction factor A.sub.u,k and a fluorescence correction factor F.sub.u,k for each element k of the preliminarily quantified unknown sample. The determination process may be part of the preliminary quantification performed in step S100.

[0080] In step S320 the same correction factors are determined, i.e. the atomic number correction factor Z.sub.s,k, the absorption correction factor A.sub.s,k and the fluorescence correction factor F.sub.s,k, for each element k of the plurality of elements of the identified standards. In this case, the correction factors may be pre-known data for each of the standard sample and pre-stored together with the standard in a storage.

[0081] In a further step S330, the method comprises, for each element, calculating a square deviation between the determined correction factor of the element of the unknown sample and the corresponding correction factor of the corresponding element of one of the standard samples.

[0082] The method further comprises calculating S340 a sum of square deviations between the respective correction factors Z.sub.u,k, A.sub.u,k and F.sub.u,k of the element of the unknown sample and the respective correction factors Z.sub.s,k, A.sub.s,k and F.sub.s,k of the standard sample. The developed square deviation score used in the present case is designed to not cancel out in contrast to the matrix correction which is a product of the correction factors. In this sense, a superior measure is provided to improve the accuracy. The correction factors obtain as well information on the concentration of the element k and all other concentrations in the sample and is therefore a good measure to perform an accurate element-based selection process.

[0083] To even further improve the selection criterion, the correction factors can also be combined with the concentration of the corresponding element in the standard sample. It has been shown particularly accurate, when a term log(c.sub.s,k) is calculated and the term is subtracted from the calculated (sum of) square deviations. The concentration c.sub.s,k corresponds to the concentration of the element k in the standard sample. The term is based on the insight that small concentrations in the standard to larger concentration are less reliable and that the statistical error of the peak in the spectrum is smaller for larger concentrations. Therefore, an improved similarity score is provided since very small concentrations become unlikely due to the behavior of the log-function for small c.sub.s,k.

[0084] The terms can also be weighted with respect to each other, the determining of the similarity score include setting of weights between the calculated term log(c.sub.s,k) and the calculated (sum of) square deviation. The weights can be represented in the following formula developed by the applicant:

S=w.sub.1.Math.[(Z.sub.u,k−Z.sub.s,k).sup.2+(A.sub.u,k−A.sub.s,k).sup.2+(F.sub.u,k−F.sub.s,k).sup.2]−w.sub.2.Math.log c.sub.s,k,

wherein w.sub.1>0 refers to a preset first weight factor, w.sub.2>0 refers to a preset second weight factor.

[0085] This similarity score includes all above reported said benefits and thus provides an improved similarity score compared to other previously used similarity scores.

[0086] As was already presented above, the weight factors can be used to iteratively perform the quantification when the sum of concentrations of the quantification result has turned out to be within a (narrow) threshold, e.g. [98, 102] around 100% by successive adaption of the weights of the similarity score to improve the selection of concentrations of the standards in this manner. This was already described above with respect to the step of S600 and it is herewith referred to the above description.

[0087] FIG. 3 illustrates a schematic device 10 for determining an element concentration of an EDS/WDS spectrum of an unknown sample. The device comprises at least one processor 20 and a storage 30 that is operatively connected to the processor 20. The storage 30 includes the standard samples and further data, for example the correction factors of the elements of the Z.sub.s,k, A.sub.s,k and F.sub.s,k of the elements in the standard sample. The processor 20 is configured to perform the method as described above with respect to FIGS. 1 and 2.

[0088] FIG. 4 shows a schematic view of a spectroscopic system 100 for generating an EDS/WDS spectrum of a sample 150. The system 100 includes a device 10 for determining an element concentration of an EDS/WDS spectrum of the sample 150 as described above. The device 10 may receive the EDS/WDS spectrum as input. A sample holder may be provided on which the sample 150 is secured. The sample 150 may be a homogeneous material, for which the concentration composition of the elements therein is to be quantified.

[0089] The system 100 may comprise an aperture 110 to locally excite the sample 150 via primary particles 120, e.g. via an electron beam or via radiation, and a detector 140 to record the secondary particles 130, i.e. an X-ray response, from the sample 150. The detector 140 is provided to locally record the outputted response from the particular elements of the sample 150. The aperture 110 is in top position with respect to the sample 150 as indicated but for example may also be inclined and/or positioned sideways with respect to the sample 150. The detector 140 is inclined and/or positioned sideways with respect to the sample 150.

[0090] The device 10 for determining an element concentration 10 may receive and/or generate the spatially resolving EDS/WDS spectrum based on the recorded data of the sample 150 by the detector 140. The processor can quantify the EDS/WDS spectrum with improved accuracy. The system 100 may share all the advantages as described above for the corresponding method.

REFERENCE SIGNS

[0091] 10 device [0092] 20 processor [0093] 30 storage [0094] 100 spectroscopic system [0095] 110 aperture [0096] 120 primary particles [0097] 130 secondary particles [0098] 140 detector [0099] 150 sample [0100] S100 preliminary quantification [0101] S200 identify standard sample [0102] S300 determine similarity score [0103] S400 select standard for each element [0104] S500 quantification of the sample [0105] S600 calculate sum of concentrations [0106] S310 determine correction factors of sample of unknown composition [0107] S320 identify correction factors of sample of unknown composition [0108] S330 calculated square deviation [0109] S340 calculate sum of square deviations [0110] S350 calculate concentration term [0111] S360 set weights for summation