EP1144986A2

EP1144986A2 - Energy dispersion x-ray fluorescence analysis of chemical substances

Info

Publication number: EP1144986A2
Application number: EP00901071A
Authority: EP
Inventors: Alexander Henrich; Hans-Helmut Itzel; Peter Hoffmann; Hugo Ortner
Original assignee: Merck Patent GmbH
Current assignee: Merck Patent GmbH
Priority date: 1999-01-23
Filing date: 2000-01-07
Publication date: 2001-10-17
Also published as: WO2000043761A2; JP2002535647A; US6496562B1; WO2000043761A3

Abstract

The invention relates to a method for classifying and identifying by means of energy dispersion X-ray fluorescence analysis chemical substances whose X-ray fluorescence lines cannot be detected and which therefore cannot be classified by energy dispersion X-ray fluorescence analysis alone. Said method is characterized in that the sample to be analyzed is analyzed in its original packaging or natural state without prior processing in a sample vessel. According to the method the sample is: a) positioned in front of the measuring aperture in a sample chamber of an X-ray fluorescence apparatus; b) measured; and c) classified and identified by application of multivariate, statistical techniques to the measurement signals obtained, i.e. to the Compton and Rayleigh scattering.

Description

Energy dispersive X-ray fluorescence analysis of chemical substances

The present invention relates to the differentiation and classification by means of X-ray fluorescence analysis of chemical substances whose X-ray fluorescence lines cannot be detected and which therefore cannot be classified by energy-dispersive X-ray fluorescence analysis (EDRFA) alone, through the packaging and without having to carry out a sampling.

Rapid identity verification of laboratory chemicals is necessary in many places. This mainly applies in chemical companies within the so-called material flow management. In this case, material flow means taken back chemicals that are returned to the chemical factory by the end users or intermediaries.

According to the Closed Substance Cycle and Waste Management Act, each material flow is to be regarded as waste until a control plausibly characterizes each material flow. Only then can the material flow be referred to as a product or raw material, secondary raw material or finally as waste.

After the chemicals have been taken back, they are documented accordingly, checked for their composition and then reused as secondary raw materials in production, if necessary.

Energy-dispersive X-ray fluorescence analysis (EDRFA) is a fast analysis method for the qualitative and quantitative determination of elements in substances. The determination is made by evaluating the X-ray fluorescence lines. The X-ray fluorescence lines of the elements with atomic numbers between 21 and 92 can be detected and assigned by a PE (polyethylene) container. However, a large part of the substances consists of elements with atomic numbers between 1 and 20. Characterization of these elements and thus these substances is not possible with the conventional EDRFA evaluation (X-ray fluorescence line determination and evaluation) due to the lack of X-ray fluorescence lines. A statement about the substance and its composition can be made via the coherent (Rayleigh scatter) and incoherent (Compton scatter) scatter of X-rays in the substance. Correlations between the mean atomic number and the ratio between coherent and incoherent scattered radiation are known and are described, for example, by H. Kunzendorf in Nuclear Instruments and methods, 99 (1972) 61 1-612. Various EDRFA providers use the matrix correction of X-ray fluorescence lines based on the inelastic scattered radiation for the quantitative evaluation.

The object of the present invention is to safely characterize and distinguish substances from one another whose x-ray fluorescence lines cannot be detected and which therefore cannot be classified by energy-dispersive x-ray fluorescence analysis (EDRFA) alone, without additional, other analysis methods and without having to take a sample to distinguish.

It has now surprisingly been found that it is still possible to use the energy-dispersive X-ray fluorescence analysis for the classification and identification of chemical substances with an atomic number from 1 to 20, by using multivariate, statistical methods on the measurement signals obtained from the entire Compton and Rayleigh scattering range.

Until now, these substances, whose X-ray fluorescence lines cannot be detected, but only have a Compton and Rayleigh scattering range, could not be distinguished from one another, but instead were classified together in an assignment field. If you wanted to know more precisely which individual elements or substances were present, further conventional analyzes had to be carried out.

The invention therefore relates to a method for classification and identification by means of energy-dispersive X-ray fluorescence analysis of chemical substances whose X-ray fluorescence lines cannot be detected and which cannot therefore be classified solely by energy-dispersive X-ray fluorescence analysis (EDRFA), which is characterized in that the sample to be analyzed in its original packaging or as such without prior preparation in a sample container

a) is positioned in an X-ray fluorescence system in front of the measurement opening in a sample chamber, then measured and

b) by applying multivariate statistical methods to the measurement signals obtained, i.e. of the Compton and Rayleigh scattering range, is classified and identified.

The principal component analysis (PCA) for recognizing the differences between the substances and / or the Regularized Discriminance Analysis (RDA) for the differentiation and classification of the substances are used as multivariate, statistical methods.

As already mentioned, the identity check of laboratory chemicals is particularly important when taking back chemicals. As can be seen from the statistics, it is mainly small packs that are returned in large quantities to the chemical companies. For this reason, substances in small packs are often documented and analyzed on a small pack sorting system (KSA). In the case of the analysis of different substance flows, each individual substance in its packaging must be analyzed. Opening the packaging and taking a sample may be due to the risk to people and the environment that can be excluded when dealing with old chemicals (Recycling Management and Waste Prevention Act (Krw- / AbfG) 1994 (BGBI. I, 1354); Regulation No. 259/93 des Council for the supervision and control of shipments of waste in, into and out of the European Community 1993 (OJ L 30, 1); Regulation introducing the European Waste Catalog (EWC Regulation) 1996 (BGBI. I 1428); Regulation on Determination of waste requiring special monitoring (BestbüAbfV) 1996 (BGBI. I 1366); Ordinance for the determination of waste requiring monitoring for recycling

(BestbüAbfV) 1996 (BGBI. I 1377); Second general administrative regulation on the Waste Act (TA Abfall) 1991 (GMBI, p. 139, ber. 496); Sales Ordinance on Protection against Hazardous Substances (GefStoffV) 1993 (BGBI. I 1782) does not take place in the room in which the sorting system is located and the analyzes are carried out. Energy dispersive X-ray fluorescence analysis (EDRFA) has emerged as a suitable method for the analysis through the unopened packaging.

The analysis is therefore preferably carried out through the packaging, different packaging materials (glass or polyethylene packaging) being able to be present and having to be taken into account accordingly in the assignment.

When checking these substances, the aim is not to completely identify the substances, including the main and secondary constituents. However, a plausible assignment of the substance spectrum recorded by the packaging to the spectrum of the substance name noted on the packaging label is expected. This type of analytics is called mapping analytics.

Substances containing elements with an atomic number (OZ)> 22 (Ti) can be characterized depending on the packaging size by the PE packaging based on their element lines. The detection of the peaks, determination of the peak parameters (peak position, half-width, area) takes place automatically, as does the subsequent comparison of the XRF data with the information from the database. It is new, too, that these substance groups can also be distinguished much better using multivariate, statistical methods. For this purpose, the X-ray fluorescence range of the element is calculated with an OZ> 22 and the Compton and Rayleigh scattering range with the multivariate statistical methods.

However, as already mentioned, a large part of the substances consists of elements with atomic numbers between 1 and 20, which have no detectable X-ray fluorescence lines. Characterization of these elements and thus the substance (such as between NaCI and NaCN or K ₂ CO ₃ and KF) is not possible with a conventional EDRFA evaluation due to the lack of X-ray fluorescence lines. The XRF measurement of such substances only provide scattered radiation spectra that cannot be assigned to the conventional EDRFA Allow evaluation. So far, these substances have been classified in a common "assignment field".

However, this means that only a very rough classification can take place here. For a more precise classification, additional analysis methods must be used in these cases, sometimes with prior sampling and preparation of the sample. However, these further studies are time consuming and costly. These additional analyzes can now be avoided by the present invention.

According to the invention, a further distinction can now be made for these substances within the assignment field, which is defined with the conventional X-ray fluorescence (RFA) evaluation, by using multivariate, statistical methods on the Compton and Rayleigh scattering range.

The main component analysis (PCA) and the regulating discriminant analysis (RDA) are preferred as multivariate, statistical methods. These methods are known to the person skilled in the art as such and are dealt with in detail in many references (cited as an example for the RDA: JH Friedman, J. Amer. Statistical Association, 1989. Vol. 84, No. 405, 165-175 ).

A further assignment can be made by directly applying multivariate statistical methods to the Compton and Rayleigh scattering range. According to the invention, the various methods can be applied to the scatter spectra alone or both in succession.

With the help of the main component analysis, further subclasses can be identified in the classes defined by the conventional XRF evaluation. With PCA, spectral differences between the individual substances can be made visible in the PCA display. In example 1 (Figure 1) the result from the PCA evaluation is shown in the form of a "score plot". Here the scattered radiation area of 20 substances (in PE sample vessel) that do not provide X-ray fluorescence signals was calculated with the main component analysis. Discriminant analysis methods are used to set up mathematical models for the substance classes - with these models, substances can later be assigned to a class.

Classes of recorded spectra of different substances are visualized with the PCA, then their classes are calculated with the RDA. This means that both the spectral range or the main components calculated for the spectral range can be used as variables in the RDA. The number of main components used is determined according to the so-called "Eigenvalue 1 criterion" or by cross-validation.

The individual steps from spectra acquisition to classification are described below.

First, the sample to be analyzed is in its original packaging - opening the package is therefore not necessary and sampling is not necessary - or as such is positioned in a sample chamber in the X-ray fluorescence system in front of the measurement opening in a sample chamber without prior preparation.

The packaging or the sample vessel in which the sample to be analyzed is located can consist of a material selected from the group consisting of polyethylene, glass, aluminum, paper and cardboard.

Now the EDRFA spectrum is recorded. Then, if there are no X-ray fluorescence lines of the elements from the substance, the Compton and Rayleigh scattering range from 19.6 to 26.3 keV (note: this range only applies to excitation with an Ag tube, see Table 1) ; when using other excitation sources, this scattering range lies in an area corresponding to the excitation source) for the multivariate, statistical calculations (PCA.RDA). Subsequently - if desired - the main components are calculated with the PCA for the new substance in the new model (incl. Substance), which contains the spectrum of the substance. This step is optional. In the next step, the classes are set up and defined in the RDA with a learning data set (spectral ranges or optionally the main components determined from the previous step).

The new substance is then classified / classified

(Test data set, i.e. spectral range or main components) into a class from the learning data set (optionally the main components from the PCA can also be used for the classification instead of the spectrum). The target class must of course be included in the learning data record. The classification is based on the calculations described in the literature. The following references are cited as examples:

Friedman, J.H., "Regularized Disciminant Analysis" in J. Am. Stat. As-soc. (1989) 84, 165-175; Frank, I.E., Friedman, J.H., Classification: "oldtimers and newcomers" in J. Chemom. (1989) 33, 463-75; Wu, W., Mallet, Y., Walczak, B., Penninckx, W., Massart, DL, Heuerding, S., Erni, F., "Comparison of regularized discriminant analysis, linear discriminant analysis and quadratic discriminant analysis "applied to NIR data" in Anal. Chim. Acta (1996) 3293, 257-265; Baldovin, A., Wen, W., Massart, D.L., Turello, A. "Regularized discriminant analysis RDA - Modeling for the binary discrimination between pollution types" in Chemom. Intell. Lab. Syst. (1997) 381, 25-37.

In the last step, the comparison is made between the class assigned to the spectrum and the actual class or the substance name described on the label. If the result is the same, the substance is processed further in the form of storage and use in production.

This entire evaluation is preferably carried out automatically by using suitably adapted software, which considerably speeds up the entire analysis time and calculation time. The PCA and RDA algorithm is commercially available and can be implemented in the later evaluation data processing.

The substances to be analyzed in their packaging usually reach the EDRFA system on a conveyor belt. The position The packaging in front of the measuring opening, the recording of the EDRFA spectrum, the evaluation of the spectra, the subsequent spectrum assignment and the repositioning of the packaging on the conveyor belt are carried out fully automatically. Accordingly, the EDRFA system and the associated components (sample chamber, interfaces to the substance database, control of the EDRFA) must be designed so that automatic control of the individual components is possible.

The invention therefore also relates to the fact that the method is used within an automated system for sorting and assigning old or new packaging containing chemical substances.

The automated system preferably consists of the following components or steps:

- Conveyor belt for the substances to be analyzed in their packaging;

- EDRFA plant;

- Positioning of the packaging in front of the measuring opening in a sample chamber, the sample chamber completely enclosing the packaging;

- Measurement;

- Spectra evaluation and assignment

- Further evaluation using multivariate, statistical methods and renewed, more precise assignment;

- Reposition the packaging on the conveyor belt.

An X-ray fluorescence analysis apparatus consisting of an X-ray tube, a generator, an energy-resolving detector and evaluation electronics is preferably used.

In a preferred embodiment, the following configuration of the EDRFA system is selected: X-ray tube with generator and semiconductor detector, the measurement geometry, that is to say the angle between the excitation source, sample and detector, is selected to be variable between 45 ° and 90 °, so that the Compton and Rayleigh scatter lines are resolved in the detector.

The sample chamber must completely enclose the package, since protective mechanisms must be observed when handling ionizing radiation in accordance with the X-ray regulation. It must be ensured that the X-rays emitted do not exceed a defined limit. The sample chamber is preferably made of a material that does not increase the spectra background (scatter) in the sample chamber, can be opened and closed automatically and can be adapted to the EDRFA apparatus.

The parameters for routine operation, i.e. X-ray tube voltage and current, primary beam filter material and thickness, detector aperture, position coordinates for the pack in front of the EDRFA measurement opening, can be determined and set experimentally depending on requirements and requirements. For a later assignment of the spectra to the substances, the packaging sizes, materials and packaging positions must also be taken into account.

Since a large number of returned packs can be expected, the analysis duration should also be short. According to the invention, the measurement time for recording the spectra is preferably <30 seconds.

Example A, Table 1 describes a preferred configuration with the parameters for measurement and evaluation. This is only intended to be an exemplary list that is in no way limiting. Example A also lists the preferred general measurement conditions for the tests.

In Scheme 1, the EDRFA measurement geometry and coordinate system for packaging positioning is preferred for an inventive one

Embodiment to see.

The method according to the invention provides a fast, reliable and effective analysis method for identifying chemical substances through the packaging. In principle, the EDRFA is based on such substances from elements with numbers between 1 and 20, which were previously not distinguishable in this way, expanded. As a result, a significantly improved characterization and assignment can be achieved when taking back chemicals, without having to use further analysis methods with complex sample preparation and sampling.

Even without further explanations, it is assumed that a person skilled in the art can use the above description to the greatest extent. The preferred embodiments are therefore only to be understood as a descriptive disclosure, and in no way as a limitation in any way.

The full disclosure of all applications and publications listed above and below are incorporated by reference into this application.

The following examples are intended to illustrate the invention in more detail.

Example A

Table 1 describes the configuration with the parameters for measurement and evaluation, which was used in the following examples.

Table 1 Equipment of the EDRFA plant for the waste-specific allocation in the return of substances

Model description

Preamplifier with FET input stage, pulsed, optoelectronic feedback,

Immersion cryostat with 25 μm inlet window made of Be, 30 I liquid gas container

RS-3001 X-ray generator, tube hood with a window, 5 m HV cable, adapter and tube hood holder

RACK-300 analog electronics in 19 "rack with analog power supply, active filter amplifier with triangular pulse shaping, baseline restorer,

Pulse pileup rejector and pulse striker, detector high voltage supply, continuously adjustable from 0 - 1000 V.

TSI-MCA analog / digital converter and pulse height analyzer for ISA bus

TSI-AQL program package for EDRFA under MS Windows TM

ICP-300A industrial PC in 19 "case with 250 W power supply, additional fan, passive bus board with 8 ISA, 2 ISA / PCI and 4 PCI slots, single board computer with Pentium / 133 CPU, 256 kB cache, 32 MB RAM , 2 x ser., 1 x par. Interface E-IDE interface, 1, 44 MB floppy disk drive, 1, 2 GB hard disk, CD-ROM drive, PCI graphics card Matrox-Millenium (2 MB), MF keyboard, MS Mouse, MS DOS 6.22 and MS Windows 3.11 ™

EIZO-57S SVGA color video monitor with 43.2 cm (17 ") picture tube, TOC 95 model Eizo Flexscan T57S

Equipment cabinet in 19 "standard to accommodate the X-ray generator, the system electronics and the industrial PC

Primary beam collimator (3 mm beam diameter, material: AI)

Primary jet filter (materials Cu, Mo, Ag) Model description

WECO KL chiller (GWK) 04

SCAN for software package for chemometric classification of Windows ^® , substances, from Minitab version 1.1

The measurement conditions for the substance classification tests can be selected as follows:

Tube high voltage 45 kV

Tube current 4 mA

Primary beam collimator diameter 2mm

Primary jet filter material Ag

Primary jet filter thickness 0, 12 mm

Detector aperture material AI

Detector aperture 3 mm

Detector time constant 2 μs

X-ray height - packaging base 1.5 cm

Measuring time for the classification tests 20 s

Packaging position in front of the EDRFA measurement opening 10/0 (in mm, to the right of the detector center)

Angle (primary beam - sample - detector) approx. 60 °

example 1

The measurement parameters and conditions can be found in example A.

1 .1

The Compton and Rayleigh scattered radiation range (19.7 - 26.2 keV) of 20 substances that do not provide an X-ray fluorescence signal are calculated with the main component analysis (PCA). Figure 1 shows the result from the PCA evaluation in the form of a "score plot". 1.2

The results from the RDA (Regularized Discriminance Analysis) calculation for the 20 substances measured in 1.1 are shown in Figure 2 (calculated with the Compton and Rayleigh scattering range) and in Figure 3 (calculated with the first three main components from the PCA) .

1.3

Table 2 lists the different substances from this series of measurements with their physical data.

Table 2 Different substances and their physical data from the no element group (measurement series 1)

Example 2

The measurement parameters and conditions can be found in example A.

2.1

In this example 5 chromium compounds were examined. The element line as well as the Compton and Rayleigh scattering range were evaluated with the main component analysis (PCA). Figure 4 shows the result of the PCA evaluation in the form of a "score plot". 2.2

The results from the RDA (Regularized Discriminance Analysis) calculation for this series of tests (see point 1.2) are shown in Figures 5 and 6.

2.3

Table 3 lists the substances from the chromium group and their physical data.

Table 3: Investigated substances from the chromium group and their physical data (measurement series 2)

Example 3

3.1

Measurement series 3 was carried out using substances from the iron group in accordance with the measurement parameters and conditions described in Example A.

In this example 7 iron compounds were examined. Both the element line as well as the Compton and Rayleigh Scattering range evaluated with the main component analysis (PCA). Figure 7 shows the result from the PCA evaluation in the form of a "score plot".

3.2

The results from the RDA (Regularized Discriminance Analysis) calculation for this series of tests (see point 1.2) are shown in Figures 8 and 9.

3.3

Table 4 shows the physical data of this series of measurements.

Table 4: Investigated substances from the iron group and their physical data (measurement series 3)

The examples with the PCA calculations show that there are 20 substances from the no element group and selected substances in the new data space, which is spanned by the main components Have chromium and iron separated from the one-element groups. This separation is not achieved with the conventional EDRFA evaluation of the spectra recorded by the packaging.

The best possible separation of the substances in the new data space and the lowest scatter within the groups is achieved for the no element group by the PCA calculation taking into account the Compton and Rayleigh scattering range. For the one-element groups, good separation of the groups from one another and little scatter within the groups is achieved by means of the PCA calculation using a combination of the fluorescence line region of the element and the Compton and Rayleigh scatter region.

The examples with the RDA calculations show that EDRFA spectra recorded by the packaging can be distinguished from one another with the aid of the respective RDA model if the spectra have spectral similarities which are recognizable for the RDA. For the one-element groups chromium and iron, the substances can be assigned to a previously defined class with the spectral range (fluorescence line, Compton and Rayleigh scattering range) as well as with the significant HK from the PCA as variables. The classification with the spectral range is even better for the Eisen group (calculated by internal cross-validation of the Eisen data set).

The RDA classification of substances in the no element group for their classes also works.

Scheme 1:

Claims

PATENT CLAIMS

1.Procedure for classification and identification by energy-dispersive X-ray fluorescence analysis of chemical substances whose X-ray fluorescence lines cannot be detected and which therefore cannot be classified by energy-dispersive X-ray fluorescence analysis (EDRFA) alone, characterized in that the sample to be analyzed is in its original packaging or as such without previous preparation in a sample vessel a) is positioned in an X-ray fluorescence system in front of the measurement opening in a sample chamber, b) is then measured and c) using multivariate, statistical methods on the measurement signals obtained, ie the X-ray spectral ranges (Compton and Rayleigh

Scattering range) is classified and identified.

2. The method according to claim 1, characterized in that when positioning the sample to be analyzed, the angle between the excitation source, sample and detector, also called measurement geometry, is selected to be variable between 45 ° and 90 °, so that the Compton and Rayleigh scatter lines are resolved in the detector.

3. The method according to claim 1 or 2, characterized in that the sample to be analyzed is measured and classified in the closed packaging, ie without sampling.

Method according to one of claims 1 to 3, characterized in that the classification is carried out by the packaging or a sample vessel consisting of a material selected from the group consisting of polyethylene, glass, aluminum, paper and cardboard.

5. The method according to any one of claims 1 to 4, characterized in that in step c) by using the methods of main component analysis (PCA) and / or regularized discrimination analysis (RDA) on the X-ray spectral ranges (Compton and Rayleigh scattering range) the measured signals are classified / identified for the sample to be analyzed.

6. The method according to claim 5, characterized in that spectral differences of the individual substances are made visible in the PCA representation by the application of the main component analysis.

7. The method according to claim 5 or 6, characterized in that a classification / identification of the sample to be analyzed is carried out by using the Regularized Discriminance Analysis (RDA) makes substance identification possible by direct calculation of a spectrum or spectral range, in which case the The test substance is assigned to a previously determined and defined class.

8. The method according to any one of claims 5 to 7, characterized in that for a classification / identification, the RDA method is applied to the main components obtained by the PCA.

9. The method according to any one of claims 1 to 8, characterized in that an X-ray fluorescence analysis apparatus consisting of an X-ray tube, a generator, an energy-resolving detector and evaluation electronics is used.

10. The method according to any one of claims 1 to 9, characterized in that the measuring time for recording the spectrum <30 seconds is.

1 1. The method according to any one of claims 1 to 10, characterized in that it is used within an automated system for sorting and assigning old or new packaging containing chemical substances.

12. The method according to claim 11, characterized in that the automated system consists of the following components or steps:

Conveyor belt for the substances to be analyzed in their packaging;

EDRFA plant;

Position the packaging in front of the measuring opening in a sample chamber, the sample chamber being the

Packaging completely encloses;

Measurement;

Spectra evaluation and assignment; further evaluation using multivariate, statistical methods and renewed, more precise assignment;

Reposition the packaging on the

Conveyor belt.