rus
Issue #2/2018
E.V.Gorsky, A.M.Livshits
Analysis of metal alloys on low-resolution emission spectrometers. Correction of inter-element effects
Analysis of metal alloys on low-resolution emission spectrometers. Correction of inter-element effects
A technique for correction of the influence of third components in analysing aluminium alloys and high-alloy steels with the use of low-resolution emission spectrometers with spark excitation is considered. The results of the calculation of additive and multiplicative coefficients of influence for analysed elements are presented. It is shown that the use of regression methods for processing of the analysis results makes it possible to significantly reduce the errors in determining the contents of elements in the considered groups of alloys.
DOI: 10.22184/2227-572X.2018.39.2.110.113
DOI: 10.22184/2227-572X.2018.39.2.110.113
Теги: aluminium alloys analytical unit emission spectrometer алюминиевый сплав высоколегированная сталь эмиссионный спектрометр
In up-to-date factory laboratories, atomic emission spectral analysis is widely used to control the composition of metal alloys [1, 2], the equipment for which is mainly developed in two directions. The first one is the development of spectrometers with high resolution and high sensitivity to the detected radiation, which are intended for an accurate analysis of the composition of alloys including small admixtures of chemical elements. This is mainly achieved by utilization of stable and intensive sources of excitation. Spectral units with a relatively high dispersion (less than 1 nm/mm) are used, which enables to resolve close-lying spectral lines in order to reduce the influence of various chemical elements on the measurement results. At the same time, the amount of scattered light is reduced but the dimensions and cost of the devices increase.
The second direction for the emission spectrometer market development is associated with the creation of small and inexpensive devices with a relatively low dispersion and simpler excitation sources. Such equipment is highly demanded by customers although has a number of shortcomings. The following factors have an impact on the measurement results: inter-elements influences due to the overlapping of spectral emission lines on each other, the uneven supply of constituent components to the discharge region, the reabsorption of analytical spectral lines etc. [3].
Usually the main reason for errors in concentration determination are so called third component influences. To compensate for these influences in emission spectroscopy, mathematical regression methods [4] implemented in the software of devices have been widely used. For example, a series of PAPUAS-4 spectrometers (dispersion 1.5 nm/mm) is equipped with software that uses a mathematical methods for correcting the inter-elements effect when processing the results of analysis of aluminium [5], copper [6] and nickel alloys [7] as well as high-alloy steels [8]. The operation of spectrometers in factory environment conditions has shown the effectiveness of such methods.
In the meantime, low-dispersion spectrometers may face limitations due to a stronger overlap of the spectral lines. The purpose of the research was to consider the possibility of applying regression methods for correction of inter-element effect in a spectrometer with a relatively low dispersion (4–5 nm/mm).
To accomplish the task, a spectrometer mock-up based on a concave diffraction grating with spectrum registration by means of linear CCD array [9] was developed and assembled. To excite the spectra, a high-voltage spark discharge was used in the air atmosphere (Table 1).
The research was made for alloys based on aluminium and high-alloyed steel. For the aluminium alloys, the samples of four sets of standard samples were used: 18 (AMG grade), 23 (AK12 grade), 207 (AK5M2 grade) and VSA1 (primary aluminium). For steel alloys, the samples of two sets of high-alloy steels were used: 31 (grade X12H20T3p) and 36 (grade X12m). The emission spectra of the alloys were studied, and for each chemical element analytical lines, which are maximally free from overlapping peaks of other analytes, were chosen. Tables 2 and 3 show the wavelengths used in the methods for determining the composition of aluminium alloys and high-alloy steels: λ1 and λ2 are the wavelengths of the analytical line and the corresponding comparison line, and Е1 and Е2 are their excitation potentials in electron-volts, respectively.
The algorithm of data processing of the correction of inter-element effects consisted of a number of steps [5]. Let the relative intensities of the analytical lines be obtained for a certain set of standard samples. To calculate the concentration of an element, the formula below was used:
,
where Cр is the concentration of the element p being investigated, Rj is the relative intensity of the line; aj are the so-called basic coefficients that are obtained by the least square method in calculating the concentration curve (the dependence of the rated concentration on the intensity) without taking into account the inter-element effects. In the software used to determine this basic curve, polynomials of a degree no higher than the third one are used (j = 1, 2, 3).
The following coefficients ka are responsible for the so-called additive effects depending on the concentration of the influencing element only. Usually, these effects are caused by the imposition of the spectral line of the influencing element on the analytical line, or on the site where its background level is determined. Here, Ca is the concentration of the corresponding influencing element.
The last term in the formula describes the so-called multiplicative effects with the km coefficients which depend on the product of the concentrations of the influencing element and the analysed one. When there is a mutual influence of elements, for the correct operation of the calculation algorithm, instead of the concentration of the analysed element, the intensity of its analytical line is considered. Coefficients in the software are calculated by the iterative procedure.
It should be noted that special care should be taken in the calculation of the coefficients of influence without using any ‘excess’ ones because it is possible to obtain a regression model in which random deviations of measurements are taken into account as systematic influences. In addition, with sufficient initial data, the system becomes overdetermined so the concentration values are accurately calculated in a certain small range, and if outside, there is the risk of obtaining unpredictable results.
As a result of the performed experiments, chemical elements which make a significant influence on the results of the analysis were identified. Tables 4 and 5 show the additive and multiplicative coefficients of influence calculated with a regression methods for the analysed elements in aluminium alloys and high-alloy steels.
The presented results show that in the used spectrometer with low dispersion, the amount of additively influencing elements has significantly increased in comparison with devices with a higher dispersion of 1.5 nm/mm [5].
In Fig.1 as an example, a comparison of the certified and determined concentrations is given without correction (a) and with correction (b) of inter-element effects for the magnesium line Mg 293.654 nm/Al 308.215 nm in aluminium alloys, and Fig.2 presents similar dependences for the manganese line Mn 293.306 nm/Fe 239.562 nm in high-alloy steels.
Practical verification of the developed methods has shown that they can be successfully used for alloys that are close to those investigated by composition. This limits the scope of regression methods for spectrometers with comparatively low dispersion. However, for quality control in factories, in which a limited number of alloys are usually produced, considered spectrometer which uses regression methods for correction, can become in common use.
REFERENCES
1. Buravlev Yu.M. Atomic emission spectroscopy of metals and alloys. DonNU. 2000. 437 p.
2. Thomsen V.B.E. Modern spectrochemical analysis of metals. Ohio: ASM International, 1996. 204 p.
3. Buravlev Yu.M. The effects of sample composition and size on the results of spectral analysis of alloys. Kiev. Tekhnika. 1970. 212 p.
4. Orlov A.G. Calculation methods in quantitative spectral analysis. Leningrad. Nedra. 1977. 108 p.
5. Gorsky E.V., Livshits A.M., Peleznev A.V. Accounting for the Influence of the ‘Third’ Elements in the analysis of aluminium alloys on the PAPUAS-4 spectrometer // Factory Laboratory. 2006. No 3. P. 11–15.
6. Gorsky E.V., Livshits A.M., Peleznev A.V. Features of the use of the emission spectrometer PAPUAS-4IM in the analysis of copper alloys // Factory Lab. 2007. No 5. P. 25–27.
7. Livshits A.M., Gorsky E.V., Mavrin N.B., Palkin Yu.A. Features of analysis of nickel alloys with the emission spectrometer PAPUAS-4 // Control. Diagnostics. 2011. No 10. P. 55–58.
8. Gorsky E.V., Livshits A.M. Accounting for inter-elemental influences in the analysis of high-alloy steels on the emission spectrometer PAPUAS-4 // Factory Laboratory. Material Diagnostics. 2017. 83 (2). P. 26–30.
9. Holst G.C., Lomheim T.S. CMOS/CCD sensors and camera systems // JC. Publishing / 2011. 335p.
The second direction for the emission spectrometer market development is associated with the creation of small and inexpensive devices with a relatively low dispersion and simpler excitation sources. Such equipment is highly demanded by customers although has a number of shortcomings. The following factors have an impact on the measurement results: inter-elements influences due to the overlapping of spectral emission lines on each other, the uneven supply of constituent components to the discharge region, the reabsorption of analytical spectral lines etc. [3].
Usually the main reason for errors in concentration determination are so called third component influences. To compensate for these influences in emission spectroscopy, mathematical regression methods [4] implemented in the software of devices have been widely used. For example, a series of PAPUAS-4 spectrometers (dispersion 1.5 nm/mm) is equipped with software that uses a mathematical methods for correcting the inter-elements effect when processing the results of analysis of aluminium [5], copper [6] and nickel alloys [7] as well as high-alloy steels [8]. The operation of spectrometers in factory environment conditions has shown the effectiveness of such methods.
In the meantime, low-dispersion spectrometers may face limitations due to a stronger overlap of the spectral lines. The purpose of the research was to consider the possibility of applying regression methods for correction of inter-element effect in a spectrometer with a relatively low dispersion (4–5 nm/mm).
To accomplish the task, a spectrometer mock-up based on a concave diffraction grating with spectrum registration by means of linear CCD array [9] was developed and assembled. To excite the spectra, a high-voltage spark discharge was used in the air atmosphere (Table 1).
The research was made for alloys based on aluminium and high-alloyed steel. For the aluminium alloys, the samples of four sets of standard samples were used: 18 (AMG grade), 23 (AK12 grade), 207 (AK5M2 grade) and VSA1 (primary aluminium). For steel alloys, the samples of two sets of high-alloy steels were used: 31 (grade X12H20T3p) and 36 (grade X12m). The emission spectra of the alloys were studied, and for each chemical element analytical lines, which are maximally free from overlapping peaks of other analytes, were chosen. Tables 2 and 3 show the wavelengths used in the methods for determining the composition of aluminium alloys and high-alloy steels: λ1 and λ2 are the wavelengths of the analytical line and the corresponding comparison line, and Е1 and Е2 are their excitation potentials in electron-volts, respectively.
The algorithm of data processing of the correction of inter-element effects consisted of a number of steps [5]. Let the relative intensities of the analytical lines be obtained for a certain set of standard samples. To calculate the concentration of an element, the formula below was used:
,
where Cр is the concentration of the element p being investigated, Rj is the relative intensity of the line; aj are the so-called basic coefficients that are obtained by the least square method in calculating the concentration curve (the dependence of the rated concentration on the intensity) without taking into account the inter-element effects. In the software used to determine this basic curve, polynomials of a degree no higher than the third one are used (j = 1, 2, 3).
The following coefficients ka are responsible for the so-called additive effects depending on the concentration of the influencing element only. Usually, these effects are caused by the imposition of the spectral line of the influencing element on the analytical line, or on the site where its background level is determined. Here, Ca is the concentration of the corresponding influencing element.
The last term in the formula describes the so-called multiplicative effects with the km coefficients which depend on the product of the concentrations of the influencing element and the analysed one. When there is a mutual influence of elements, for the correct operation of the calculation algorithm, instead of the concentration of the analysed element, the intensity of its analytical line is considered. Coefficients in the software are calculated by the iterative procedure.
It should be noted that special care should be taken in the calculation of the coefficients of influence without using any ‘excess’ ones because it is possible to obtain a regression model in which random deviations of measurements are taken into account as systematic influences. In addition, with sufficient initial data, the system becomes overdetermined so the concentration values are accurately calculated in a certain small range, and if outside, there is the risk of obtaining unpredictable results.
As a result of the performed experiments, chemical elements which make a significant influence on the results of the analysis were identified. Tables 4 and 5 show the additive and multiplicative coefficients of influence calculated with a regression methods for the analysed elements in aluminium alloys and high-alloy steels.
The presented results show that in the used spectrometer with low dispersion, the amount of additively influencing elements has significantly increased in comparison with devices with a higher dispersion of 1.5 nm/mm [5].
In Fig.1 as an example, a comparison of the certified and determined concentrations is given without correction (a) and with correction (b) of inter-element effects for the magnesium line Mg 293.654 nm/Al 308.215 nm in aluminium alloys, and Fig.2 presents similar dependences for the manganese line Mn 293.306 nm/Fe 239.562 nm in high-alloy steels.
Practical verification of the developed methods has shown that they can be successfully used for alloys that are close to those investigated by composition. This limits the scope of regression methods for spectrometers with comparatively low dispersion. However, for quality control in factories, in which a limited number of alloys are usually produced, considered spectrometer which uses regression methods for correction, can become in common use.
REFERENCES
1. Buravlev Yu.M. Atomic emission spectroscopy of metals and alloys. DonNU. 2000. 437 p.
2. Thomsen V.B.E. Modern spectrochemical analysis of metals. Ohio: ASM International, 1996. 204 p.
3. Buravlev Yu.M. The effects of sample composition and size on the results of spectral analysis of alloys. Kiev. Tekhnika. 1970. 212 p.
4. Orlov A.G. Calculation methods in quantitative spectral analysis. Leningrad. Nedra. 1977. 108 p.
5. Gorsky E.V., Livshits A.M., Peleznev A.V. Accounting for the Influence of the ‘Third’ Elements in the analysis of aluminium alloys on the PAPUAS-4 spectrometer // Factory Laboratory. 2006. No 3. P. 11–15.
6. Gorsky E.V., Livshits A.M., Peleznev A.V. Features of the use of the emission spectrometer PAPUAS-4IM in the analysis of copper alloys // Factory Lab. 2007. No 5. P. 25–27.
7. Livshits A.M., Gorsky E.V., Mavrin N.B., Palkin Yu.A. Features of analysis of nickel alloys with the emission spectrometer PAPUAS-4 // Control. Diagnostics. 2011. No 10. P. 55–58.
8. Gorsky E.V., Livshits A.M. Accounting for inter-elemental influences in the analysis of high-alloy steels on the emission spectrometer PAPUAS-4 // Factory Laboratory. Material Diagnostics. 2017. 83 (2). P. 26–30.
9. Holst G.C., Lomheim T.S. CMOS/CCD sensors and camera systems // JC. Publishing / 2011. 335p.
Readers feedback
