JPH045549A - Emission spectrochemical analysis of steel - Google Patents

Abstract

PURPOSE:To rapidly and exactly analyze the grain size distribution of the inclusions in steel by discharging pulses to a sample, determine the correlation between the spectral rays of an iron element and the element to be analyzed, ranking abnormal value pulses, and counting the number of the abnormal value pulses belonging to the respective ranks. CONSTITUTION:The emission intensity of the element to be analyzed increases extraordinarily with respect to the emission intensity of the iron elements of the spectral rays when the pulses are discharged to the sample. The detected pulse groups are processed by a regression analysis and the correlation equation of the lower tangent of the pulse groups is deformed to obtain an upper tangent. The pulses exceeding this upper tangent are decided as the abnormal value pulses. The abnormality degree is determined with every abnormal value pulse. The abnormality values are ranked and the number of the abnormal value pulses belonging to the respective ranks is determined. This number is compared with the previously detected number of the abnormal value pulses of a standard sample having the known grain size distribution of the inclusions as a reference value, by which the grain size distribution is determined. The grain size distribution is rapidly and exactly analyzed in this way.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to an optical emission spectroscopic analysis method for analyzing the components of steel materials, and in particular to an optical emission spectroscopic analysis method for measuring the particle size distribution of inclusions in steel. Concerning spectroscopic analysis methods.

[Prior Art] The addition of aluminum to steel serves as an important factor in determining the quality of the steel, in addition to its function as a deoxidizing agent. The forms of aluminum in steel are acid-soluble aluminum (hereinafter referred to as Sol, Ajll) solid-solved in steel, and acid-insoluble aluminum (hereinafter referred to as 1nso1.
There are two forms: A and Q).

In the steelmaking process, the amounts of Sol and Ail are determined for each type of steel, and in actual operations, Sol and Ail are determined based on analysis results.
The amount of aluminum added is adjusted so that the amount of aluminum is within the standard range. Most of the added aluminum dissolves into Sol, A11, but some of it becomes In5o1. Becomes Al1. In5o1. When a large amount of Ajll is present in steel, alumina inclusions are generated, and the product is likely to have defects such as surface scratches.

In particular, when a large number of large-grain alumina inclusions are present in steel, cracks are likely to occur starting from the inclusions, and the fatigue properties of the product are significantly reduced. For this reason, steel products such as deep-drawn materials require high cleanliness, so it is necessary to accurately grasp the particle size distribution of alumina inclusions in steel at each stage of the steel manufacturing process.

Generally, methods for measuring the average particle size of inclusions in steel include sand analysis and microscopy. In the sand analysis method,
A sample is dissolved in acid, inclusions such as alumina in the residue are screened out, and the particle size distribution of these inclusions is measured. However, the Sandoz analysis method requires complicated operations and requires analysis time of 2 to 5 days, making it impractical.

[Problem to be solved by the invention] The microscopy method is specified in JIS standard GO555.

In this method, the sample must be mirror-finished, and it takes one to two days for sample preparation and measurement, making it impossible to obtain analytical results quickly.

In recent years, computer image analysis methods have been developed to speed up analysis, but they are prone to errors due to polishing scratches and adhesion of dust. Furthermore, computer image analysis methods have drawbacks such as difficulty in determining the type of inclusions.

The present invention was made in view of the above circumstances, and an object of the present invention is to provide an emission spectroscopic analysis method for steel that can quickly and accurately analyze the particle size distribution of inclusions in steel.

[Means for Solving the Problems] The steel emission spectroscopic analysis method according to the present invention applies a pulse discharge to a sample, detects the emission intensities of the spectroscopic spectral lines of the iron element and the analysis target element for each pulse discharge, and performs detection. The correlation between the emission intensity of the iron element and the emission intensity of the analyte element is determined using a statistical method, and from this correlation, the degree of abnormality is determined for each abnormal value pulse of the analyte element in the sample. Rank the abnormal value pulses using the degree of abnormality, determine the number of abnormal values/curs belonging to each rank, and use the number of each abnormal value pulse as an index to determine the particle size distribution of inclusions containing the target element to be analyzed. It is characterized by

[The action sample contains soluble aluminum and insoluble aluminum (
(alumina inclusions) coexist, or the two exhibit different behavior in luminescence intensity. In general, the excitation efficiency of insoluble aluminum is higher than that of soluble aluminum, so the presence of a large amount of insoluble aluminum causes a positive error in the PDA method.

In the steel emission spectroscopic analysis method according to the present invention, when a pulse discharge is applied to a sample, a large number of abnormal value pulses are detected. In these abnormal value pulses, the emission intensity of the element to be analyzed becomes abnormally higher than the emission intensity of the iron element.

The pulse group detected in this manner is processed by a regression method, and the lower tangent line of the pulse group is determined as a correlation equation. Furthermore, this downward tangent correlation equation is transformed using statistical analysis methods,
Find the upper tangent. This upper tangent provides a threshold value for distinguishing abnormal value pulses from other normal value pulses, and a pulse that exceeds the upper tangent is determined to be an abnormal value pulse.

Next, the degree of abnormality is determined for each abnormal value pulse.

The degree of abnormality of the abnormal value pulse is the degree of abnormality in how the emission intensity value of the target element deviates from the upper tangent (degree of abnormality b/a)
is the hail index. All the abnormal value pulses are ranked in terms of the degree of abnormality b/a, and the particle size distribution of inclusions is determined using the number of abnormal value pulses belonging to each rank as an index. In this case, pulse discharge is performed on a standard sample whose particle size distribution of inclusions is known, and abnormal value pulses are detected in advance for each particle size.
These abnormal value pulses are used as reference values. The particle size distribution of inclusions in the sample is determined by comparing these reference values with the actually measured abnormal value pulses.

[Embodiments] Various embodiments of the present invention will be specifically described below with reference to the accompanying drawings.

In this example, samples are taken from the bottom and middle parts of a continuously cast slab, and each sample is analyzed by emission spectroscopy to measure the particle size distribution.

After a pre-discharge of 5 seconds between the sample and the electrode, an additional 5
Pulse discharge is performed for seconds, and the spectral lines are received by a photomultiplier tube to detect the emission intensities of iron and aluminum elements, respectively. In this case, the frequency of pulse discharge is 40
It is 0 hertz.

The photomultiplier tube of the emission spectrometer is connected to the input side of the data processing device. FIG. 1 shows the relationship between the luminescence intensities of iron element and aluminum element obtained by discharging the sample. In Figure 1, the number of plots may be simplified for convenience, or there may actually be 200 luminescence plots in one analysis.
It consists of 0 plot groups. The horizontal axis of the screen shows the luminescence intensity of the iron element (intensity increases from left to right on the screen), and the vertical axis shows the luminescence intensity of aluminum element (intensity increases from the bottom of the screen to the top) shows.

The CPU of the data processing device has various programs for executing statistical analysis. All the data of the luminescence plots are temporarily stored in the memory section of the data processing device, and the data are analyzed using various statistical analysis techniques.

Next, the detected data is analyzed using various statistical methods,
The procedure for finding a regression line for a group of plots will be explained.

Two-point regression method [I]] Find the sum of the emission intensities of the elements (hereinafter referred to as "iron intensity"), and divide this by the number of pulses (2000) to find the average value AV of the iron intensity.

[n] Exists in a region where the iron strength is equal to or higher than the average value Av, and
The fifth plots with the lowest emission intensity of the aluminum element (hereinafter referred to as "aluminum intensity") are extracted, and the average value FH of these five iron intensities and the average value AH of the aluminum intensities are determined, respectively.

[m] Exists in a region where the iron strength is less than the average value Av, and
The fifth plots with the smallest emission intensity of the aluminum element (hereinafter referred to as "aluminum intensity") are extracted, and the average value FL of these five iron intensities and the average value AL of the aluminum intensities are determined, respectively.

[IV] The intersection point P of the average values (FH, AH) representing the upper region and the average value (FL) representing the lower region.

Find the equation of the straight line that passes through the intersection Q of AL) and the two points. This linear equation can be expressed as the following equation (1) as a correlation equation representing the downward tangent line of iron strength and aluminum strength.

(Aff)-(Fe)xA, 10B+...(
1) However, (1) is the aluminum strength, (Fe) is the iron strength, A1 is the slope of the straight line on the XY coordinates, and B1 is the X
The intercepts of straight lines on the Y coordinate are shown.

Iron Column Least Squares Regression Method [I]] The sum of the emission intensities of the elements (hereinafter referred to as "iron intensity") is determined, and this is divided by the number of pulses (2000) to determine the average value AV of the iron intensity.

[11] Divide the average value Av by 10 to find the column width. The plot group existing in the region where the iron strength is below the average value AV is equally divided into 10 columns L1 to LIO.

[m]] Iron strength average value FVI and aluminum strength average value AVI of the n-th plots from the lowest aluminum strength among the plots existing in 1 column L.
seek.

[IV] For the second column L2 to the tenth column LIO, the iron strength average values FV2 to FV are determined by the same procedure, respectively.
VIO and aluminum strength average value AV2 ~ AVIO
seek.

[V] Average value (FVI) representing each cikaram.

10 intersection points corresponding to AVI) to (FVIO, AVIO) are subjected to linear regression using the method of least squares to obtain a straight line equation. This linear equation can be expressed as the following equation (2) as a correlation equation representing the lower tangent line of iron strength and aluminum strength.

(AΩ) - (Fe)
The intercepts of straight lines on the Y coordinate are shown.

Regression convergence correlation coefficient determination method [1] From the correlation diagram, iron strength and aluminum strength are subjected to linear regression using the least squares method to obtain equation (3) below.

(Ai)) - (Fe) XA3 +83
- (3) However, A indicates the slope of the straight line on the XY coordinates, and B3 indicates the intercept of the straight line on the XY coordinates. In this case, the correlation coefficient representing the degree of dispersion of the plot group is small.

[II] Discard the plot groups that exist in the area above the straight line corresponding to equation (3) above, perform linear regression on the iron strength and aluminum strength using the least squares method for the plot groups that exist in the area below the straight line, and calculate the following: Find equation (4).

This increases the correlation coefficient.

(AN) (F e)XA4' +B4 -(
4) However, (All) is aluminum strength, (F e
) is the steel strength, A4 is the slope of the straight line on the XY coordinates, B
4 indicates the intercept of a straight line on the XY coordinates.

[m] Increase the correlation coefficient by repeating the linear regression and upward rejection operations as described above, stop the repeated calculations when the correlation coefficient exceeds a predetermined value, and find the regression equation at that time. The final regression equation is shown in equation (5) below.

(Aj)) - (Fe) , find the following equation (6).

(IQ) - (Fe)XA6+B6 , , (6) where A6 is the slope of the straight line on the
Each shows the intercept of a straight line on the coordinates. In this case, the correlation coefficient representing the degree of dispersion of the plot group is small.

[II] Discard the plot groups that exist in the area above the straight line corresponding to the above equation (6), linearly regress the iron strength and aluminum strength using the least squares method for the plot groups that exist in the area below the straight line, and calculate the following: Find equation (7).

This increases the correlation coefficient.

(AjJ)-(Fe)XA7+87-
(7) However, A7 indicates the slope of the straight line on the XY coordinates, and B7 indicates the intercept of the straight line on the XY coordinates.

[Inl As described above, the correlation coefficient is increased by repeating the linear regression and upward rejection operations, and when the correlation coefficient exceeds a predetermined value, the repeated calculation is stopped and the regression line at that time is determined. The distance d from this provisional regression line to each plot (remaining plot) is determined, and its standard deviation σ6 is determined using the following equation (8). However, N is the number of residual plots.

σ, -Σd2/ (N-1) ... (8) [
IV] Plots with more than twice the standard deviation σ are rejected as abnormal values, or plots with a distance d from the provisional regression line that is 10% from the farthest are rejected. After rejecting abnormal values, linear regression is performed again using the least squares method to obtain a regression line. This final regression line is expressed by the following equation (9).

(AN)-(Fe)xA, +B, -(9) where A indicates the slope of the final regression line on the XY coordinates, and B9 indicates the intercept of the final regression line on the XY coordinates.

Regression Convergence Pulse (Plot) Litter Method [I] From the correlation chart, iron strength and aluminum strength are linearly regressed by the least squares method to obtain the following equation (10).

(A, & )−(F e)XA+o+B+o −(1
0) However, AHO is the slope of the straight line on the XY coordinates,
BIO each indicates a straight line segment on the XY coordinates.

[II] Discard the plot groups that exist in the area above the straight line corresponding to the above equation (10), linearly regress the iron strength and aluminum strength using the least squares method for the plot groups that exist in the area below the straight line, and calculate the following: Find equation (11).

(All)-(F e) ×A+++Bz-(11) However, A11 is the slope of the straight line on the XY coordinates, Bll
indicate the intercept of a straight line on the XY coordinates.

[III] Reduce the number of residual plots by repeating the linear regression and upward rejection operations as described above, and stop the repeated calculation when the number of plots becomes less than a predetermined number (for example, 100), and then calculate the regression line at that time. seek. This final regression line is expressed by the following equation (12).

(Ai))-(Fe)XA12+B12-(1
2) However, A3. represents the slope of the final regression line on the XY coordinates, and BI2 represents the intercept of the final regression line on the XY coordinates.

Regression Convergence Pulse (Plot) Bedding Constant Abnormal Pulse Rejection Method [I] From the correlation diagram, the iron strength and aluminum strength are linearly regressed by the least squares method to obtain the following equation (13).

(AN)-(Fe)XA12+B12-(1
B) However, AI3 is the slope of the straight line on the XY seat ridge,
BI3 indicates the intercept of a straight line on the XY coordinates.

[II] Discard the plot groups that exist in the area above the straight line corresponding to the above equation (13), perform linear regression on the iron strength and aluminum strength using the least squares method for the plot groups that exist in the area below the straight line, and calculate the following: Find equation (14).

(AN) - (F e) XA14 + B14 - (14) However, A14 is the slope of the straight line on the XY seat machine,
indicate the intercept of a straight line on the XY coordinates.

[I [I] Reduce the number of residual plots by repeating the linear regression and upward rejection operations as described above, and stop the iterative calculation when the number of plots becomes less than a predetermined number (for example, 100). Find a provisional regression line. The distance d from the provisional regression line to each residual plot is determined, and the standard deviation σ is determined using the above equation (8).

[IV] Plots that exceed twice the standard deviation σ are rejected as abnormal values. After rejecting outliers, linear regression is performed on the remaining plot groups using the least squares method to obtain a final regression line. This final regression line is expressed by the following equation (15).

(AII) −(Fe) XA15+ Bus −
(15) However, AI indicates the slope of the final regression line on the XY ridge, and B15 indicates the intercept of the final regression line on the XY coordinates.

Determine the regression line (lower tangent) of the plot group using one of the six methods described above.

Next, the above regression line (hereinafter general formula (Ai)) - (Fe)
A case will be described in which the particle size distribution of alumina inclusions is determined by the following method using xA+B).

[I] Determine the threshold value of the abnormal value pulse existing in the upper region using either method (1) or (2) below.

(2) Find equation (A11)-(Fe)XAXN+B.

However, N is a constant.

The above equation is shown as straight line G in FIG. The degree of abnormality b/a is determined for each abnormal value pulse, and each abnormal value pulse is ranked based on the degree of abnormality b/a. Abnormality level b/a
is determined by the following equation (16).

b/a-[(8Ω) -1(Fe) x AX N+
However, this calculation of the degree of abnormality is performed only for abnormal value pulses existing in the region above the straight line G.

■ Find the equation (AN)=(F e)XAXN+C.

However, N is a constant, C-FHXA10B.

The above equation is shown as straight line H in FIG. The degree of abnormality b/a is determined for each abnormal value pulse, and each abnormal value pulse is ranked according to the degree of abnormality b/a. Abnormality level b/a
is determined by the following equation (17).

b/a= [(Aff) −f(Pe) X A x
N+C1]/(Pe) (17) However, this degree of abnormality is calculated only for abnormal value pulses that exist in the region above the straight line H.

[n] The ranked abnormal value pulses are totaled for each rank, and the number of abnormal value pulses belonging to each rank is determined.

Each rank is, for example, particle size is 5 μm or less, 5 to 10
μm, divided into 5 μm increments such as 10 to 15 μm. The correlation between the ranking based on the degree of abnormality and the particle size distribution of inclusions is determined in advance using a standard sample with a known particle size distribution. Find the diameter distribution.

In FIGS. 3 and 4, the horizontal axis represents the grain size of alumina inclusions in the bottom part of the steel ingot and the middle part of the steel ingot, and the vertical axis represents the amount of inclusions at each grain size measured by the method of the present invention. , and are distribution charts obtained by examining the particle size distribution of inclusions.

In Figures 5 and 6, the horizontal axis shows the particle size of alumina inclusions in the bottom and middle parts of the steel ingot, and the vertical axis shows the amount of inclusions at each particle size measured by the sand analysis method. , and are distribution charts obtained by examining the particle size distribution of inclusions.

As is clear from the figure, the analysis results obtained by the method of the present invention are in good agreement with the results obtained by the sand analysis method, and the method of the present invention can accurately measure the particle size distribution of each part of the ingot.

According to the method using the above abnormal value pulse abnormality degree determination method, the time required from the start to the end of analysis is approximately 30 seconds (2
(in the case of multiple analysis), and it takes approximately 15 minutes including the sample preparation time.
This method can fully meet the demand for faster measurement of the particle size distribution of alumina inclusions.

In addition, although the above example describes the case where the particle size distribution of alumina inclusions is measured, the present invention is not limited to this only (the particle size distribution of other types of inclusions such as Mn5SiO□, TiO2, etc.) is measured. It can also be used to measure distribution.

[Effects of the Invention] According to the present invention, the particle size distribution of inclusions in steel can be accurately measured. Further, according to the present invention, the time required from the start to the end of analysis can be shortened, and analysis results can be obtained more quickly than with conventional methods. For this reason,
It is particularly effective in quality control when refining clean steel such as deep-drawn steel, and can greatly contribute to improving the quality of steel products.

[Brief explanation of drawings]

FIGS. 1 and 2 schematically show a group of light emission pulses, and are diagrams for explaining the procedure for measuring particle size distribution by the abnormal value pulse abnormality determination method, and FIGS. 3 and 4 respectively show the present invention. Distribution diagram showing the particle size distribution of inclusions according to the method, No. 5
Figure 6 and Figure 6 are distribution charts showing the particle size distribution of inclusions by the sand analysis method, respectively, for comparison. Applicant's agent Patent attorney Takehiko Suzue 1 history Marugametsu 91 light j ki cliff, fig. life νμ Fusagido Tom, Ichiriku 1) b2 Kui Y1! Mystery (1m) Figure 3 Seven A and Mr. Beto 51) Stone 1 Town Higa L (, pm) Figure Figure 1 → - Middle i & Majesty t Hede D Old Nine Diameter (1m) Figure 4 @ Tun middle! Toy Y & Monoro Kai (Hm) Figure

Claims (1)

[Claims] A pulse discharge is applied to the sample, and the emission intensities of the spectral lines of the iron element and the target element are detected for each pulse discharge, and the emission intensity of the iron element and the luminescence of the target element are calculated using statistical methods for the detected luminescence intensities. The intensity correlation is determined, and the degree of abnormality is determined for each abnormal value pulse of the target element in the sample from this correlation.The abnormal value pulses are ranked using these abnormality degrees, and the abnormal value pulses belonging to each rank are determined. A method for optical emission spectroscopic analysis of steel, characterized in that the particle size distribution of inclusions containing the target element to be analyzed is determined using the number of abnormal value pulses as an index.