JP6994749B2 - Plating liquid judgment method and data processing equipment - Google Patents

Description

The present invention relates to a method for determining a plating solution that changes with use, and a data processing apparatus.

When managing the treatment liquid that is used repeatedly, it is common to perform instrumental analysis such as a photometer or liquid chromatograph, quantitatively measure the change status of the contained components, and make adjustments such as replenishment of deficient components. Is. In the liquid chromatograph, the unknown sample is measured after measuring the standard sample, and if the unknown sample has a peak whose retention time matches the peak of the known component of the standard sample, the component can be qualitatively and quantified. ..

For components that cannot be determinized without a standard sample, a method for quantifying unknown components using a spectrum library that stores spectral data of known components is disclosed. (Patent Document 1)

In addition to the metal component which is the main component of the plating solution, various additives are blended for the purpose of smoothing the plating surface, controlling physical properties, uniform electrodeposition, and the like. As the plating is repeated, not only these additives are consumed, but also their concentrations decrease by reacting with coexisting substances to form or decompose compounds. In addition, foreign matter increases in the plating solution due to the contamination of equipment such as a plating tank and eluents from the object to be plated. Since such a change in the additive concentration and an increase in the foreign matter concentration cause plating defects, the additives in the plating solution and the foreign matter concentration should be grasped, and each component such as the additive in the plating solution before the defect occurs. It will be necessary to replenish the plating solution and replace the plating solution itself. A method for quantifying additives and decomposition products thereof in a plating solution and evaluating the deterioration behavior of the plating solution is disclosed. (Non-Patent Document 1)

As a data processing method, a technique for creating a one-dimensional table of signal intensity data obtained by a liquid chromatograph, a gas chromatograph, etc., whose components are known and fluctuates with the passage of time, and performing multivariate analysis is disclosed. (Patent Document 2).

Japanese Patent No. 2924267 Japanese Patent No. 5327388

Keisuke Ogura and 4 others, "Evaluation of Additive Deterioration Behavior of Copper Sulfate Plating Using High Speed Liquid Chromatography", Surface Technology Vol. 65, No. 1, pp. 35-40, 2014

When the treatment liquid such as the plating liquid is replenished or replaced at regular intervals and measured by a liquid chromatograph, the detected components react with not only the additives but also the coexisting substances in the plating liquid. There are various components such as compounds formed in the above, equipment such as plating tanks, and eluates from the object to be plated, and it is difficult to prepare standard substances for all components and perform qualitative and quantitative analysis.

In addition, in the process of aging due to the use of the treatment liquid, not limited to the plating liquid, problems may occur in the object treated with the treatment liquid due to a decrease in the required components or active ingredients of the treatment liquid and an increase in foreign substances. However, since the treatment liquid contains a plurality of additives whose composition is not disclosed, it is difficult to identify all the components.

As described above, when creating a control index for a treatment liquid containing unknown components, it is necessary to extract data on the components that affect the properties of the treatment liquid or the analysis equipment derived from those components. It becomes.

An object of the present invention is a determination method and a data processing apparatus for managing a plating solution by performing multivariate analysis using data which is considered to have a large influence on the properties of the processing solution even if the components are not known. Is to provide.

In order to solve the above problems, in the plating solution determination method of the present invention, the plating solution as a sample is measured by an analytical instrument, and the signal intensity of the detector with respect to the characteristics such as time or wavelength is acquired to determine the properties of the sample. In the evaluation method for judgment, samples whose properties are known are grouped into normal groups and defective groups according to their properties and measured as learning samples, and the signal strength of the detector for the sample properties and measurement conditions and characteristics is saved as a data file. , A peak table selection step that extracts multiple signal strengths as peaks from the saved data file, and a peak table creation step that creates a peak table with the extracted signal strength and characteristics and the allowable range of characteristics. Also, a peak table division step of dividing the peak table into a training table and a test table so as to include two or more data files of the normal group and the bad group, and a multivariate analysis are performed from the created peak table to set a discrimination threshold. In the discrimination threshold setting step , the data file in which the test table is grouped according to the discrimination threshold is appropriate using the test table created in the peak table division step. It has a step of determining whether or not it is discriminated into the normal group and the defective group, maintaining the discrimination threshold if it is properly determined, and resetting the discrimination threshold if it is not properly determined. It is characterized in that the quality of an unknown sample is judged using the discriminant threshold.

Further, in the peak selection step of each of the above data files, the data file having the largest number of extracted peaks is selected from the result of waveform processing the signal intensity data for the characteristics of each data file and extracting the signal intensity peaks above the threshold value. It is characterized by having a process of performing.

Further, in the discrimination threshold setting step, an arbitrary peak is extracted from the created peak table, and the property is calculated from the combination of peaks having the maximum difference in the discrimination score created from the discrimination threshold between the grouped properties. It is characterized by having an optimization process that clarifies the difference between the two.

In order to solve the above-mentioned problems, the data processing apparatus of the present invention measures the data obtained by measuring the plating solution as a sample whose properties are known as a learning sample grouped into a normal group and a defective group according to the classification of the properties. A data storage unit that saves the output signal strength of the detector for the properties of the sample and measurement conditions and characteristics as a data file, and multiple signal strengths are selected as peaks from each saved data file, and the selected peaks and characteristics are allowed. Create a peak table with a range , divide the peak table into a training table and a test table so that each table contains two or more data files of normal group and bad group, and perform multivariate analysis from the created peak table. And set the discrimination threshold, and using the test table, it is determined whether the data files grouped by the test table are properly discriminated into the normal group and the defective group, and it is appropriate. It is characterized by having a data processing unit that maintains the above-mentioned discrimination threshold when the judgment is made, resets the discrimination threshold when the judgment is not appropriate, and judges the quality of the sample using the set discrimination threshold. There is.

According to the present invention, even if the components are not known, it is determined to manage the plating solution by performing multivariate analysis using the data of the analyzer derived from the components which are considered to have a large influence on the properties of the plating solution. Methods and data processing devices can be obtained.

(A) It is a figure which shows the structural example of the liquid chromatograph analyzer as an example of the analyzer which concerns on this application (b) is a figure which shows the structural example of the data processing apparatus which concerns on this application. An example of a chromatogram of a plating solution (a) a chromatogram of a new plating solution (b) a chromatogram of a plating solution after repeated use (c) a chromatogram of a plating solution in which a defect has occurred It is a figure which shows the evaluation / management process flow. It is a figure which shows an example of a peak table. It is a figure which shows the peak selection process flow. It is a figure which shows the evaluation / management process flow including the optimization of a peak table element.

In the test liquid evaluation method according to the present invention, an example of performing plating liquid management using a liquid chromatograph will be described.

In addition to the metal component, which is the main component, the plating solution contains various additives for the purpose of smoothing the plating surface, controlling physical properties, uniform electrodeposition, and the like. With repeated plating, these additives are not only consumed, but also decompose to decrease in concentration or react with coexisting substances to form compounds. In addition, foreign matter increases in the plating solution due to the contamination of equipment such as a plating tank and eluents from the object to be plated. Since such a change in the additive concentration and an increase in the foreign matter concentration cause plating defects, the additives in the plating solution and the foreign matter concentration should be grasped, and each component such as the additive in the plating solution before the defect occurs. It is necessary to replenish the plating solution and replace the plating solution itself.
In addition, for management of replenishment, replacement, etc. of the plating solution, generally, quantitative analysis of known components is performed, or plating is actually performed using a dedicated jig such as an electrodeposition test, and the electrodeposition state is obtained. By evaluating the above, the excess and deficiency of additives and the influence of foreign substances were evaluated.

In this embodiment, a liquid chromatograph is used for a plating solution capable of performing normal plating and a plating solution for which foreign matter increases due to repeated plating or a decrease in the concentration of each component of an additive is known to be the cause of plating failure. This is an example in which the result of multivariate analysis using the obtained chromatogram is used for the evaluation of the plating solution. The data file is composed of a chromatogram at the time of measurement, measurement conditions, waveform processing results, signal intensity transition (peak) of the detector, and the like. The chromatogram is a plot of the transition of the signal intensity of the separated components, with the elution time as a characteristic on the horizontal axis and the signal intensity on the vertical axis. The elution time is called the retention time. The signal strength of the chromatogram is proportional to the concentration of the eluted components during its retention time. Since there are a plurality of components that elute at the same holding time depending on the analysis conditions of the liquid chromatograph, a plurality of components may coexist in one peak.

FIG. 1A is a schematic view showing a configuration example of a liquid chromatograph analyzer as an example of the analyzer according to the present invention.
In the liquid chromatograph 100, the mobile phase contained in the mobile phase container 1 is sent by the pump 2. The measurement sample injected by the autosampler 3 is moved by the mobile phase and held in the column 4 kept at a constant temperature in the column oven 5. The sample held in the column 4 is sequentially eluted and separated according to the properties of its components. The separated components are detected by the detector 6, and the signal of each component is collected as a chromatogram in the arithmetic processing unit 7 and displayed on the display unit 10. The data processing unit 7 includes a data holding unit 8 and an arithmetic processing unit 9. The detector 6 includes a UV detector, a three-dimensional detector, a differential refractive index detector, a mass detector, and the like, but the detector 6 is not particularly limited.

The holding time varies depending on the condition of the column, the device, and the like. As a preparation before measurement, it is necessary to measure an arbitrary calibration sample and correct the fluctuation of the holding time in advance in order to perform accurate measurement.

The treatment liquid evaluation method according to the present invention will be described below.
The evaluation / control process using discriminant analysis for evaluating the state of the plating solution will be described with reference to the drawings.

First, as the judgment threshold setting step shown in FIG. 3, the measurement condition setting step S10, the learning sample measurement step S20, the peak selection step S30, the peak table creation step S40, the peak table division step S50, the discrimination threshold calculation step S60, and the threshold correctness / rejection are performed. There is a determination step S70 and a determination threshold setting step S80.

First, the measurement conditions of the liquid chromatograph are determined in the measurement condition setting step S10. The measurement conditions of this liquid chromatograph are applied to all the samples for which the discrimination threshold for evaluation and control is obtained and the samples sampled from the process for process control.

Next, as described above in the learning sample measurement step S20, a plurality of samples grouped into the normal group and the defective group are prepared for each group, and the measurement conditions of the liquid chromatograph set in the measurement condition setting step S10 are used. Measure. As a measurement result, a chromatogram of signal intensity with respect to the elution time (characteristic) as shown in FIG. 2 is obtained. Here, the normal group is a group of plating solutions that do not cause defects in the plating film or plating substrate immediately after preparation or after plating, and the defective group is a group of plating solutions that cause defects in the plating film or plating substrate after plating or immediately before waste liquid. It is a group of plating solutions. FIG. 2 shows an example of measurement data (hereinafter referred to as a chromatogram) of the plating solution by a liquid chromatograph. FIG. 2 is a chromatogram measured by sampling the plating solution from the state of the new solution to the occurrence of plating defects. FIG. 2 (a) is a chromatogram of a new plating solution, FIG. 2 (b) is a chromatogram of a plating solution after repeated use, and FIG. 2 (c) is a chromatogram of a plating solution in which plating defects have occurred due to repeated use. The numbers described in each chromatogram are peak numbers, and the same numbers are the peaks with the same retention time. The peak numbers were numbered in order from the peak with the shortest retention time.

Comparing the intensities of the peaks, not only the gradually increasing peaks (peak numbers 2, 4, 5, 6, 7 in FIGS. 2 (a) to (c)) but also the gradually decreasing peaks (FIGS. 2 (a) to (c)). There are also peak numbers 3) of c) and peaks that increase or decrease (peak numbers 1 and 8 of FIGS. 2A to 2C). These increases and decreases in peak intensity are due to the consumption of additives, the reaction and decomposition with other components, the reaction between the components in the plating solution, or the elution from the environment such as the plating tank.

Next, in the peak selection step S30, the peak is selected by the procedure described later.
Next, in the peak table creation step S40, as shown in FIG. 4 based on the peak selected in the peak selection step S30, the peak number assigned in the peak selection step S30 and the holding time of the numbered peak are detected. Create a peak table that contrasts the peak areas. The allowable range of the holding time when the holding time deviates due to measurement conditions or other factors is determined by the degree of variation in the repeated measurement in the measurement condition setting step S10, and is reflected in the holding time of the peak table.

Next, in the peak table division step S50, the peak table is divided into a training table and a test table. The division into the training table and the test table should include two or more data files of the normal group and the bad group in each table so that the sampling date and the strain are not biased.
Next, in the discrimination threshold calculation step S60, a discrimination function is created using the training table created in the peak table division step S50, and the discrimination score of each data file of the training table is calculated based on the created discrimination function. The calculated discrimination score is plotted based on the above-mentioned group data of the normal group and the defective group, and the discrimination threshold that is the threshold for discrimination between the normal group and the defective group is set.
Next, in the threshold value correctness determination step S70, it is determined whether or not the threshold value is appropriate by using the test table created in the peak table division step S50. At the time of this determination, if the data files grouped in the test table are properly discriminated into the normal group and the defective group, the calculated threshold is used for re-preparation and replacement of the plating solution in the discrimination threshold setting step S80. It is set as a pass / fail determination threshold used as a process control index. If it is not properly determined, the process returns to the peak selection step S30, the peak selection step S30 to the threshold value correctness determination step S70 are repeated, and the threshold value for proper determination is calculated.

As the second pass / fail determination step, there are an unknown sample measurement step S90, a discrimination score calculation step S100, and a pass / fail determination step S110.

Using the measurement conditions set in the measurement condition setting step S10 in the unknown sample measurement step S90, the sample sampled from the step is measured by a liquid chromatograph, and the data file is saved.
Next, in the discrimination score calculation step S100, the peak area is calculated by performing waveform processing on each data file saved in the unknown sample measurement step S90 using the same criteria as in the peak selection step S30 described later. The peak for which the peak area is calculated is compared with the peak number set in the peak table creation step S40 from the peak elution time, and the area value of the peak number having the same peak elution time is created in the discrimination threshold calculation step S60. The discrimination score is calculated by substituting it into the discrimination function.
Next, in the quality determination step S110, the discrimination score calculated in the discrimination score calculation step S100 is compared with the threshold value set in the discrimination threshold setting step S80, and the quality of the sample sampled from the step is determined.

Subsequently, the details of the peak selection step S30 will be described with reference to FIG.
In the peak selection step S30, the samples grouped into the normal group and the defective group are grouped, and the chromatograms of the individual data files measured in the learning sample measurement step S20 are baselined according to the peak shape in the waveform processing step S510. Set. The parameters for setting the baseline include noise value, peak width, slope of peak start point and end point, and are selected according to the characteristics and purpose of the data file.
After setting the baseline, select whether to use the signal strength (peak height) to create the discriminant function or the peak area as the integrated value of the signal strength. Here, an example using the peak area will be described.
Next, in the signal strength / peak area calculation step S520, the area of each peak based on the baseline set in the waveform processing step S510 is calculated.

Next, in the peak determination step S530, the preset threshold value and the area of each peak calculated in the signal intensity / peak area calculation step S520 are compared, and the threshold value or more is set as the determination peak, and it is determined whether or not the peak is a determination peak. , The peak regarded as the determination peak is extracted in the peak extraction step S540.
Next, in the peak number counting step S550, the number of peaks included in each data file extracted in the peak extraction step S540 is counted.
Next, in the data file selection step S560, the number of peaks extracted in the peak extraction step S550 included in each data file is compared, and the data file containing the largest number of peaks is selected.
Next, in the peak number numbering step, the determination peaks of the data files selected in the data file selection step S560 are numbered in the order of elution time. The number of each peak numbered here becomes the peak number of the peak table created in the peak table creation step S40.

In the steps from the peak table creation step S40 to the discrimination threshold setting step S80 of the judgment threshold setting step described in the first embodiment, discrimination analysis is performed using all the elements of the peak table, and the discrimination threshold is set. However, the elements of the peak table may be optimized.
FIG. 6 shows an evaluation / management process using discriminant analysis for evaluating the state of the plating solution, including the process of optimizing the elements of the peak table.

The measurement condition setting step S10 to the peak table dividing step S50 in FIG. 3 perform the same steps.
The method of optimizing the peak table in the next peak table optimization step S650 is shown below.
1) Discriminant function K ₀ (0) created using the area values of all peaks in the training table 2) Mean value of discriminant score and discriminant score (g ₀ and u ₀ ) from each test table of normal group and bad group grouped ) Is calculated g ₀ : Average value of the discrimination score of the normal group calculated using the area values of all peaks in the test table u ₀ : Mean value of the discrimination score of the defective group calculated using the area values of all peaks in the test table 3 ) Discrimination between normal group and bad group in the test table Difference in average score D _m (n) is calculated D _m (n) = | g _m - _um |
g _m : Average discrimination score of normal group u _m : Average discrimination score of defective group m: Number of peaks not added to discrimination function calculation n: Peak number excluded from discrimination function calculation and discrimination score calculation All peaks in peak table When the area value of is used, D ₀ (n) = | g ₀ -u ₀ |
4) Exclude the area value of peak number 1 from the training table 5) Discrimination function K ₁ (0) Create 6) Exclude the area value of peak number 1 from the test table 7) Mean value of the discrimination score of the test table g ₁ , U ₁ Calculation 8) Calculate the difference between the average value of the discrimination scores between the normal group and the bad group in the test table D ₁ (1) = | g ₁ -u ₁ |
9) Repeat steps 4) to 8) in order to exclude the area value of each peak up to the total number of peaks N in the training table, and create _a discriminant function K1 (n). 4) to 8) are repeated in order up to the number N, excluding the area value of each peak, and D ₁ (n): n = 1 to N, N = peak extraction step S504 for calculating D ₁ (n). Total number of peaks 11) Extract the maximum value D ₁ (n) max from N D ₁ (n) calculated in 10), specify the n (nth peak), and set it as n _max 12) Training table And exclude the area value of the peak number n _max from the test table 13) Repeat 4) to 8) for the peak numbers 1 to N excluding the peak number n _max , and calculate D ₂ (n) 14) 13) The maximum value D ₂ (n) max is extracted from the N-1 D ₂ (n), and the n (nth peak) is specified and set as n _max . 15) 4) to 13) are D _m ( The discriminant function created from the training table excluding the peak area value of the peak number when m of n) becomes N-1 and D _m (n) max is specified 16) D _m (n) max. Let it be K _m (n) max.
17) Similarly, calculate the average value of the discrimination score using the data file excluding the peak area of the corresponding peak number from the test table, and confirm that it can be discriminated into the normal group and the defective group. 18) 17) In this case, K _m (n) max is used as the discriminant function for good / bad judgment, and K0 (0) using all peak areas is used as the discriminant function. K _m set in 18) above. (N) max is set in the discriminant function setting step S80 as the discriminant function optimized in the peak table optimization step S650.
In the subsequent pass / fail determination step, the unknown sample measurement step S90 to the pass / fail determination step S110 in FIG. 3 are determined by the same procedure.

Here, as the signal strength, either a signal strength that can be processed in the waveform processing step or an area value that is an integral value with respect to the characteristics of the signal strength may be used.

The data processing apparatus according to the embodiment of the present invention will be described with reference to the drawings by taking a liquid chromatograph as an example.
The data processing device of the liquid chromatograph uses a separation means for separating a plurality of components contained in the sample as a characteristic of time, and an analyzer equipped with a detection means for acquiring the signal strength of each of the separated components, and uses a plurality of samples. On the other hand, it is a device that performs data processing of signal strength data of components separated by time. The data processing apparatus 7 of FIGS. 1A and 1B includes a data holding unit 8, an arithmetic processing unit 9, a control unit 71, a recording unit 73, and the like.

The mobile phase is pumped by the pump 2 according to the instruction from the control unit 71. The measurement sample injected by the autosampler 3 is moved by the mobile phase and held on the column 4. The sample held in the column 4 is sequentially eluted and separated depending on the properties of its components. The separated components are detected by the detector 6 and stored in the data holding unit 8 as a data file together with information such as measurement conditions as a chromatogram of the signal intensity of the detector for the characteristics such as time and wavelength of each component. The chromatogram of each data file of the learning sample stored in the data holding unit 8 is first numbered by the signal processing unit 91 according to each step of FIG. Next, the arithmetic processing unit 92 sets the discrimination threshold according to each step of the determination threshold setting step of FIG. Next, the unknown sample measured in the same manner as described above follows the steps of FIGS. 3 to 6 in the signal processing unit 91 and the arithmetic processing unit 92, and is determined by the determination unit 93 as to whether or not it is necessary. The determination result is displayed on the display unit 10.

The above-mentioned data processing device is an independent data processing device attached to a device that can obtain a chromatogram that acquires a plurality of components contained in a sample as a signal strength of a detector for a characteristic, and the detection means includes time and time. As long as it is possible to acquire the signal strength using the wavelength as a characteristic, the type is not particularly limited. For example, a spectroscopic visible / ultraviolet detector, a spectroscopic fluorescence detector, a mass detector, a differential refractometer detector, or the like may be used.

It goes without saying that the present invention is not limited to the above embodiments and extends to various forms and equivalents included in the idea and scope of the present invention. For example, the measuring device includes a liquid chromatograph, a gas chromatograph, a capillary electrophoresis, a photometer, etc. that acquire the signal intensity of the detector with respect to characteristics such as time or wavelength, and the sample to be evaluated also uses a plating solution, a culture solution, etc. The target is a treatment liquid that controls changes over time.

100 Liquid chromatograph 1 Mobile phase container 2 Pump 3 Auto sampler 4 Column 5 Column oven 6 Detector 7 Data processing device 8 Data holding unit 9 Calculation processing unit 10 Display unit 71 Control unit 72 Measurement condition setting unit 73 Recording unit 91 Signal processing Unit 92 Calculation processing unit 93 Judgment unit

Claims (4)

In an analysis method in which the plating solution is measured by an analytical instrument, the signal strength of the detector with respect to the characteristics is acquired, and the properties of the plating solution are determined from the signal strength.
The plating solution whose properties are known is classified into a normal group and a defective group according to the classification of the properties, and measured as a learning sample.
A peak selection step of storing the signal strength of the detector for the properties of the plating solution and the measurement conditions and characteristics as a data file and extracting a plurality of signal strengths from each of the saved data files.
A peak table creation step of creating a peak table based on the extracted signal strength, characteristics, and allowable range of characteristics, and
A peak table division step of dividing the peak table into a training table and a test table so that each table contains two or more data files of the normal group and the bad group.
It has a discrimination threshold setting step of performing multivariate analysis from the created peak table.
In the discrimination threshold setting step, using the test table created in the peak table partitioning step, the data files grouped by the test table according to the discrimination threshold are properly discriminated into the normal group and the defective group. It has a step of determining whether or not the data is correct, maintaining the discrimination threshold if the judgment is appropriate, and resetting the discrimination threshold if the judgment is not proper.
The quality of the plating solution is determined using the discrimination threshold.
A plating solution determination method characterized by this. The peak selection step is
The claim is characterized by having a step of selecting the data file having the largest number of extracted peaks from the result of waveform processing the signal strength data for the characteristics of each data file and extracting the peaks of the signal strength equal to or higher than the threshold value. Item 1. The plating solution determination method according to Item 1. The discrimination threshold setting step is
Having an optimization step of extracting an arbitrary peak from the peak table at the time of the multivariate analysis and calculating from the combination of the peaks in which the difference in the discrimination score calculated from the discrimination threshold between the grouped properties is maximized. The plating solution determination method according to claim 1 or 2, wherein the plating solution is determined. It is a data processing device that measures the plating solution with an analytical instrument, acquires the signal strength of the detector with respect to the characteristics, and determines the properties of the plating solution from the signal strength.
The data processing device is
The data obtained by classifying the plating solutions with known properties into normal groups and defective groups according to the property classification and measuring them as learning samples,
A data storage unit that stores the signal strength of the detector for the properties of the plating solution, measurement conditions, and characteristics as a data file.
Select multiple signal strengths as peaks from each of the saved data files.
Create a peak table with the selected peaks and characteristics,
The peak table is divided into a training table and a test table so that each table contains two or more data files of the normal group and the bad group.
Multivariate analysis is performed from the peak table created above, a discrimination threshold is set, and
Using the test table, it is determined whether or not the data files grouped in the test table are properly discriminated into the normal group and the defective group according to the discriminant threshold value, and if it is properly determined, the discriminant threshold value is determined. If the judgment is not correct, reset the judgment threshold and reset the judgment threshold.
A data processing apparatus including a data processing unit that determines the quality of the plating solution using the set discrimination threshold value.

Images