JP2000131228A - Spectroscopic measurement method using ultraviolet rays and near infrared rays - Google Patents

Abstract

PROBLEM TO BE SOLVED: To accurately obtain a characteristic value of a sample from ultraviolet absorbance data and near infrared absorbance data, by eliminating influence of respective independent fluctuation factors in a near infrared spectroscopy process and an ultraviolet spectroscopy process, preferably, by eliminating influence of fluctuation factors common to both the processes. SOLUTION: A dichroic mirror 63 has a characteristic transmitting near infrared rays and reflecting ultraviolet rays. The mirror 63 converges light from two light sources 61, 62 to one light beam. Then, the light beam is transmitted through an interference filter 64. An interference filter disc 65 has plural interference filters 64 having different transmission wavelengths. The transmitted light beam through the interference filter 64 passes through a cell 66 and is received by a sensor. The sensor is a composite element comprising a silicon photodiode for receiving light in an ultraviolet area and a germanium photodiode for receiving light in a near infrared area. The sensor converts a light intensity into a current intensity, while an A/D converter converts the current intensity into a digital value.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for measuring characteristic values (concentrations, etc.) of a sample that absorbs characteristics in an ultraviolet region and a near infrared region.

[0002]

2. Description of the Related Art It is necessary to control the concentration of a chemical solution in a wet process such as cleaning and etching of a semiconductor wafer, for example, in a semiconductor manufacturing process. Regarding the measurement of the concentration of a chemical solution, in the measurement method described in Japanese Patent Application Laid-Open No. 3-175341 by the present applicant, a plurality of components are quantified based on the degree of change due to ion hydration using near infrared rays.
For example, the hydrogen peroxide spectrum is very similar to the water spectrum in the near-infrared region, and the slight difference is measured with high precision to quantify the hydrogen peroxide. However, there is a problem that the measurement accuracy is inevitably low for components having no clear characteristic absorption in that region, for example, for the determination of hydrogen peroxide in the case of a mixed aqueous solution of ammonia and hydrogen peroxide.

[0003]

However, a double bond (C = C, C = O, N = N) is added to the molecule of the component to be measured in the drug solution.
It is known that strong absorption is present in the ultraviolet region when is present. The above example of hydrogen peroxide also has strong absorption in the ultraviolet region. Therefore, it is considered that components having strong absorption in the ultraviolet region are measured with ultraviolet light, components having characteristic absorption in the near infrared region are measured with near infrared light, and both components are measured with high accuracy. Can be However, in many cases, components having strong absorption in the ultraviolet region also have some spectral change effect in the near-infrared region, while components having characteristic absorption in the near-infrared region also have some absorption in the ultraviolet region. Has an effect on the spectrum. Therefore, it is necessary to combine the data in the ultraviolet region and the data in the near-infrared region to correct the respective interferences and use the corrected data in the subsequent concentration determination.

An object of the present invention is to provide a measuring method for accurately determining a characteristic value from ultraviolet light absorbance data and near infrared light absorbance data.

[0005]

It is desirable that data measurement of near infrared rays and data measurement of ultraviolet rays can be performed by the same spectroscope. However, since there is no high-sensitivity, high-stable sensor that has sensitivity across both regions, and because there is no high-intensity, high-stable light source that has emission intensity across both regions, it is close to the ultraviolet region. The spectral process will be different in the infrared region. However, for stable measurement,
It is necessary to irradiate the same measurement point with both types of light beams. If this condition is not satisfied, the component concentration conditions will differ depending on the location of the measurement object, especially in liquids, etc., due to changes in scattering conditions due to bubbles, etc., correction of near-infrared data based on ultraviolet data, The reverse correction operation does not work. Therefore, in the spectroscopic measurement method according to the present invention, the same measurement target point is irradiated with both types of light beams. In the spectroscopic measurement method using ultraviolet light and near-infrared light according to the present invention, in a spectrometer for measuring the transmission intensity or reflection intensity of a sample by an ultraviolet spectroscopic process and a near-infrared spectroscopic process, the absorbance data of a sample having a known characteristic value And multiple wavelengths of near infrared.
Next, the multi-wavelength ultraviolet absorbance data obtained by the measurement is converted so as not to be affected by apparatus fluctuations related to the ultraviolet spectroscopy, and the multi-wavelength near-infrared absorbance data obtained by the measurement is converted to near-infrared spectroscopy. Is converted so as not to be affected by device fluctuations related to. In this way, the effects of independent fluctuation factors for the near-infrared spectroscopy process and the ultraviolet spectroscopy process are removed from the data. Using the converted ultraviolet and near-infrared data as explanatory variables, a multiple regression equation for obtaining the characteristic value of the sample is obtained. The characteristic value is, for example, the concentration of the liquid. When obtaining the characteristic value of the unknown sample, the influence of the device fluctuation is removed from the measured data, and the data is substituted into a multiple regression equation (calibration curve). Preferably, the ultraviolet data and the near-infrared data converted so as not to be affected by apparatus fluctuations are further converted so that the variances of the ultraviolet data and the near-infrared data match. Then, the data of the ultraviolet ray and the near infrared ray whose variances are matched are used as explanatory variables of the multiple regression equation. Preferably, for the ultraviolet data and near-infrared data converted so as not to be affected by device fluctuations, the data is further converted so as not to be affected by a common variation factor in both the ultraviolet spectroscopy process and the near-infrared spectroscopy process. I do. Then, the ultraviolet and near-infrared data converted so as not to be affected by the common fluctuation factors are used as explanatory variables of the multiple regression equation. Preferably, the ultraviolet data and the near-infrared data converted so as to match the variance of the data, and further, the data so as not to be affected by a common variation factor in both the ultraviolet spectroscopy process and the near-infrared spectroscopy process. Convert. Then, the ultraviolet and near-infrared data converted so as not to be affected by the common fluctuation factors are used as explanatory variables of the multiple regression equation.

[0006]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A measuring method according to an embodiment of the present invention will be described below with reference to the accompanying drawings. In this measurement,
Ultraviolet and near-infrared light are used to determine the characteristic value (eg, concentration) of a liquid that absorbs both ultraviolet and near-infrared light.
Specific examples of the chemical solution measured using this method include a mixed solution of ammonia and hydrogen peroxide, a mixed solution of hydrochloric acid and hydrogen peroxide, a mixed solution of sulfuric acid and hydrogen peroxide, a mixed solution of hydrofluoric acid and hydrogen peroxide, There are a mixture of hydrofluoric acid and ozone, a mixture of hydrofluoric acid, nitric acid and acetic acid, and a mixture of phosphoric acid, nitric acid and acetic acid. Generally, it is a mixture containing molecules having a double bond of C = C, C = O, N = N.

[0007] The spectrometer used for the measurement includes an ultraviolet ray measuring section and a near infrared ray measuring section. Ultraviolet light and near-infrared light are transmitted through the same liquid sample introduced into the spectrometer cell common to both measurement units, and the transmission intensity is measured by the ultraviolet measurement unit and the near-infrared measurement unit. Therefore, the measurement is performed by irradiating the same measurement target point of the sample in the cell with the luminous flux of the ultraviolet ray and the near-infrared ray, so that the correction in the measurement is facilitated.

[0008] Since the measurement is performed by irradiating the same measurement target point of the cell with ultraviolet rays and near-infrared rays, factors affecting the measurement data are those relating to the ultraviolet spectroscopy process, those relating to the near infrared spectroscopy process, and those relating to the ultraviolet spectroscopy process. And those related to near-infrared spectroscopy. Therefore, instead of obtaining a conversion formula using the ultraviolet absorbance data and the near-infrared absorbance data as they are as explanatory variables, the ultraviolet absorbance data is converted into data excluding the variation specific to the ultraviolet spectroscopic means, and the near-infrared data is similarly calculated. The infrared absorbance data is also converted into data excluding the fluctuation amount peculiar to the near-infrared spectroscopic means. In this manner, for the independent portions in the near-infrared spectroscopy process and the ultraviolet spectroscopy process, the error fluctuation amount caused by the device is removed at the initial stage in both the near-infrared spectroscopy and the ultraviolet spectroscopy. Preferably, the variances of the converted ultraviolet data and near-infrared data are then matched. In addition, preferably, the influence of a variation factor relating to both the ultraviolet spectral process and the near infrared spectral process is excluded. In this manner, the error fluctuation amounts resulting from the common fluctuation factors are converted after removing the error fluctuation amounts relating to the independently fluctuating portions.

The multiple regression equation (calibration curve) used for the measurement is obtained as follows. First, absorbance data of a sample having a known characteristic value at a plurality of wavelengths of ultraviolet light and near infrared light is measured. Then, the multi-wavelength ultraviolet absorbance data obtained by the measurement is converted so as not to be affected by the device fluctuation related to the ultraviolet spectroscopy, and the multi-wavelength near-infrared absorbance data obtained by the measurement is related to the near-infrared spectroscopy. Conversion is performed so as not to be affected by device fluctuation. Regarding the data thus converted, preferably, the variances of the converted ultraviolet data and near-infrared data are matched. Preferably, the conversion is performed so as not to be affected by a variation factor common to both the ultraviolet spectral process and the near infrared spectral process. Finally, a multiple regression equation for determining the characteristic value of the sample (for example, the concentration of the liquid) is determined by using the above-described converted and / or dispersion-matched ultraviolet and near-infrared data as explanatory variables. Therefore, the multiple regression equation does not directly use the ultraviolet light absorbance data and the near infrared light absorbance data as explanatory variables. Before obtaining the multiple regression equation, the UV absorbance data is converted to data excluding the fluctuation amount peculiar to the ultraviolet spectroscopic means, and the near infrared absorbance data is also excluded from the fluctuation amount peculiar to the near infrared spectroscopic means. Data. Preferably, data variance and common fluctuation factors are also processed. In the measurement of a sample having an unknown characteristic value, the influence of apparatus fluctuation is removed from the measurement data. Next, multiple regression equation (calibration curve)
To obtain the characteristic value of the unknown sample.

As a conversion method for eliminating the above-mentioned error fluctuation of each of the above-mentioned ultraviolet absorbance data and near-infrared absorbance data, or a common error fluctuation between the ultraviolet absorbance data and the near-infrared absorbance data, a general method such as a principal component analysis method is used. Techniques are available. In the principal component analysis method, a principal component may be obtained in a direction in which the most effective data can be obtained, and the first to N-th principal components may be employed, and linear transformation may be performed in that direction. In such a case, it is assumed that the fluctuation amount peculiar to the spectroscopic unit mostly exists in the main components not adopted. Further, the present applicant may perform linear conversion using a method described in Japanese Patent Application Laid-Open No. 3-209149. Here, JP-A-3-209
No. 149 is used. In this method, the ultraviolet lamp variation data and the ultraviolet sensor variation data are regarded as vectors, and their directions are often known in advance, so that linear conversion is performed in a direction orthogonal to the directions. First, the output changes Δ ₁ , Δ ₂ ,... At a plurality of wavelengths with respect to the temperature fluctuation of the sample per unit are expressed as a vector T of a dimension equal to the number of measured wavelengths as follows. T = (Δ ₁ , Δ ₂ ,...) Similarly, the output change with respect to the sample scattering fluctuation per unit and the equipment temperature fluctuation is also expressed as vectors S and M. Next, a vector P that satisfies the following three equations, P · T = 0 P · S = 0 P · M = 0, is obtained. In this case, there are three independent solutions of P. Let them be P ₁ , P ₂ and P ₃ . P ₁ , P ₂ , and P ₃ are vectors forming a subspace orthogonal to the vectors T, S, and M. Next, the measured data obtained for the known sample is converted into a vector A = (A
_1, A _2, ...) and then, the data X ₁ which is not affected by the error variation by projecting the next operation in this subspace, X _2, X ₃
Convert to X ₁ = P ₁ · A X ₂ = P ₂ · A X ₃ = P ₃ · A X ₁ , X ₂ , and X ₃ project A to a subspace represented by P ₁ , P ₂ , and P ₃ , respectively. This data is not affected by the above-mentioned error fluctuation at all.

Hereinafter, three measurement examples will be described. (Measurement Example 1) The concentration in a mixed aqueous solution of ammonia and hydrogen peroxide is measured. This mixed solution is used for cleaning a silicon wafer in a semiconductor manufacturing line. As a characteristic of the liquid, ammonia evaporates as a gas, and hydrogen peroxide decomposes to water. Both change to be thin. Since this mixing ratio is related to the cleaning ability and damage to the silicon wafer, it is desired to accurately measure the concentrations of both on-line.

FIG. 1 shows the configuration of a chemical concentration measuring device.
The liquid from the cleaning tank 2 is sucked by the pump 4 and the spectroscope 6
To the cell 66. FIG. 2 shows the structure of the spectroscope 6.
The light source in the near infrared region is a halogen lamp 61, and the light source 62 in the ultraviolet region is a D2 (deuterium) lamp. The dichroic mirror 63 has a property of transmitting near infrared rays and reflecting ultraviolet rays. The light from the two light sources 61 and 62 is converted into one light beam by the dichroic mirror 63 and then transmitted through the interference filter 64. The interference filter disk 65 is formed by arranging a plurality of filters 64 having different transmission wavelengths, and the disk 65 rotates 20 times per second. The configuration of the filter 64 is eight wavelengths in the near infrared region, and specifically, 980, 1040, 1080, 1110, 1150, 1200,
1255,1300nm is used. In the ultraviolet region, 6
The wavelength is 220, 230, 260, 280, 300, 330 nm. The light that has passed through the interference filter 64 passes through the cell 66. The cell length is 10 mm. The light passing through the cell 66 is received by the sensor 67. The sensor 67 is a composite element (manufactured by Hamamatsu Photonics) of a silicon photodiode and a germanium photodiode. Light in the ultraviolet region is received by the former, and light in the near infrared region is received by the latter. Sensor 67
Converts the light intensity into a current and converts the intensity into a digital value by AD conversion.

In the measurement, water is put in the cell 66 as a reference sample, and its transmission intensity is measured first. The data W ₂₂₀ , W ₂₃₀ ,..., W ₁₃₀₀ of the above-mentioned 14 wavelengths are stored in the memory as the intensity value W _λ of each wavelength. Next, the sample to be measured is placed in the cell 66. The intensity value and S _lambda, determine the absorbance A _lambda by the following calculation. The resulting intensity _A λ divided into the ultraviolet region and the near-infrared region. The _A λ in the ultraviolet region marked U _λ, referred to as the _A λ of the near-infrared region N _{λ. A λ = -LOG 10 (S λ} / W λ) here, the variation ratio of each wavelength including the sensitivity variation of the intensity change and a silicon photodiode sensor deuterium lamp 62 in the ultraviolet region C _lambda, (C ₂₂₀ , C ₂₃₀ , C ₂₆₀ , C ₂₈₀ , C
₃₀₀ , C ₃₃₀ ), this example is as follows. This fluctuation can be caused by the temperature of the equipment installed as a disturbance factor.To obtain this data, put the equipment in a constant temperature bath and change the set temperature. Become. (1.2, 1.1, 1.05, 1.04, 1.04, 1.00) When measuring a mixture of ammonia and hydrogen peroxide, the absorption of hydrogen peroxide is too strong at wavelengths shorter than 270 nm.
The three wavelengths of 0, 230, and 260 nm are measured but not used in the following calculations. Only the variation ratio C _λ at the remaining three wavelengths of 280, 300, and 330 nm is written as follows. (1.04, 1.04, 1.00)

Next, a plurality of ultraviolet absorbance data are converted so as not to be affected by apparatus fluctuations related to ultraviolet spectroscopy. In this conversion, the present applicant has disclosed in
The method described in Japanese Patent Publication No. 49 is used. The purpose of measuring the ultraviolet region is mainly to measure the concentration of hydrogen peroxide.
One parameter is obtained from the above three absorbances so as to obtain the maximum change in the absorbance and remove fluctuations in the apparatus. If the rate of change in absorbance per unit concentration of hydrogen peroxide is expressed in parenthesis as in the above rate of change, the following is obtained. (0.30, 0.15, 0.01) The data expressed by these two parentheses is converted into a three-dimensional vector A,
B can be considered. A = (0.30, 0.15, 0.01) B = (1.04, 1.04, 1.00) Data obtained by conversion orthogonal to the B vector and projected in a direction as close as possible to the direction of the A vector is a parameter to be obtained. If X = (x ₁ , x ₂ , x ₃ ), it can be obtained from the following equation. From FIG. 3, X
= A−qB, and q that satisfies X · B = 0 may be obtained. In summary, X = A-((AB) / (BB)) B. By substituting a specific value, X = (0.1428, -0.007, -0.1411), ie, K = 0.1428U ₂₈₀ -0.007U ₃₀₀ -0.1411U ₃₃₀ , can be converted so as not to be affected by the change of the B vector. .

Regarding the data in the near infrared region, the fluctuation ratio C _λ of each wavelength including the halogen / tungsten lamp intensity fluctuation in the near infrared region and the sensitivity fluctuation of the germanium photodiode sensor is _expressed by (C ₉₈₀ , C ₁₀₄₀ , C _1080). ,
C ₁₁₁₀ , C ₁₁₅₀ , C ₁₂₀₀ , C ₁₂₅₅ , C ₁₃₀₀ ), the following was obtained in this example. This fluctuation can be caused by the temperature of the equipment installed as a disturbance factor.To obtain this data, put the equipment in a constant temperature bath and change the set temperature. Become. (1.00, 1.04, 1.01, 1.03, 1.01, 1.02, 1.04, 1.03) The spectrum in the near infrared region is greatly affected by the temperature fluctuation of the aqueous solution. become. (0.0053, -0.0020, -0.0009, 0.0007, 0.0171, -0.003
0, -0.0090, -0.0010)

Next, a plurality of near-infrared absorbance data are converted so as not to be affected by apparatus fluctuations related to near-infrared spectroscopy. In this conversion, as in the case of ultraviolet spectroscopy, the present applicant uses the method described in JP-A-3-209149. The reason for measuring the near infrared region is mainly to measure the concentration of ammonia. One parameter is obtained from the eight absorbances so that the change in absorbance due to the change is obtained to the maximum and the fluctuation in the apparatus and the fluctuation in the temperature of the sample can be removed. If the rate of change in absorbance per 1 wt% of ammonia concentration is expressed in parenthesis as in the above rate of change, the following is obtained. (-0.0037, 0.0088, 0.0010, -0.0003, -0.0065, -0.001
(8, 0.0003, 0.0047) The data expressed in the three parentheses can be regarded as an 8-dimensional vector.

D = (-0.0037, 0.0088, 0.0010, -0.0003,
-0.0065, -0.0018, 0.0003,0.0047) E = (1.00, 1.04, 1.01, 1.03, 1.01, 1.02, 1.04, 1.0
3) F = (0.0053, -0.0020, -0.0009, 0.0007, 0.0171, -0.
0030, -0.0090, -0.0010), it is orthogonal to the E and F vectors and D
Data obtained by a conversion projecting in a direction as close as possible to the direction of the vector is a parameter to be obtained. If it is expressed as Y = (y ₁ , y ₂ , y ₃ , y ₄ , y ₅ , y ₆ , y ₇ , y ₈ ), it can be obtained as follows.

From FIG. 4, a vector obtained by a linear combination of the E vector and the F vector is defined as a G vector. The G vector is represented by the following equation. G = sE + tF Y = DG, s and t such that Y · G = 0
Should be obtained. These parameters s and t are Y · G = 0.
In this case, the length can be easily obtained from the condition that the length of the vector Y is minimized. The square value L of the length of Y is represented by the following equation. Assuming that the components of the D, E, and F vectors are d _I , e _I , and f _I , L = Σ (d _I −se _I −tf _I ) ² This L value is given by ∂L / ∂s = 0 0L / Δt = 0 is satisfied. When expanded, the following equation is obtained. ^{_{sΣe I 2 + tΣe I f I}} = Σe I d I sΣe I f I + tΣf I 2 = Σf I d I _{Therefore, s = Σe I d I /} (Σe I 2 Σf I 2 - (Σe I f I) 2) t = Σf _I d _I / (Σe _I ² Σf _I ^2- (Σe _I f _I ) ² ). Using these s and t, a Y vector is obtained from Y = DG = D− (sE + tF). When applied to actual values, s =
0.000654, t = -0.37416, Y = (-0.00237, 0.00737, 0.00000, -0.00071, -0.0007
6, -0.00359, -0.00375, O.00365). That _{J = -0.00237N 980 + 0.00737N 1040 -0.00000N} 1080 -
0.00071N ₁₁₁₀ -0.00076N ₁₁₅₀ -0.00359N ₁₂₀₀ -0.003
If converted to 75N ₁₂₅₅ + 0.00365N ₁₃₀₀ , conversion can be performed without being affected by changes in the E vector and F vector.

Next, the data K of the ultraviolet ray and the data J of the near infrared ray which have been converted so as not to be affected by the fluctuation of the apparatus are made to have the same variance, and the concentrations of both ammonia and hydrogen peroxide are accurately obtained. . Here, the S / N of the data of J and K is matched. That is, if necessary, S /
A weighting factor is applied so that N is close. Here _{M = K / S U N =} J / S N, S U is the standard deviation of the error in the ultraviolet spectral portion that has been determined in advance experimentally, S _N is near that had been determined in advance experimentally This is the standard deviation of the error of the infrared spectroscopy unit. FIG. 5 is a graph in which N is set on the X axis and M is set on the Y axis, and data of various mixed liquids of ammonia and hydrogen peroxide are plotted. Lines are drawn for the same ammonia concentration. FIG. 7 shows N on the X axis and M on the Y axis.
It is the graph which plotted the data of various mixed liquids of ammonia and hydrogen peroxide, and draws a line about the same hydrogen peroxide concentration.

It is preferable that the data K and J are not affected by a variation factor common to both the ultraviolet spectroscopy process and the near-infrared spectroscopy process using, for example, a method described in Japanese Patent Application Laid-Open No. 3-209149. Convert to The portion that covers both the near-infrared spectroscopy process and the ultraviolet spectroscopy process is processed after removing the amount of error variation for each independent portion. A typical example of a portion that spans both is around an object to be measured, and becomes a cell when the sample is a liquid. Fluctuation factors common to both the ultraviolet spectroscopy process and the near-infrared spectroscopy process include sample temperature fluctuation, cell window plate contamination, cell length fluctuation, and sample scattering fluctuation when transmitted through the same cell. If the liquid is not a liquid, the light scattering conditions depending on the surface state of the measurement point are the same. Since it is necessary to use both near-infrared data and ultraviolet data efficiently, their S / N
Are preferably matched. In this way, the variation over both the near-infrared data and the ultraviolet data is removed. In the data after the removal in this manner, there is sufficient information to be used for the quantitative calculation, such as the concentration of the components, and the interference information to the quantitative calculation is efficiently removed.

Next, a multiple regression equation for obtaining a characteristic value (for example, density data) of the sample that is desired to be obtained is obtained using the data converted so as to remove the fluctuation factor as an explanatory variable. FIG.
The results of fitting the data of FIG. 5 to a three-dimensional curved surface by the least squares method with the ammonia concentration on the Z axis are shown. The surface equation was expressed by the following equation. Ammonia concentration = 86.983N-0.58731M-98.6907N ² +
^{0.208974M 2 + 15.821MN + 606.98M 3 -0.0531N} 3 -88.
958M ² N-0.4704MN ²

FIG. 8 shows the result of fitting the data of FIG. 7 to a three-dimensional curved surface by the least squares method with the hydrogen peroxide concentration on the Z axis. The surface equation of the data was expressed by the following equation. Hydrogen peroxide concentration = -109.32N + 14.743M + 2406.97N ² -
^{3.3155M 2 -205.873MN-12427.1M 3 + 0.41837N} 3 +94
6.616M ² N + 19.5979 MN ² These two curved surfaces are the calibration curve formula, measure the mixed solution of ammonia and hydrogen peroxide of unknown concentration, find the N value and M value, and substitute them into this formula. For example, the concentration of ammonia and the concentration of hydrogen peroxide can be accurately obtained. As described above, in Measurement Example 1, one conversion data K is obtained from a plurality of ultraviolet data, and one conversion data J is obtained from a plurality of near-infrared data. Using the two data of K and J as explanatory variables, conversion formulas of the ammonia concentration and the hydrogen peroxide concentration were respectively obtained by a least squares regression calculation.

(Measurement Example 2) The ultraviolet absorbance and the near infrared absorbance are measured in the same manner as in Measurement Example 1. Next, a plurality of ultraviolet absorbance data are converted using, for example, a method described in Japanese Patent Application Laid-Open No. 3-209149 so as not to be affected by apparatus fluctuations related to ultraviolet spectroscopy. That is, the absorbance at three wavelengths in the ultraviolet region is determined by the variation ratio of the ultraviolet spectrometer.
Projective transformation to a space orthogonal to the direction of (1.04, 1.04, 1.00). In this case, this space becomes a two-dimensional plane. In Measurement Example 1, the three light absorbances were projected and converted into one axis. In this case, the condition that the amount of change of hydrogen peroxide is maximized is added. In reality, the change amount of hydrogen peroxide changes its direction little by little for each concentration. Therefore, the direction of one vector is represented by the direction of one vector when the change amount of the concentration is small, Some lack of occurs. Thus, in Measurement Example 2, the three absorbances are not immediately converted to one data, but are converted to two data.

Assuming that base vectors in a two-dimensional space orthogonal to B = (1.04, 1.04, 1.00) are X ₁ and X ₂ , BX ₁ = 0 BX ₂ = 0 X ₁ X ₂ = 0 X ₁ and X ₂ satisfying the following condition may be obtained. Since there are six variables and three expressions, it is not decided uniquely, but one X
_{If 1} or X ₂ is obtained, the later solution is expressed by a linear combination equation. The following are the values determined. X ₁ = (-0.3976, -0.3976, 0.8270) X ₂ = (-0.3976, 0.3976, 0.0000) K ₁ = -0.3976U ₂₈₀ -0.3976U ₃₀₀ + 0.8270U ₃₃₀ K ₂ = -0.3976U ₂₈₀ + 0.3976U ₃₀₀ With the conversion, the data can be converted into data that is not affected by the change in the B vector.

Similarly, a plurality of near-infrared absorbance data are
Conversion is performed so as not to be affected by device fluctuations related to near-infrared spectroscopy. That is, the data in the near-infrared region is converted into six data Y _i (i = 1 to 6) obtained by removing two variations of the sample temperature variation and the device temperature variation of the near-infrared spectrometer from the absorbance at eight wavelengths. I do. The device temperature change rate is E = (1.00, 1.04, 1.01, 1.03, 1.01, 1.02, 1.04, 1.0
3), and the temperature change rate of the sample is F = (0.0053, -0.0020, -0.0009, 0.0007, 0.0171, -0.
0030, -0.0090, -0.0010), and Y _i (i = 1 to 6) satisfying the following expression is obtained. This is also not uniquely determined, but if one Y _i is obtained, the subsequent solution can be expressed by a linear combination thereof. E · Y _i = 0 F · Y _i = 0 Y _i · Y _j = 0 Y _i · Y _i = 1 (where i = 1 to 6, j = 1 to 6, i ≠
j)

Y _i (i = 1 to 6) obtained from actual values are described below. Y ₁ = (-0.1022, -0.6481, 0.0000, 0.0000, 0.0000, 0.0
000, 0.0000, 0.7546) Y ₂ = (-0.0737, -0.3986, 0.8435, 0.0000, 0.0000, 0.0
000, 0.0000, -0.3524) Y ₃ = (-0.2710, -0.2053, -0.2097, 0.8917, 0.0000, 0.0
_{000, 0.0000, -0.2130) Y 4} = (0.8544, -0.3235, -0.1460, 0.1122, -0.3244, 0.0
000, 0.0000, -0.1621) Y 5 = (- 0.0999, -0.2529, -0.2247, -0.1965, 0.1375, 0.8
745, 0.0000, -0.2308) Y 6 = (- 0.0228, -0.2479, -0.2093, -0.1648, 0.3323, -0.2
743, 0.7967, -0.2160) _{therefore, J 1 = -0.1022N 980 -0.6481N 1040} -0.0000N 1080 -0.
0000N ₁₁₁₀ -0.0000N ₁₁₅₀ -0.0000N ₁₂₀₀ -0.0000N
_{1255 + 0.7546N 1300 J 2 = -0.0737N} 980 -0.3986N 1040 + 0.8435N 1080 -0.
0000N ₁₁₁₀ -0.0000N ₁₁₅₀ -0.0000N ₁₂₀₀ -0.0000N
_{l255 -0.3524N 1300 J 3 = -0.2710N 980} -0.2053N 1040 -0.2097N 1080 +0.
8917N ₁₁₁₀ -0.0000N ₁₁₅₀ -0.0000N ₁₂₀₀ -0.0000N
_{1255 -0.2130N 1300 J 4 = 0.8544N 980} -0.3235N 1040 -0.1460N 1080 +0.
1122N ₁₁₁₀ -0.3244N ₁₁₅₀ -0.0000N ₁₂₀₀ -0.0000N
_{1255 -0.1621N 1300 J 5 = -0.0999N 980} -0.2529N 1040 -0.2247N 1080 -0.
1965N ₁₁₁₀ + 0.1375N ₁₁₅₀ + 0.8745N ₁₂₀₀ -0.0000N
_{1255 -0.2308N 1300 J 6 = -0.0228N 980} -0.2479N 1040 -0.2093N 1080 -0.
1648N ₁₁₁₀ + 0.3323N ₁₁₅₀ -0.2743N ₁₂₀₀ + 0.7967N
₁₂₅₅ -0.2160N ₁₃₀₀

With respect to K ₁ , K ₂ and J _{1 to} J ₆ obtained above, the S / N of the data of K and J is made to match. M _i = K _i / S _U (where i = 1 to 2) N _i = J _i / S _N (where i = 1 to 6) Here, S _U was obtained in advance by experiments. S _N is the standard deviation of the error of the ultraviolet spectroscopic unit, and S _N is the standard deviation of the error of the near-infrared spectroscopic unit obtained in advance by experiments.

Using the M _i and N _i obtained as described above, the following equation for converting the concentration of ammonia and hydrogen peroxide is obtained by a least squares regression calculation. Ammonia concentration = Σp _i M _i + Σq _j N _j Hydrogen peroxide concentration = Σr _i M _i + Σs _j N _j (where i = 1 to 2, j = 1 to 6) p _i , q _j , r _i , s _j Is a parameter of a concentration conversion formula, which is obtained by a least squares regression calculation. Thus, M _i
Using N _i and, the ammonia concentration and hydrogen peroxide concentration determined.

Here, the eight data of M _i and N _i are used as they are in the density conversion formula, but if there is an error correlation between these data, they are converted again at this stage.
At this stage, for example, the error of the common optical element of the ultraviolet spectroscopy unit and the near-infrared spectroscopy unit is eliminated. In particular,
A luminous flux of a deuterium lamp 62, which is a light source of ultraviolet light in FIG. 2,
Halogen tungsten lamp 61 as a near infrared light source
Of the dichroic mirror 63 that makes the same light beam the same light beam, or a change in characteristics of the cell 66 through which ultraviolet light and near-infrared light pass as the same light beam (cell length, stain on the cell window plate). Here, an example is shown in which dirt on the window plate of the cell 66 has been removed. (M ₁ , M ₂ , N ₁ , N ₂ , N ₃ , N ₄ , N ₅ , N ₆ )
V = (1.22, 1.31, 1.01, 1.00, 1.15, 1.01, 1.00, 1.0
1), a base vector Z in a space orthogonal to this direction is obtained. V · Z _i = 0 Z _i · Z _j = 0 Z _i · Z _i = 1 (where i = 1 to 7, j = 1 to 7, i
≠ j)

Z _i (i = 1 to 7) obtained from the actual values are described below. Z ₁ = (-0.6377, 0.0000, 0.0000, 0.0000, 0.0000, 0.0
_{000, 0.0000, 0.7703) Z 2} = (0.4909, -0.7706, 0.0000, 0.0000, 0.0000, 0.0
000, 0.0000, 0.4064) Z 3 = (- 0.2618, -0.2811, 0.8975, 0.0000, 0.0000, 0.0
000, 0.0000, -0.2167) Z 4 = (- 0.2132, -0.2289, -0.1765, 0.9164, 0.0000, 0.0
000, 0.0000, -0.1765) Z 5 = (- 0.2041, -0.2192, -0.1690, -0.1673, 0.9084, 0.0
000, 0.0000, -0.1690) Z 6 = (- 0.1529, -0.1641, -0.1266, -0.1253, -0.1441, 0.9
387, 0.0000, -0.1266) Z 7 = (- 0.1345, -0.1444, -0.1113, -0.1102, -0.1267, -0.1
113, 0.9464, -0.1113) _{therefore, U 1 = -0.6377M 1 -0.0000M 2} -0.0000N 1 -0.0000N 2
−0.0000N ₃ −0.0000N ₄ −0.0000N ₅ + 0.7703N ₆ U ₂ = + 0.4909M ₁ −0.7706M ₂ −0.0000N ₁ −0.0000N ₂
−0.0000N ₃ −0.0000N ₄ −0.0000N ₅ + 0.4064N ₆ U ₃ = −0.2618M ₁ −0.2811M ₂ + 0.8975N ₁ + 0.0000N ₂
−0.0000N ₃ −0.0000N ₄ −0.0000N ₅ −0.2167N ₆ U ₄ = −0.2132M ₁ −0.2289M ₂ −0.1765N ₁ + 0.9164N ₂
−0.0000N ₃ −0.0000N ₄ −0.0000N ₅ −0.1765N ₆ U ₅ = −0.2041M ₁ −0.2192M ₂ −0.1690N ₁ −0.1673N <sub>2
+ 0.9084N 3 -0.0000N 4 -0.0000N 5 -0.1690N</sub> 6 U 6 = -0.1529M 1 -0.1641M 2 -0.1266N 1 -0.1253N 2
−0.1441N ₃ + 0.9387N ₄ −0.0000N ₅ −0.1266N ₆ U ₇ = −0.1345M ₁ −0.1444M ₂ −0.1113N ₁ −0.1102N ₂
−0.1267N ₃ −0.1113N ₄ + 0.9464N ₅ −0.1113N ₆ U _i is obtained by the above equation, and the density conversion equation is as follows. Here, the parameters p _i and r _i are obtained by a least squares operation expression. Ammonia concentration = Σp _i U _i hydrogen peroxide concentration = Σr _i U _i (here i = 1 to 7)

(Measurement Example 3) A measurement example in which a mixed aqueous solution of hydrofluoric acid and ozone was measured by this method is shown. This mixture
In semiconductor manufacturing lines, it is used for cleaning and etching of silicon wafers. As a characteristic of the liquid, since hydrofluoric acid dissolves silicon, its concentration gradually decreases, and ozone is unstable and decomposes in tens of minutes. This mixing ratio is closely related to the throughput of silicon wafers, and it is desired to accurately measure the concentrations of both on-line. The configuration of the device is almost the same as that of FIG. The liquid from the cleaning tank 2 is sucked by the pump 4 and the liquid
Introduce to 6. The material of the cell 66 is sapphire.
The temperature is adjusted to 5 ° C. Cell 6 to prevent condensation
The area around 6 is dehumidified with a desiccant. The structure of the spectroscope is the same as in FIG. The wavelength used is the same as that in the measurement example 1.

FIGS. 9 and 10 show the ultraviolet spectra of ozone and hydrofluoric acid, respectively. FIG. 11 shows an ultraviolet spectrum of pure water. Ozone has absorption with a peak near 260 nm. Hydrofluoric acid has a slightly reduced absorption below 230 nm compared to water. FIG. 12 shows a near-infrared spectrum of hydrofluoric acid. Since it is based on water, it is a difference spectrum with water. Ozone has no noticeable change in the near infrared spectrum.

Water is placed in a cell as a reference sample, and its transmission intensity is measured first. As the intensity value W _λ at each wavelength, 14 data W ₂₂₀ , W ₂₃₀ , W ₂₆₀
ＷW ₁₃₀₀ is stored in the memory. Next, a sample to be measured is placed in the cell, and measurement is performed, and the intensity value is defined as _Sλ . Next, the absorbance _Aλ is determined. _{A λ = -LOG 10 (S λ} / W λ) obtained divide the _A λ in the ultraviolet region and the near-infrared region. The A _lambda in the ultraviolet region represented as U _lambda, expressed as the A _lambda in the near-infrared region N _lambda.

Next, a plurality of ultraviolet absorbance data are converted so as not to be affected by apparatus fluctuations related to ultraviolet spectroscopy. Fluctuation ratio C of each wavelength including deuterium lamp intensity fluctuation in ultraviolet region and sensitivity fluctuation of silicon photodiode sensor
_{If λ} is described as (C ₂₂₀ , C ₂₃₀ , C ₂₆₀ , C ₂₈₀ , C ₃₀₀ , C ₃₃₀ ), the following is obtained in this example. This fluctuation can be caused by the temperature of the equipment installed as a disturbance factor.To obtain this data, put the equipment in a constant temperature bath and change the set temperature. Become. (1.2, 1.1, 1.05, 1.04, 1.04, 1.00) The measurement in the ultraviolet region is mainly for measuring the concentration of ozone. 230 nm, 260 nm, and 300 nm are adopted as the ozone absorption peak and the wavelengths at both ends thereof. 220, 280,330nm
The three wavelengths are measured but not used in the following calculations. Write out only the remaining variation ratio C _I as follows: (1.1, 1.05, 1.04) One parameter is obtained from the three absorbances so that the change in absorbance due to ozone can be obtained to the maximum and the fluctuation of the apparatus can be removed. If the rate of change in absorbance per unit concentration of ozone is expressed in parenthesis as in the above rate of change, the following is obtained. (0.12, 0.30, 0.01) The data represented by these two parentheses can be regarded as a three-dimensional vector. A = (0.12, 0.30, 0.01) B = (1.1, 1.05, 1.04) If B = (1.1, 1.05, 1.04), it is a parameter required to obtain data obtained by a transformation orthogonal to the B vector and projected in a direction as close as possible to the direction of the A vector. is there. X = (x ₁ , x ₂ ,
x ₃ ), it can be obtained from the following equation. From FIG. 3, by expressing X = A−qB, q such that X · B = 0 is obtained.
Should be obtained. In summary, X = A-((AB) / (BB)) B. Substituting specific values, X = (- 0.02824, 0.15850 , -0.1302) i.e. if converted _{K = -0.02824U 220 + 0.15850U 280 -0.1302U} 330, so as not to be affected by changes in B vector conversion it can.

Next, a plurality of near-infrared absorbance data are converted so as not to be affected by apparatus fluctuations related to near-infrared spectroscopy. Regarding the data in the near infrared region, the spectrum in the near infrared region is mainly for measuring the concentration of hydrofluoric acid.
One parameter is obtained from the eight absorbances so that the change in absorbance due to the change is obtained to the maximum and the fluctuation in the apparatus and the fluctuation in the temperature of the sample can be removed. If the rate of change in absorbance per 1 wt% of hydrofluoric acid is expressed in parentheses as in Measurement Example 1, the results are as follows. (-0.00065, 0.00055, -0.00175, -0.00027, 0.00175,
(0.00114, 0.00062, 0.00033) The subsequent calculations are the same as in Measurement Example 1. _{J = -0.30577N 980 -0.40249N 1040 -0.24703N 1080} -
0.08442N ₁₁₁₀ + 0.39903N ₁₁₅₀ + 0.57953N ₁₂₀₀ -0.290
If converted 21N ₁₂₅₅ -0.31154N _1300, can be converted so as not to be affected by temperature fluctuations in the device change and the sample.

Next, using the two values of J and K, the concentrations of both ozone and hydrofluoric acid are accurately measured. The S / N of the data of J and K should be matched before regression calculation. Here _{M = K / S U N =} J / S N, S U is the standard deviation of the error in the ultraviolet spectral portion that has been determined in advance experimentally, S _N is near that had been determined in advance experimentally This is the standard deviation of the error of the infrared spectroscopy unit.

Next, using the converted data as an explanatory variable, a multiple regression equation for obtaining a characteristic value desired for the sample, for example, density data is obtained. The phase equations corresponding to FIGS. 6 and 8 were expressed by the following equations. Ozone concentration = 0.2651N + 3.6389M-34.3107N ² +0.1011
M ² -1.5671MN Hydrofluoric acid concentration = 99.101N + 0.2341M + 12.4511N ² -2.7895
M ² −1.3474MN These two curved surfaces are of a calibration curve type, and a mixture of ozone and hydrofluoric acid of unknown concentration is measured to obtain N and M values.
By substituting into this equation, the concentration of ozone and the concentration of hydrofluoric acid can be accurately obtained.

In the above embodiment, the measurement of the transmission intensity has been described. However, measurement and data processing may be similarly performed for the measurement of the reflection intensity. Note that Japanese Patent Laid-Open No. 4-2
The concentration measuring method described in Japanese Patent No. 49748 is also the same as the above-described measuring method in that the concentration of a mixed solution of hydrogen peroxide and ammonia is measured. However, the ultraviolet ray measuring section and the infrared ray measuring section are independent of each other, and the measuring cells are provided respectively. This measurement method has the following problems. (1) Since the obtained concentration conversion formula is divided by the ammonia concentration value α and the ultraviolet absorbance value β as parameters, the value obtained at the boundary may have a step or may not change smoothly. (2) Since the density conversion equation changes according to the parameters α and β, the data processing algorithm becomes complicated. (3) Since the components that make up the device, such as the light source and the sensor, are different between the near-infrared spectroscopy unit and the ultraviolet spectroscopy unit, it is necessary to treat near-infrared absorbance and ultraviolet absorbance as the same data (explanatory variables). There is impossible. Specifically, the S / N of the data, the intensity, the variation due to the device, and the variation due to the sample are different. On the other hand, in the present invention, measurement is performed using ultraviolet light and near infrared light using a common cell. Further, the above problem does not occur.

[0039]

According to the present invention, the transmission intensity of a sample is measured at the same measurement target point of the cell by an ultraviolet spectroscopic process and a near-infrared spectroscopic process. And preferably further remove from the data the effects of variables common to both the ultraviolet and near-infrared spectroscopy processes. Thereby, the characteristic value (for example, the concentration of the liquid) of the sample can be accurately obtained from the ultraviolet light absorbance data and the near infrared light absorbance data.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration of a chemical solution concentration measuring device.

FIG. 2 is a diagram schematically showing a structure of a spectroscope in a chemical solution concentration measuring device.

FIG. 3 is a diagram showing a relationship among A, B, and X;

FIG. 4 is a diagram showing a relationship among D, E, F, and G;

FIG. 5 is a scatter plot in which data of various mixed solutions of ammonia and hydrogen peroxide are plotted with N on the X axis and M on the Y axis.

FIG. 6 is a curved surface diagram of ammonia data.

FIG. 7 is a scatter plot in which data of various mixed solutions of ammonia and hydrogen peroxide are plotted with N on the X axis and M on the Y axis.

FIG. 8 is a curved surface diagram of hydrogen peroxide data.

FIG. 9 UV spectrum of ozone

FIG. 10 Ultraviolet spectrum of hydrofluoric acid (HF 30%)

FIG. 11 Ultraviolet spectrum of pure water

FIG. 12: Near infrared spectrum of hydrofluoric acid

[Explanation of symbols]

2 Spectrometer 61 Halogen lamp, 62 Deuterium lamp, 63 Dichroic mirror, 66
Cell, 67 sensors.

Claims (5)

[Claims] 1. A spectrometer for measuring the transmission intensity or reflection intensity of a sample by an ultraviolet spectroscopic process and a near-infrared spectroscopic process, wherein the absorbance data of the sample having a known characteristic value is measured at a plurality of wavelengths of ultraviolet and near-infrared. UV absorbance data of multiple wavelengths obtained by
Converted so as not to be affected by device fluctuations related to ultraviolet spectroscopy, and converted near-infrared absorbance data of multiple wavelengths obtained by measurement so as not to be affected by device fluctuations related to near-infrared spectroscopy. A spectral measurement method using ultraviolet light and near infrared light, wherein a multiple regression equation for obtaining a characteristic value of a sample is obtained using near infrared data as an explanatory variable. 2. The data of the ultraviolet rays and the data of near-infrared rays which are converted so as not to be affected by device fluctuations.
2. The method according to claim 1, further comprising: converting the variance of the ultraviolet data and the near-infrared data so as to be identical, and using the ultraviolet and near-infrared data whose variances are identical as explanatory variables of the multiple regression equation. Spectrometry method. 3. The data of the ultraviolet ray and the data of near-infrared ray which are converted so as not to be affected by apparatus fluctuation,
Furthermore, the data is converted so that it is not affected by the common fluctuation factors in both the ultraviolet spectroscopy process and the near-infrared spectroscopy process, and the converted ultraviolet and near-infrared data are converted so as not to be affected by the common fluctuation factors. The method according to claim 1, wherein the method is an explanatory variable. 4. The ultraviolet data and the near-infrared data, which are converted so as to match the variance of the data,
Furthermore, the data is converted so that it is not affected by the common fluctuation factors in both the ultraviolet spectroscopy process and the near-infrared spectroscopy process, and the converted ultraviolet and near-infrared data are converted so as not to be affected by the common fluctuation factors. 3. The method according to claim 2, wherein the method is an explanatory variable. 5. The spectrometry method according to claim 1, wherein the characteristic value is a concentration of a liquid.