JPH0414298B2 - - Google Patents

Description

[Detailed Description of the Invention] [Technical Field of the Invention] The present invention relates to a spectrum analyzer that quantifies a specific component using the resonance or emission spectrum of a sample containing a mixture of gases or liquids. This invention relates to a spectral data processing method for removing the influence of the interference spectrum of mixed substances superimposed on the spectrum of

[Technology background]

Analytical techniques for determining the concentration of components in a sample are highly advanced, such as absorption spectrum analysis using gas, liquid, or solid spectrophotometers, and emission spectrum analysis using thermal excitation, discharge excitation, laser excitation, etc. of metals, metal oxides, etc. Because of its reliability, it is widely used in the chemical industry to measure impurities or composition in products and as a sensor for plant control equipment. It is considered to be a useful method due to its high height and short measurement time.

Figure 1 is a diagram showing the principle of an analysis device that uses an absorption spectrum. In the figure, 1 is a light source such as a wavelength tunable laser, 2 is a sample cell, 3 is a photoelectric converter, and P _S is a light source 1.
, L is the optical path length passing through the sample cell 2 , and P _D is the power of the light incident on the photoelectric converter 3 .

In this case, the following relationship exists between P _D and P _S.

P _D (ν) = P _S e ^- 〓 ⁽ 〓 ⁾ ...(1) Here, ν is the wave number, and τ (ν) is called the absolute absorption amount, which is given by the following equation according to the Beer-Lambert law. It will be done.

τ (ν) = CLα (ν) + β (ν) ... (2) Here, C is the concentration of the target substance to be determined, L is the optical path length, and α (ν) is the absorption coefficient spectrum of the target substance, which is unique to the substance. The spectrum, β(ν), is the interference spectrum.

In addition, when quantitatively measuring the concentration of a trace amount of a component, the signal level of the detected absorption spectrum is also small, making measurement difficult. Therefore, the spectrum waveform is usually differentiated using derivative spectroscopy to remove minute irregularities. This is done based on the emphasized higher-order spectra. The spectrum in this case is represented by S(v).

By the way, for example, when measuring the absorption spectrum of the atmosphere containing a trace amount of the target substance and calculating the concentration of the target substance from it, the reference of the target substance is included in the sample spectrum S _X (ν) obtained by measurement. Not only the spectrum S _R (ν), but also the interference spectra S _I1 of K substances whose sizes are unknown but whose shapes are known.
(ν), S _I2 (ν), ..., S _IK (ν), an interference spectrum S _U (ν) whose size and shape are unknown, and random noise n (Caussian distribution with zero mean value) are mixed. It is thought that there are. Here, in the case of the spectrum analyzer having the configuration shown in FIG. 1, S _X (ν) is expressed by the following equation.

_{S X (} ^ν _{) = C} It is. Note that C _X represents the concentration of the target substance in the sample, and C _K represents the concentration of the substance with known interference spectrum S _IK (ν) in the sample.

FIG. 2 conceptually shows an example of a mixture of such various spectra. In the figure, the shaded A represents spectrum information of the target substance, B represents the interference spectrum, and T represents the spectral region. The target substances of spectrum A are low molecular weight substances such as methane CH ₄ , NO _x , SO _x , whereas interference spectrum B is mainly water H ₂ O. The concentration of CH ₄ in the atmosphere is usually
The concentration of H 2 O is about 1 ppm, whereas the concentration of H ₂ O is an order of magnitude higher.

In Figure 2, there is only one type of interference spectrum to simplify the explanation, but in reality there is only one type of interference spectrum.
As shown in equation (3), multiple types of interference spectra and random noise are superimposed. The interference spectrum includes not only the absorption spectrum of existing gases but also flat spectra caused by fogging, dust, etc. on lenses and windows. In this situation,
When the laser frequency ν is fixed, absorption by the target gas and absorption by a different gas cannot be distinguished.

[Conventional technology and problems]

Next, some typical conventional techniques will be explained.

(a) Spectrophotometric type This method is the method explained in Figure 1,
Using a light source such as a white light source and a spectrometer or a wavelength-tunable semiconductor laser, the frequency (wavelength) of the freely adjustable light is adjusted and fixed to a value specific to the target gas or liquid. The purpose is to find out the concentration of ingredients. In this method, as described above, if a substance other than the target substance absorbs at the same frequency, it becomes an interference spectrum and causes disturbance, which affects the measured value. Therefore, high sensitivity cannot be obtained.

(b) Ultraviolet region correlation spectrophotometer This method attempts to measure component concentrations from the ultraviolet absorption spectrum of gases. The spectrometer slit of the above spectrophotometer is provided with a correlated slit array having two types of patterns corresponding to the spectral arrangement of the target gas species, and by using these slits alternately, it is possible to have high sensitivity only to the concentration of a specific species of gas. This eliminates the influence of interference spectra of different gases. However, since the pattern of this correlation slit array is fixed,
It lacks flexibility for different gas types, and it is also difficult to provide a smoothing function to suppress noise (smoothing, a type of digital filter processing).

(c) Non-dispersive infrared gas analyzer This method uses white infrared light to measure the absorption by the target gas by utilizing the thermal expansion of the same type of gas as the target gas. It is. This method is limited to cases where the level of the interference spectrum is constant within the spectral region to be used, and in principle only direct absorption methods can be applied, and higher-order derivative spectroscopy with a good signal-to-noise ratio cannot be used. . Also, because mechanical methods are used, accuracy is low, with sensitivity often 10 ppm.
That's about it.

[Purpose of the invention]

An object of the present invention is to remove the influence of an interference spectrum caused by a known substance other than the target substance from a spectrum measured for a sample in which a plurality of substances including the target substance are mixed in a spectrum analyzer, The object of the present invention is to provide a means that makes it possible to determine the concentration of a target substance with high sensitivity.

In other words, in component concentration analysis of mixed substances,
The signal information directly detected by the sensor is the spectrum S of a function such as the wave number ν of light or the sweep control voltage.
(ν), which is the component concentration C _n of M mixed substances, m=1, 2,..., M: S(ν)= _M 〓 ^m=1 C _n S _n (ν)... (4) When expressed as, one specific substance (target substance)
This method realizes a method of calculating only C _i of the component concentration without being interfered with by the values of other component concentrations.

[Principle of the invention]

For example, in the above-mentioned equation (3), _the spectrum analyzer based on the present invention uses the following approximate equation S _X ( _ν ⁾ _{= C _ _} ^{_ _} _{_} Inner product of <S ^* , S _X >
Create. That is ^, _{＜ S} ^{* ,} _{S X ＞} ⁼ _C For K, if we can find S ^* such that the inner product <S ^* , S _Ik >=0, <S ^* , S _R >≠0 ...(7), then C _X = <S ^* , S _X >/ <S ^* , S _R > ...(8) As the interference spectrum S _Ik , k=1, 2, ...,
The concentration _CX of the target substance can be determined without the influence of K.

Such S ^* is called an adjoint spectrum in the present invention, and S ^* (ν)=S _R (ν) − _k 〓 ^k=1 < S _R (μ), U _Ik (μ) > U _Ik (ν ) ... is given by (9). However, K spectra U _Ik
(ν), (k=1, 2,..., K) are orthonormal systems synthesized from K known interference spectra S _Ik (ν), (k=1, 2,..., K), S ^* has degrees of freedom times an arbitrary constant.

By the way, in the above description, <S ^* , S _X >, <S ^* , S _R
Symbols <x(η), y(η)> such as > represent an inner product operation performed between two spectra x(η) and y(η), and the result is a real number. Specifically, the spectral region (η ₁ , η ₂ ) is expressed as r=∫ ⁿ² _o1 x (η) y (η) w (η) dη ...(10), and the spectrum is When expressed as a string of data, it is calculated by R= _N 〓 ⁿ⁼¹ X _o Y _o W _o (11). w(η) and W _o are positive numbers and are window functions for weighting. By appropriately selecting these function values, it is possible to gradually reduce errors caused by the finite spectral region being set.

In the present invention, by making w(η) and W _o particularly small at the edges of the spectral region, the limit of resolution when the spectral region is finite or a collection of discrete discontinuous regions can be avoided. It's improving. FIG. 3 shows an example of a window function.

The adjoint spectrum S ^* is, as shown in Figure 4,
Component S ^* at P wave numbers ν ₀ , ν ₁ , ..., ν _P-1 within the spectral region T set as the measurement range
(ν ₀ ), S ^* (ν ₁ ), ..., S ^* (ν _P-1 ). That is, S ^* (ν) = (S ^* (ν ₀ ), S ^* (ν ₁ ), ...,
S ^* (ν _P-1 ))...(12) Similarly, the interference spectrum S _Ik , k=1,
2, ..., K also S _Ik = (S _Ik (ν ₀ ), S _Ik (ν ₁ ), ..., S _Ik
(ν _P-1 ))...(13)

S ^* (ν) is the K interference spectra S _I1 , S _I2 ,
..., is selected from the orthogonal complement space of the linear subspace spanned by S _Ik . Specifically, from S _I = {S _I1 , S _I2 , ..., S _Ik } ... (14), the orthonormal basis U _I = {U _I1 , U _I2 , ..., U _Ik } ... (15) and calculate the above equation (9).

The above orthonormal basis U _Ik , k=1, 2,..., K is
It is obtained as follows using V _Ik as an intermediate result.

U _I1 = S _I1 / | S _I1 | V _I2 = (S _I2 − < U _I1 , S _I2 > U _I1 ) U _I2 = S _I2 / | S _I2 | V _I3 = (S _I3 − < U _I2 , S _I3 ＞U _I2 −＜U _I1 , S _I3 ＞U _I1 ) U _I3 = V _I3 / | V _I3 | 〓 V _Ik = (S _Ik − _k-1 〓 ^j-1 ＜S _Ik , U _IJ , ＞U _IJ ) U _Ik = V _Ik / | V _Ik | ...(16) Specifically, S _Ik (ν) = (S _Ik (ν ₀ ), S
_Ik
(ν ₁ ), ..., S _Ik (ν _P-1 )), and if S _Ik is flat, ν _i (i=0, 1, ..., P
-1) S _Ik (ν _i )=C (C:const) ...(17) If S _Ik is S _io , S _Ik (ν _i )=Sin(2πi/P-1 ...(18) , V _Ik = {S _Ik − _k-1 〓 ^J=1 ( _P=1 〓 ⁱ⁼⁰ S _Ik (ν _i )・U _IJ (ν _i )} ...(19) can be calculated and prepared in advance.

With the adjoint spectrum S ^* obtained in this way, <S ^* , for all k=1, 2,...,K,
Since the relationship S _Ik >=0 holds, the interference spectrum of a known mixed substance other than the target substance whose spectral shape is predicted in advance does not affect the concentration measurement of the target substance.

[Embodiments of the invention]

Next, details of the present invention will be explained based on examples.

FIG. 5 is a block diagram of one embodiment of a spectrum analyzer based on the present invention. In the figure, 4 is a spectrometer,
5 is a wavelength tunable semiconductor laser; 6 is a sweep control unit;
7 is a sample cell, 8 and 9 are reflecting mirrors, 10 is a semi-transparent mirror, 11 is a photoelectric converter, 12 is a reference cell, 13 is a photoelectric converter, 14 is a processing device, 15 is an input signal processing section, 16 is a sample spectrum Storage unit, 17 is a reference spectrum storage unit, 18 is an interference spectrum storage unit, 18a is an adjoint spectrum generation unit, 19 is a window function storage unit, 20 is a constant storage unit, 21 is a concentration calculation unit, 22 is an input/output device, L _X is the optical path length of the sample cell 7, L _R is the optical path length of the reference cell 12, C _X is the concentration of the target substance in the sample, C _R is the concentration of the target substance in the reference cell,
S _X represents the sample spectrum, and S _R represents the reference spectrum.

The semiconductor laser 5 is driven by a sweep control section 6 and sequentially sweeps discrete wave number positions ν _i under digital control within an appropriately selected spectral region T as shown in FIG. Information regarding the wave number ν _i currently being swept or the sweep control voltage is notified to the processing device 14 for synchronization.

The light output from the semiconductor laser 5 is reflected by the reflecting mirrors 8 and 9 with respect to the sample cell 7, passes through the sample by an optical path length _LX , and is absorbed by the sample substance. The light emitted from the sample cell 7 is split into two by a semi-transparent mirror 10, one of which is input to the photoelectric converter 11, and the other passes through the target substance in the reference cell 12 by an optical path length L _R and is then input to the photoelectric converter. 13.

Each output signal from the photoelectric converters 11 and 13 is input to a processing device 14, amplified and A/D converted in an input signal processing section 15, and further normalized by factors related to the power of transmitted light to obtain a sample spectrum.
S _X and reference spectrum S _R data are generated. These S _X and S _R data are stored in the sample spectrum storage section 16 and the reference spectrum storage section 17, respectively.

On the other hand, the data of orthogonal interference spectra U _I1 ... U _IR created in advance are stored in the interference spectrum storage section 18, and the window function W is similarly stored in the window function storage section 19. Also L _R , C _R ,
Constants such as _LX are also set in the constant storage section 20.

The concentration calculation unit 21 calculates the accompanying spectrum S ^* from the above S _R and U _I1 ... U _IR , and then calculates the above S _X , S _R ,
Using each data of W, L _R , C _R , and L _X , the following formula is calculated to determine the concentration C _X of the target substance in the sample.

_{C X} = L _R C _{R / L X}・< ^{S *} ,S
displayed or printed. Note that the above equation (20) corresponds to the configuration of the spectroscopic device 4 in Fig. 5, and is different from the above equation (8) because the quantity related to the length of the optical path and the value of the reference gas concentration are extracted. There is.

To show a specific example of concentration calculation processing, there is only one interference spectrum S _Ik , that is, K=1, U _I1 is flat, and the point of wave number ν _i in the spectral region is i=0, If we assume that there are 1, 2..., 255, that is, 256 P, then from equation (19'), Then, if S _R = (S _R0 , S _R1 , ..., S _R256 ) ... (22), then in equation (9), ＜S _R , U _I1 ＞U _I1 = ( _R , _R , …, _R )
...(24) The value of each element of vector S ^x is the element of vector S _R
It is obtained by subtracting the average value _R from S _R ¹ , S ² _R . . . , S ²⁵⁶ _R.

Therefore, ＜S ^* , S _X ＞／＜S ^* , S _R ＞=〓〓S _Xi S _Ri −(
〓S _Xi 〓S _Ri )／256／〓S _Ri ² −(〓S _Ri ) ² ／256……(25
) is calculated and multiplied by L _R C _R /L _X to obtain C _X.

Note that in Equation (5), the ignored spectral shape is the unknown interference spectrum S _U (ν), and even if the shape is not precisely known, it changes slowly within the spectral region in question. For example, for the interference spectrum that is known to be
can be incorporated into the adjoint spectrum S ^* by substituting it with a low-order polynomial of . As a result, it is possible to effectively improve a spectrum whose center is far away from the spectral region in question, and where the tails of its absorption lines interfere (for example, as shown in FIG. 2).

Further, the range of the wave number sweep of the semiconductor laser is not necessarily limited to one continuous region, but may be a collection of a plurality of discrete sections as long as it can be used. Of course, one section may consist of only one point. This is because even if there is an actual gas spectrum, there will be parts that contribute greatly to the measurement accuracy of gas concentration and parts that do not, so by omitting measurements of unnecessary parts, measurement time can be shortened. Furthermore, by avoiding areas with a lot of noise and interference, it is possible to prevent a decrease in accuracy.

〔Effect of the invention〕

As described above, according to the present invention, it is possible to easily measure the concentration of a trace amount of a mixed substance using a spectrum analyzer with high sensitivity and short measurement time, almost without being affected by interference spectra.

[Brief explanation of drawings]

Figure 1 is a principle diagram of an analysis device that uses absorption spectra, Figure 2 is a conceptual diagram of an example in which various spectra are mixed, Figure 3 is an explanatory diagram showing an example of a window function, Figure 4
The figure is an explanatory diagram of the wave number position in the spectral domain and the components of the adjoint spectrum S ^* , and FIG. 5 is a diagram of the configuration of an apparatus according to an embodiment of the present invention. In the figure, 4 is a spectroscopic device, 5 is a semiconductor laser, 6 is a sweep control unit, 7 is a sample cell, 11 and 13 are photoelectric converters, 12 is a reference cell, 14 is a processing device, 1
6 is a sample spectrum storage unit, 17 is a reference spectrum storage unit, 18 is an interference spectrum storage unit, 18a
19 is an adjoint spectrum generation unit, 19 is a window function storage unit,
20 represents a constant storage section, 21 represents a concentration calculation section, and 22 represents an input/output device.

Claims (1)

[Claims] 1. In a spectrum analyzer, a wavelength tunable light source, a sweep control means for sweeping the wavelength of the light from the light source within a set spectral region, and a sample containing a mixture of multiple substances using the light output from the light source. spectrum S _X and the spectrum of the target substance
Means for measuring S _R and spectra of K substances S _Ik (k=1, 2,
..., K), the adjoint spectrum such that the inner product <S ^* , S _Ik >=0, and the inner product <S ^* , S _R >≠0
S ^* and the _{concentration} of the target substance in the sample C
and a means ^for calculating based on the calculation < _S ^* , _S An interference spectrum removal processing method for mixed substances in a spectrum analyzer, characterized in that: