WO2020053089A1

WO2020053089A1 - Spectroscopic determination of gas concentrations

Info

Publication number: WO2020053089A1
Application number: PCT/EP2019/073821
Authority: WO
Inventors: Andreas Hangauer; Robert Heinrich; Alexandru POPESCU; Rainer Strzoda
Original assignee: Siemens Aktiengesellschaft
Priority date: 2018-09-11
Filing date: 2019-09-06
Publication date: 2020-03-19

Abstract

The invention relates to a measuring device for spectroscopically determining concentrations (c_fit) of gases, comprising: a tunable, monochromatic, light-emitting light source (10); a gas mixture (5) formed from the gases; a first detector (4.1), which the light hits after being partially absorbed by the gas mixture (5); a control and evaluation unit (7), which is designed and programmed to determine the concentrations of the gases from the wave-number-dependent first signal output by the first detector (4.1) and converted from analog to digital, wherein: the deviation between the wave-number-dependent first signal converted from analog to digital and a comparison signal varied by variation of the concentrations of the gases is determined until a specifiable convergence criterion is reached; the comparison signal is determined as a simulation of a characteristic part of the spectrum of the gas mixture; for each gas described by line spectra, collisional broadening parameters for the other gases of the gas mixture are determined from the particular self-broadening parameter by means of addition or multiplication of an experimentally determined constant (gp(j)) dependent on the particular pair of gases. The invention further relates to an associated infrared spectometer, to an associated method, to an associated computer program product and to an associated computer-readable medium.

Description

description

Spectroscopic Determination of Gas Concentrations Field of the Invention

The invention relates to a measuring device, a spectrometer and a method for the spectroscopic determination of the

Concentration of gases in a gas mixture. The invention also relates to an associated computer program product and an associated computer-readable medium.

Background of the Invention

The quantitative determination of gas concentrations is of great importance in various areas of industry and society. These include, for example

Continuous analysis of process gases in the petrochemical industry, calorific value determination of heating gases, exhaust gas analysis or measurement of greenhouse gases in the atmosphere. Many gases characteristically absorb light in the infrared and can therefore be detected using optical measurement techniques.

In general, the transmission of light through a gas is described by the Lambert-Beer law:

The irradiating intensity depends on the

Wavelength l absorbed to different extents. The

molecule-specific absorption coefficient gives the

Absorption depends on the wavelength l of the light. The strength of the absorption scales linearly with the concentration c and length of the path l through the gas, the intensity 1 (A) decreasing exponentially. The absorption of gases can be measured very precisely using tunable laser spectroscopy. Other methods of spectroscopy are the Fourier

Transformation spectroscopy or Raman spectroscopy.

Spectroscopy is therefore an ideal technology for

Measurement of trace gases. The quantitative measurement of

However, several components in a gas mixture is a challenge and has so far only been insufficiently solved by means of optical spectroscopy.

In a gas mixture there is a superposition of the spectra of the individual gases and a collision interaction of the different components. Equation (1) describes the transmission of a gas at a constant pressure and temperature. The strength of the light absorption is not only dependent on the pressure and temperature, but can also be influenced by the interaction of the gas molecules with other gases. On top of that, in some

Absorbance of several gases overlap in the wavelength range and make discrimination more difficult. This

Interaction effects occur in all processes of

Spectroscopy, not only in laser spectroscopy.

However, the case of laser spectroscopy is described in more detail below. However, the claimed method is readily applicable to Raman spectroscopy (which is not absorption spectroscopy) as well as

Absorption spectroscopy with other types of spectrometers, such as the Fourier transform spectrometer, etc. Here the achievable spectral resolution is lower than that of the

Laser spectroscopy, and therefore the broadening effects are somewhat less relevant, but not negligible.

The quantitative determination of such gas mixtures is

spectroscopically difficult because only the compound

Spectrum of all gas molecules can be measured and therefore a very precise knowledge of the spectra of the individual gases is necessary to avoid discrimination by skillful To enable data evaluation. That means you need a high wavelength resolution and information about the behavior of the shape of the spectra depending on the concentrations of the components in the mixture. Although the spectra of many gases are tabulated in specific databases, these do not cover all measurement conditions (interaction in the mixture) or are measured in part with limited resolution.

New laser technologies (e.g. quantum cascade and

Interband cascade lasers (QCL / ICL) enable one

high-resolution measurement of wide areas in the middle

Infrared. Your narrowband emission and broad

Tunability makes it possible to measure spectra with a resolution not previously documented, which makes the evaluation of overlapping spectra of mixtures significantly easier.

The quantitative determination of gas mixtures using

Tunable laser spectroscopy is increasingly becoming the focus.

The concentrations of several gases in a mixture are often determined by measurements with a

Gas chromatograph. Gas chromatographs have been successfully used and developed for decades

Technology. They enable the precise determination of any gas mixture. However, the disadvantages of gas chromatographs are the high calibration effort, the necessary use of carrier and reference gases and a relatively long measuring time of a few to several minutes depending on the gases to be measured.

Optical measurement methods enable the faster and more accurate detection of single gases, but are still limited in the measurement of gas mixtures because of the limited

Wavelength tunability, the limited spectral resolution and due to the described variation of

Spectra shape depending on pressure, temperature and

Gas concentration. So far, broadband light sources have often been used for the optical measurement of gas mixtures. A measurement method that has already been tried and tested many times is Fourier Transformation Infrared Spectroscopy (FTIR). This

enables the measurement of a variety of

Absorption characteristics of the respective gases, however, usually cannot have the same spectral resolution as well as the high ones

Achieve intensities and thus a sensitivity as in laser spectroscopy. There is a higher one here too

Resolution achievable, but this is associated with a very high technical outlay, since long optical paths and a very precise calibration are required. These high requirements limit this process

Laboratory applications for basic research, because the

Measurement systems are too large and too complex to use with the required resolution in an industrial environment.

The same applies to Raman spectroscopy on gases. By measuring a large number of calibrated test gases, attempts have so far been made to enable gas mixtures to be measured using standard FTIR measuring systems which can also be used in the field. The spectra can be calculated using various algorithms, e.g. neural networks, are evaluated and thus enable a quantitative determination of the components. However, this is a complex process

Training process necessary, and the results obtained are often device-specific. In addition, the

Evaluation of the components that were previously considered in the training set. The system behavior with one

Operating status is outside the training room

Undefined. In addition, this operating state cannot be easily detected.

A laser-based gas analysis process is in the

Publication US 2015/0260646 Al indicated. This is based on the use of a measuring device which uses two MEMS-VCSELs emitting in the near infrared (NIR)

(microelectro-mechanical vertical cavity surface emitting laser) the concentrations of the components in Determined hydrocarbon mixtures. The described

Measurement method refers to saturated and unsaturated hydrocarbons with one to three carbon atoms and uses reference spectra of the individual gases and reference spectra of their mixtures for data evaluation. The methodology depends on the individual spectra according to the

To separate measurement and to analyze, if possible, isolated absorption peaks of a respective component. The method is based on the correlation between the area below the absorption peak of the gas and the gas concentration.

Assuming that the area of the absorption line is directly linked to the gas concentration via a factor and that cross-effects due to absorption characteristics do not occur or other gases can be separated from the analyzed line, they achieve the concentration determination of all components in mixtures with five different ones

Hydrocarbons. The area under the

Absorption line associated factor for that

Concentration determination of the gas is also stored in a database during the evaluation.

Summary of the invention

It is an object of the invention to provide a solution for that

Specify spectroscopy, which enables an improved quantitative determination of gas concentrations in gas mixtures.

An invention is presented below, which

quantitative determination of gas concentrations in mixtures improved by means of direct spectroscopy by measuring spectra over a wide wavelength range and taking into account the deviations in the spectral shape of individual components due to the interactions in the gas mixture.

According to the invention the object is achieved with the

Device, the spectrometer, the method, the

Computer program product and the computer readable medium of independent claims solved. Advantageous further developments are specified in the dependent claims.

According to the invention, the external broadening parameters of gases determined from self-broadening parameters are used for a simulation-based determination of the concentration of the gases in a gas mixture.

The invention claims a measuring device for

spectroscopic determination of concentrations of gases, comprising:

a tunable, monochromatic, light-emitting light source,

a gas mixture formed from the gases (consisting of target gas and foreign gases or interfering gases),

a first detector which the light strikes after being partially absorbed by the gas mixture,

a control and evaluation unit, which is designed and programmed, from that of the first detector

delivered analog-digital converted,

to determine the concentration of the gases based on the wave number,

being the deviation between the analog-digital

converted, wavenumber-dependent first signal and one by varying the concentrations of the gases themselves

changing comparison signal is determined until a predeterminable convergence criterion is reached,

the comparison signal as a simulation of a

characteristic part of the spectrum of the gas mixture is determined,

whereby for each gas described by line spectra (also referred to as target gas) external distribution parameters for the other gases (also referred to as external gases or interfering gases) of the gas mixture from the respective self-expansion parameter by means of addition or

Multiplication of an experimentally determined constant dependent on the respective pair of gases can be determined. The invention offers the advantage of determining the concentration of gases in a gas mixture more precisely.

In a further embodiment, the deviation is determined using the mathematical method of least squares.

In one development, the light source can be modulatable, the first signal being demodulated and the

Comparison signal adapted to the demodulated first signal is simulated.

In a further embodiment, the light source can be a quantum cascade laser or an interband cascade laser.

The invention also claims a high resolution Fourier transform infrared spectrometer orrichtung with an inventive measuring _V.

The invention also claims a method for

spectroscopic determination of concentrations of gases, whereby:

a gas mixture formed from gases is provided, light is emitted by a tunable, monochromatic light source,

the light on a first detector after partial

Absorption by the gas mixture hits,

from the analog-to-digital converted, wavenumber-dependent first signal emitted by the first detector

Concentration of the gases is determined

being the deviation between the analog-digital

changing comparison signal is determined until a predefinable convergence criterion (for example in the simplest case a threshold value) is reached, the comparison signal being determined as a simulation of a characteristic part of the spectrum of the gas mixture,

whereby for each gas described by line spectra (also referred to as target gas) external distribution parameters for the other gases (also referred to as foreign gases or interfering gases) of the gas mixture from the respective self-expansion parameter by means of addition or

Multiplication of an experimentally determined constant dependent on the respective pair of gases can be determined.

In a further embodiment of the method, the

Deviation can be determined using a least squares method.

The invention also claims a computer program product comprising a computer program, the computer program being stored in a storage device according to the invention

Measuring device can be loaded, with the computer program performing the steps of a method according to the invention when the computer program is executed on the measuring device.

In addition, the invention claims a computer-readable medium on which a computer program is stored, the computer program being loadable into a memory device of a measuring device according to the invention, with the computer program taking the steps of an inventive method

Procedure are executed when the computer program is executed on the measuring device.

Further special features and advantages of the invention will become apparent from the following explanations of several exemplary embodiments with the aid of schematic drawings.

Brief description of the drawings Show it:

Fig. 1: a block diagram of a measuring device for

spectroscopic determination of

Gas concentrations,

Fig. 2: a graph of determined

Processing parameters y and their

Measurement deviations for investigated interfering gases,

3: a diagram to illustrate the effect of a percentage broadening on the line profile of an exemplary methane absorption,

Fig, a graph of the relative error dropped at the

Methane determination and

Fig. 5: a table of the concentrations of a

calibrated gas mixture with seven components

Detailed description of the invention

The measuring method and the measuring device are described, in which a monochromatic light source that can be tuned over wide wavelength ranges, for example a

Quantum Cascade Laser (QCL), high-resolution direct

Absorption spectra of individual gases and their

Multicomponent mixtures in the medium infrared range (MIR) can be measured. The analysis of mixtures of different hydrocarbons, but also the analysis of individual ones

Components of an exhaust gas mixture are typical

Examples of use.

The quantitative determination of the individual components of

Gas mixtures are carried out using a process that allows the interaction of the gases with one another to be taken into account. This procedure is not on the high resolution

Spectroscopy (eg QCL based spectroscopy) limited and can also be used for FTIR or Raman spectroscopy. The underlying measurement technology and the

the associated analysis methodology is explained below.

Depending on the application, and for measuring the gases

Matching tunable monochromatic target gases

Light sources used. Laser diodes, VCSEL, MEMS-VECSELs, DFG and OPO based laser sources, frequency combs in the infrared, individual QCLs, individual ICLs, with external ones

Resonators equipped variations of these lasers, but also lasers combined into an array can be used.

It is crucial that the light sources used cover the absorption spectrum of all target components of the gas mixture. It does not matter that the spectral

Coverage is complete, but all relevant

Absorption lines and areas are recorded.

It is advantageous to use laser arrays, since they are inexpensive and robust in order to cover the widest possible spectral range. There are several lasers, e.g.

DFB (distributed feedback) QCLs or ICLs, applied on one chip, each in a different one, however

neighboring, emitted wavelength range. These laser arrays have a very narrow-band emission spectrum and high powers in the MIR, so that they have a high

Enable wavelength resolution and high sensitivity of the measurement.

In addition to the high laser power, the high sensitivity is made possible by the strong absorption of many gases in the middle infrared. At 3.3 gm and 6.8 gm, hydrocarbons have the basic modes for stretching and bending vibrations

Carbon-hydrogen compounds, which is why they absorb particularly strongly in this wavelength range. The lasers in the array used can be controlled individually. The intensity and wavelength of the emitted radiation can be continuously tuned by varying the operating current and the laser temperature. Through a combination The continuous individual spectra of the different lasers is thus covered over a wide absorption range of the respective gases.

Fig. 1 shows a block diagram of an optical one

Measuring device for the spectroscopic determination of

Gas concentrations. The one from the tunable

monochromatic light source 10, for example a

Laser / QCL array, radiation emitted is via mirror 1 and beam splitter 2 via three optical paths A, B, C and

Radiation-focusing parabolic mirrors 3 are directed to the detectors 4.1, 4.2, 4.3.

The powers measured by the detectors 4.1, 4.2, 4.3 are recorded with an analog-digital converter and the measured signals are analyzed with analysis software which determines the gases and their concentrations. Furthermore, the current and the voltage impressed on the light source are recorded in parallel with the optical signals via a control circuit 11.

The first path A (= measuring path) contains the measuring cell 5, through which the laser radiation is transmitted and the respective sample gas is absorbed. The sample will be at

Continuous gas flow through the measuring cell 5 measured, the gas being stabilized to negative pressure and temperatures above room temperature. The suppression leads to a better distinguishability of the spectra of the individual components due to the reduced pressure broadening and consequently less overlap of the

Absorption lines. The first detector 4.1 is located in the first path A.

A temperature stabilization of the measuring cell 5 enables reproducible measurements over a long period of time. An increased temperature avoids possible condensation of the

Target gases, such as long-chain hydrocarbons in the measuring system. In the second path B, the light output is measured with the second detector 4.2 without prior absorption. This signal is used to standardize the

Light source power. Since the incident light output is variable, this can be based on this signal

reference.

The third path C with an etalon 6 serves as a relative

Wavelength reference. The etalon 6 causes one

periodic modulation of the

transmitted radiation. On the basis of the equidistant minima and maxima of the modulated power, the emitted

Wavelength of the light source 10 can be determined relatively to the wavelength axis of the absorption spectra

linearize. The third detector 4.3 is located in the third path C.

All three detector signals are not using one

shown analog-digital converter recorded and evaluated on a control and evaluation unit 7. The

Evaluations can be displayed on a display unit 8.

The evaluation of the spectra requires detailed knowledge of the individual spectra of the gas components, which is illustrated below using an example. The spectra of gases are based on the molecule-specific absorption at certain wavelengths, which excite vibrations and rotational vibrations in the infrared gas. The absorption for a certain wavelength can be determined by a so-called

"Line profile" can be described (e.g. the Voigt profile), which essentially consists of the parameters position,

Line width and half width the line is characterized. Other parameters such as pressure-dependent widening or displacement enable the line to be described for different measuring conditions. The transmission function of the gas is given by the Lambert-Beer law

T (n, r, T, a, .., Ck) = exp (-a (n, r, T, a, .., Ck) 1), (2) with 1 the length of the interaction between the sample gas and the light and the wavelength dependent

Absorption coefficient. This is a linear combination of the absorption coefficients of the individual gases:

with K the number of gases and their concentrations c _K. The absorption coefficient of a gas cx ^ (v), which can be described with a line spectrum, is given by a

Summation over all (Voigt) absorption lines: ί, l lines

Here u (r, T, c _lt .., c _K ) is the (Lorentz) half width,

the wave number of the center of the line, uP (T) the Doppler broadening coefficient, sPcr) the line thickness of the i-th line, the j-th gas. F also denotes the Voigt function and k _B the Boltzmann constant. An analytical formula is known for uP (T), uP and S® (To) are tabulated in databases, and for S _j ^® CO / s cr ,,), i.e. the

Temperature dependency, analytical formulas are known. The (Lorentz) half-width or "pressure broadening of the line" is proportional to pressure and the concentrations as follows:

The g are called broadening coefficients. At a reference temperature, these are tabulated in databases. Typically, however, only the

Self-widening coefficient and the air-widening coefficient saved. In this case, K = 2 and j = 1 would be the target gas and j = 2 (dry) air. From this, the self-expansion coefficient is obtained from the database as follows: g _i ⁾ (T0.100%, 0%) (6) and that for air: g ^1} (T0.0%, 100%) (7) for all absorption lines i of the target gas. The

Air absorption lines are usually not taken into account, since Cg and N ₂ absorb only very weakly.

Because of the formal correctness they should also

Broadening coefficients g [ ²⁾ (to, ioo%, o%) ₍₈₎ and

g ²⁾ (T0.0%, 100%) (9) are taken into account. Here, the roles of self-expansion and external expansion coefficient are reversed. The

Tabulated line parameters are determined once from measurements. This succeeds for the foreign

Broadening coefficients are only good if the spectra of the individual gas components do not overlap. The present invention deals with the other case in which the spectra of the individual gas components overlap.

The measurements with the measuring device make it possible to have exactly that effect on the line width in gas mixtures

examine. The process shows that the Considering the processing a better

Concentration determination in gas mixtures can be achieved.

As an exemplary measurement application, the concentration of hydrocarbons in a mixture of up to 7 components should be determined. The methane and ethane contained in the mixture have a typical line spectrum.

From the measurement of the pure spectrum of methane, the

undisturbed line width U _j (rq, T0,100%, 0%, ...) can be determined. The associated self-expansion coefficients are also stored in the HITRAN database. In a mixture

however, the lines widen depending on the

Concentrations of all gases. For hydrocarbons

this broadening has hardly been examined among one another. The widening of the

Line widths of methane in binary mixtures with other hydrocarbons as well as nitrogen can be determined. When determining, the simplification was made that all

Lines of the jth gas widen depending on the interfering gas species and its concentration c _P by the same factor g _P 0 ⁾ . Here, the

External broadening coefficients for the gas p from g _P ® and the self-broadening coefficient determined as follows:

Alternatively, a multiplicative form can be used if it proves that this better describes the measured spectra:

Note that either equation (11a) or equation (11b) is used and that the constants g _p ^} are defined differently in both equations. The For the sake of simplicity, the same symbols are used here

used because this simplifies the following illustration.

The self-expansion coefficients

for the jth gas are determined from the measured clean gas spectrum using a curve fit. For iterative

Equations (2) - (5) are used to simulate the transmission spectrum. No further equations are required, since in equation (5) only one C _j = 100% and all other terms are zero. Since only the self-broadening term is relevant in equation (5), this can be determined individually for all lines i. It is necessary that the other line parameters are known and that the concentration, pressure and temperature are firmly determined. The (as yet unknown) numbers g _p ® are not

needed.

In order to determine the numbers g _P ö ⁾ , after all clean gas spectra have been measured and the self-broadening coefficients for all lines have been determined for the line spectra, spectra are now measured for all binary mixtures (from gas j and gas p). Here are

typically mixing ratios of 50:50, 10:90 or 90:10 are used.

Now equations (2) - (5) and equation (11a) or alternatively equation (11b) are used again

iterative curve fit for each measured binary mixture, only this time with the unknown g _P ö ⁾ and, if gas p also has a line spectrum that absorbs in the relevant spectral range, also

as a parameter to be estimated.

This procedure avoids the effort, the individual processing for each line depending on the interference gas and to determine its concentration. Known evaluation methods either ignore the broadening or try

Correct deviations by data-driven evaluation (machine learning).

Using equations (5) and (11a) or (11b), the broadening of the line depending on the concentration of an interfering gas c _P and an interfering gas-specific factor g _p ® can now be determined for the theoretical line width for 100% concentration of the interfering gas.

The diagram of FIG. 2 shows the determined broadening parameter and its measurement deviation for each investigated interfering gas (= perturber) for an exemplary band of neighboring absorption transitions of methane. The mean value Ymean of all considered methane lines is formed as follows:

The effect of a percentage widening of the

Lorent shared the line profile of an exemplary one

Methane absorption can be simulated with the Voigt profile and is shown as an example in the diagram in FIG. 3. The line parameters are taken from the HITRAN database.

Under exemplary conditions for temperature, pressure and light path through the gas, the transmission is simulated for 80%, 100% and 120% of the pressure-broadened line width. For illustration purposes, a

No influence of foreign gases.

The possibility of simulating the spectrum now makes it possible to estimate the expected error if the

Adjustment of spectra to determine the concentration

Broadening is neglected.

The procedure is as follows. The pure spectrum of 100% methane is measured at a fixed pressure and temperature. The broadening _parameters g _pb > of the lines in the binary mixtures with other hydrocarbons are determined.

The difference between that

The self-expansion coefficient for the methane lines and the external expansion parameter in the mixture is 30 - 50% depending on the external gas. We take one as an example

Broadening by 45% for a mixture when the

Methane concentration goes to zero. So the spectrum can be simulated with a line width of 145% of the self-broadened methane line values. With increasing

Concentration of the methane decreases the broadening, since the methane molecules interact with fewer foreign gas molecules with increasing methane concentration and thus the

Self-expansion prevails. According to (5), the following relationship applies:

Here g is the assumed broadening coefficient due to a foreign gas. We set Y _x / Yi ^H4 (T0.100%) = 1.45.

Now it is checked which error occurs if a concentration-dependent broadened methane spectrum is fitted with the measured pure spectrum. It is assumed that the extraneous gas only has an influence on the line width of the methane, but in the relevant one

Spectral range itself not absorbed. The mixture therefore maintains pressure and temperature and the simulated spectrum is hypothetically scaled with a concentration c.

The following connection is finally with a least

Squares algorithm solved: minS (exp (c _si m log sim (T, p, c _methane , Y, i))) - exp (c _fit - log (

i

As a result, a concentration value Cfit is obtained, which indicates the determined methane concentration, if one

Broadening of the line spectrum is not taken into account. By means of (18) only a test spectrum is simulated, which is given in the curve fit instead of a measurement spectrum. From the curve fit results for different ones

Concentrations Csim let the relative error be free

depending on the methane concentration Cflt in the gas mixture, determine as follows:

Fig. 4 shows a diagram of the relative error fei in the methane determination if a broadened spectrum is fitted with a non-broadened reference spectrum. At maximum broadening, i.e. Cflt towards zero, the relative error fei is almost 14%. With increasing

Relative error tends towards zero in methane concentration. However, if the broadening of the fit is taken into account, the relative error fei in the concentration determination of a line spectrum can be corrected.

In addition to the broadening of methane

high resolution measurements also show that the

The broadening effect is different for the individual gases and in this example for alkanes is significantly lower for binary mixtures of the long-chain hydrocarbons (C3-C5) with one another.

To the composition of a complex gas mixture

determine, it is first necessary to determine the spectra of all pure gases as a reference. By knowing about the

Broadening of the individual gases with each other in binary

Mixture can be used to estimate which line profiles change more when mixed with several components and which weaker. For gases whose line position and parameters are known, it is now possible to use the specific one

External processing parameters the deviation in the Correct absorption determination (using the curve shown in Fig. 4).

This can also be shown with the measuring device on real measurements. The concentrations of the components in a calibrated mixture of the seven hydrocarbons under consideration are determined by measuring the absorption of the mixture over different wavelength ranges. A simple scaling (linear curve fitting) of the

Pure spectra already gives a good determination of the

respective concentrations of the seven components in the mixture, as can be seen from rows (a) and (b) of the table in FIG. 5, where row (a) is the actual

Shows concentrations of the gas mixture and line (b) shows the concentrations without taking into account the broadening.

However, it is also clear from FIG. 5 that the determination for methane deviates the most (~ 0.5% absolute). In this exemplary case, the middle one can now

Determine the broadening coefficient ymean of the methane lines, since the concentrations of the components from the first adjustment (approximately) are known, as well as the mean

External broadening coefficients are known (see FIG. 2). The following applies:

Which results in numerical values from the example:

Ymean = 1-41 Ymean (T, 0%, -, 0%, 100%, 0%) (22)

y times

The mean broadening coefficient is therefore 41% larger than the self-broadening coefficient of methane alone. If the concentration determination is repeated in a mixture with a simulated methane spectrum as a reference, the line widths of which are widened by 41% (lines with Voigt- Simulated profile, line parameters from HITRAN database), it clearly shows that the concentration now determined is significantly better than the calibrated concentration value

corresponds, while the concentration determination of the other gases remains approximately constant (FIG. 5, line (c)). The

The difference between the determined methane concentrations in the measurement without taking the broadening into account is -12.9% relative.

The simulation that estimated the relative error fell (see diagram in FIG. 4: relative error free of

Methane determination if a broadened spectrum is fitted with a non-broadened reference spectrum. With maximum broadening, the relative error is almost 14%. With increasing methane concentration, the error tends towards 0.), determines an expected deviation of -12.3%. The good agreement of the values shows that the expected deviations for the spectra differ with the

let orhersagen Model _V described. The minor

Difference between the measurement and the simulated relative error free can result from permanent measurement uncertainties, such as noise in the detector signal,

Background influence of other spectra,

Concentration uncertainty, calibrated mixture, etc.

In this case, it would now also be possible to take the broadening for ethane into account, since here too

Line spectra occur, the positions of which are known and thus their widths can also be fitted in a mixture.

The same procedure then also enables

Determination of an average broadening and thus a

Correction of the concentration value using a simulated ethane spectrum.

The methodology presented here makes two methods of concentration correction possible: 1. By means of an error curve as in FIG. 4 with a

relative error occurred when determining a methane

broadened spectrum with a non-broadened one

Reference spectrum is fitted. At maximum widening cm 0, the relative error is almost 14%. With

As the methane concentration increases, the error tends towards 0 if the concentration-dependent relative error fell for the respective gas and the measured concentration is corrected by the curve value. This assumes that the other components can be measured so precisely that the average broadening of the line spectrum can be measured using the

Concentration of the other components in the mixture can be determined according to equations (21) and (22). So it would only be necessary a composition-dependent

Correction factor according to equations (21) and (23)

determine to improve the accuracy of the concentration determination.

For the measurement example explained above (table of FIG. 5, correction of methane values without widening by -12.3% relative error), equation (20) follows with free = -12.3%, which is read from the correction curve, Cfit is the determined concentration value with the undifferentiated methane reference spectrum, the corrected one to be determined

Methane concentration Csim at 4.08%. The result shows a significantly better agreement with the calibrated

Concentration value.

2. For spectra in which line spectra of several gases or the widened lines overlap over a wide range with the non-widened (or weakly widened) spectra, the widenings and thus the

respective gas concentrations can be determined iteratively.

Analogous to the previous evaluation in the first

Iteration step determines the concentrations of the components with the non-widened reference spectra. Based on

certain concentration values of the gases can now Broadening according to equations (21) and (22) can be determined for the line spectra. In the next iteration step, the concentrations of all components are determined again, this time simulated line spectra being fitted with the determined broadening parameters. With the concentration values now obtained, a new one can be created

Determine broadening parameters and redetermine the concentration values with an adapted line spectrum. The iteration continues until the error converges to a predefinable limit.

This procedure can thus be applied to unknown mixtures, provided that one knows which gas components

are expected. For example in process gas streams from

This is a real measurement task for refineries.

In order to clarify the procedure, this will be explained step by step here again. The one described

Measuring device allows high-resolution spectra of gases and especially gas mixtures in the infrared, e.g. of hydrocarbons, to be measured with a resolution that has not previously been tabulated or only partially in databases.

Spectra of the pure gases are measured, which as

Serve to reference the concentrations of each

To determine the component in the mixture by scaling the pure spectra. An improved differentiation of the

Superimposed spectra in the mixture is made possible by reduced pressure compared to atmospheric pressure and stabilization of pressure and temperature during the measurement of the gas. Another crucial aspect is the consideration of the line widths of a gas spectrum, depending on the

Composition of the gas mixture. As shown for methane, a broadening parameter y of the lines for the interaction of methane with the other gases in the binary mixture can be determined. This parameter enables the average broadening of the methane lines in mixtures with several hydrocarbons to be estimated, and thus one Calculation of the expected error in the

Concentration determination.

Two methods for improved concentration determination are described:

1. Correct the concentration value by that in a

calculated error curve, read relative errors and

2. Fit method with iterative determination of the broadening of the line spectrum based on the concentrations of the gases in the mixture and simulation of broadened line spectra for determining the concentration.

The determination of the concentration of gas mixtures with simulated line spectra, which takes into account the broadening of the line profile as a function of the concentrations of the other components in the mixture, and simplifies assuming the same percentage broadening of the measured lines, is a new method. The method has the advantage that the line broadening is simulated depending on the concentrations of the other components in the mixture and is therefore not dependent on isolated lines. Heavily overlaid

This allows spectra to be evaluated anyway.

The method can be transferred to any gas combination that is accessible to optical spectroscopy, in particular measurements in different

Wavelength ranges can be combined (e.g. NIR and MIR). The mutual influence of the line shape of an individual gas component by the remaining components can be taken into account with the method described.

Although the invention has been illustrated and described in detail by means of the exemplary embodiments, the invention is not restricted by the disclosed examples and other variations can be derived therefrom by a person skilled in the art without departing from the scope of protection of the invention. In particular, the invention is not restricted to high-resolution spectroscopy (eg QCL-based spectroscopy), but can also be used for FTIR or in Raman spectroscopy.

Reference list

1 mirror

2 beam splitters

3 parabolic mirrors

4.1 first detector

4.2 second detector

4.3 third detector

5 measuring cell with gas mixture

6 etalon

7 control and evaluation unit

8 display unit

10 light source

A first path (measurement path)

B second path (reference path)

C third path (etalon path)

C _fit methane concentration

f Relative error _rei

M molecular mass

T transmission

g broadening parameters v wavenumber

Claims

1. Measuring device for the spectroscopic determination of

Concentrations (C _{f ± t)} of gases, showing:

a tunable, monochromatic, light-emitting light source (10),

a gas mixture formed from the gases (5) and

a first detector (4.1), which the light strikes after partial absorption by the gas mixture (5), characterized by:

a control and evaluation unit (7), which is designed and programmed, from the analog-digital converted output by the first detector (4.1),

to determine the concentration of the gases based on the wave number,

being the deviation between the analog-digital

the comparison signal as a simulation of a

characteristic part of the spectrum of the gas mixture is determined,

whereby for each gas described by line spectra external distribution parameters for the other gases of the gas mixture from the respective

Self-broadening parameters using addition or

Multiplication of an experimentally determined constant (g _P 0 ⁾ ) which is dependent on the respective pair of gases.

2. Measuring device according to claim 1,

characterized,

that the deviation can be determined using a least squares method.

3. Measuring device according to claim 1 or 2,

characterized, that the light source (10) is modulated, the first signal is demodulated and the comparison signal is simulated to match the demodulated first signal.

4. Measuring device according to one of claims 1 to 3,

characterized,

that the light source (10) is a quantum cascade laser or an interband cascade laser.

5. High-resolution Fourier transform infrared spectrometer with a measuring device according to one of the preceding claims.

6. Method for the spectroscopic determination of

Concentrations (C _{f ± t)} of gases, where:

a gas mixture formed from gases is provided, light is emitted by a tunable, monochromatic light source (10) and

the light strikes a first detector (4.1) after partial absorption by the gas mixture (5),

marked by :

that the concentration of the gases is ascertained from the first signal (4.1) emitted by the analog-digitally converted, wave number-dependent signal

being the deviation between the analog-digital

changing comparison signal is determined until a predefinable convergence criterion is reached, the comparison signal as a simulation of a

characteristic part of the spectrum of the gas mixture is determined,

Self-broadening parameters using addition or

7. The method according to claim 6,

characterized,

that the deviation is determined using a least squares method.

8. A computer program product comprising a computer program, wherein the computer program can be loaded into a storage device of a measuring device according to claim 1, wherein the steps of a method according to one of claims 6 or 7 are carried out with the computer program if the

Computer program running on the measuring device.

9. Computer-readable medium on which a computer program is stored, the computer program being in a

Storage device of a measuring device according to claim 1 is loadable, with the computer program performing the steps of a method according to one of claims 6 or 7 when the computer program is executed on the measuring device.