DE102017114422A1 - Method for continuous, time-resolved, spectroscopic analysis of a multicomponent system and spectrometer - Google Patents

Abstract

The present invention relates to a method for the continuous, time-resolved, spectroscopic analysis of a multicomponent system and a spectrometer. The method according to the invention comprises the steps:
a) irradiation of at least one partial volume range of the multicomponent system with electromagnetic radiation;
b) measurement of the intensity of the electromagnetic radiation after passage through the multi-component system for a detection period t and
c) calculating the signal-to-noise ratio (SNR) for at least one component of the multicomponent system based on the measured intensities from step b). According to the invention is a further process step
d) is provided, in which a comparison of the SNR calculated in step c) with a predetermined target SNR SNR (target) is made and, depending on the result, the detection time t for the next measurement in step b) is varied. Thus, a spectroscopic method is provided which can be used flexibly and provides a large number of statistically significant, spectroscopic information for a series of measurements.

Description

The present invention relates to a spectroscopic analysis method for investigating chemical reactions in which a high sampling rate is achieved over an adaptable detection time. Also disclosed is a spectrometer which is suitable for carrying out the method according to the invention, in particular in applications in the context of online analytics.

The key to a resource- and quality-optimized production in the chemical industry lies in the understanding of the individual reactions and in particular in the interaction between cause and effect of the available process parameters on the intended implementation. The more accurate the kinetic and thermodynamic relationships of the underlying reactions are, the more likely it is to optimize processes in practice, so that reproducibly higher space-time yields can be achieved with less energy input.

One possible way to improve the data situation and to control the quality of the manufactured products is that current procedures are followed analytically by a wide range of methods and the process management is optimized on the basis of the data collected. For this purpose, a variety of different analysis methods can be used, in particular, the spectroscopy has prevailed in the industrial environment as an online measurement technology for process control and increase process reliability. These methods are preferably used in the chemical and pharmaceutical environment, since these techniques are non-contact, robust, real-time capable and accessible both qualitative and quantitative information. In particular, the robustness of the analytical method is important in the industrial environment, since in contrast to laboratory experiments, the parameters in the industrial sector are largely controllable only with a significantly increased effort.

For spectroscopic process monitoring, methods with constant detection time per measurement series are currently used. In each individual measurement, the measurement signal is continuously collected on a detector for a defined period of time ("detection time") and stored in the form of a spectrum. The result is a series of discrete measuring points, which reflects the dynamic behavior of the process - integrated over the entire measuring time. The number of measuring points per unit time corresponds to the sampling rate. To resolve the behavior of processes with high dynamics, the highest possible sampling rate should be sought. However, this is only partially feasible, because it works with a constant detection time per measurement series. This approach is due to the requirement that even process states with low signal intensity should be recorded statistically significant. In order to guarantee the minimum required signal-to-noise ratio (SNR) for each data point of the measurement series, the detection time per data point apriori is selected to be sufficiently high. As a result, the maximum possible sampling rate is not achieved over the entire measurement series, thus "giving away" valuable information throughout the entire process.

A spectroscopic method for improving the SNR of a measurement on the basis of a comparison of the spectroscopic background of individual measurements of a measurement series is described, for example, in US Pat WO 2007/050602 A2 disclosed. In this method, a set of spectral data sets of a sample are recorded by spectroscopy. The processing of the individual data records to an entire data record is based on the background of the individual measuring points. The spectral detection is only started when the relative difference between the backgrounds of successive spectral data sets falls below a certain threshold. So only records for evaluation are used, which have a similar background. An active and automatic change of spectroscopic parameters such as the detection time between the successive measurements of the measurement series is not apparent from the document.

In this respect, there is still a need for spectroscopic methods which can be used flexibly and provide a large number of statistically significant spectroscopic information for a series of measurements.

The object is achieved for the inventive method and the measuring device according to the invention by the features of the independent claims. Preferred embodiments of the method and the measuring device are given in the subclaims.

1. a) Bestrahlung zumindest eines Teilvolumenbereiches des Mehrkomponentensystems mit elektromagnetischer Strahlung;
2. b) Messung der Intensität der elektromagnetischen Strahlung nach Passage durch das Mehrkomponentensystem für eine Detektionszeitspanne t,
3. c) Berechnen des Signal-zu-Rauschverhältnisses (SNR) zumindest für eine Komponente des Mehrkomponentensystems anhand der gemessenen Intensitäten aus Schritt b), wobei
4. d) ein Vergleich des im Schritt c) berechneten SNRs mit einem vorgegebenen Ziel-SNR SNR(Ziel) vorgenommen und in Abhängigkeit des Ergebnisses die Detektionszeitspanne t für die nächste Messung in Schritt b) variiert wird.

According to the invention, a method for the continuous, time-resolved, spectroscopic analysis of a multicomponent system, which comprises the steps several times:

1. a) irradiation of at least one partial volume range of the multicomponent system with electromagnetic radiation;
2. b) measurement of the intensity of the electromagnetic radiation after passage through the multi-component system for a detection period t,
3. c) calculating the signal-to-noise ratio (SNR) for at least one component of the multi-component system based on the measured intensities from step b), wherein
4. d) a comparison of the SNR calculated in step c) with a predetermined target SNR SNR (target) is made and, depending on the result, the detection time t for the next measurement in step b) is varied.

Surprisingly, it has been found that the above method is able to map a reaction course with a high possible number of statistically significant data points. This is in contrast to the methods described in the prior art, which work with a constant detection time for an entire series of measurements and thus provide only a constant sampling rate for the reaction. Due to the higher sampling rate, much more information can be obtained about the present chemical process and thus it is possible to make the process more efficient and / or to provide a more even product quality. Furthermore, it is possible to resolve phenomena in processes with high dynamics, which are not observable at a lower sampling rate. It has also been shown that this method can be used in particular for the observation of industrial processes, since non-significant data points (spectra) can be excluded from further processing and analysis via the exclusion criterion of the signal-to-noise ratio. For this reason, this method is accordingly suitable for the online analysis of complex chemical reactions, whereby difficult to avoid and unwanted process fluctuations are taken into account.

For the purposes of the invention, the method is continuous when the same sample is examined spectroscopically several times in succession. This excludes in particular that a spectroscopic examination takes place on a sample, the sample is exchanged and then a further spectroscopic examination takes place on a new sample. The same sample is thus measured several times in succession as a function of time, whereby, of course, different sample subvolumes can be used for the investigation. In particular, the process is continuous when the same reaction is examined spectroscopically several times in succession, for example in a tubular reactor or in a stirred tank.

Time-resolved spectra are obtained by means of the method presented. Since one and the same sample is examined spectroscopically several times in succession, the process progress of a single reaction over the recorded spectra is mapped as a function of time. For example, reaction kinetics can be obtained by knowing the start of the process, as well as the time and duration of the measurement.

A spectroscopic analysis involves the assessment of the interaction between the sample under investigation and the electromagnetic radiation used. In this case, the most varied properties of the interaction of the sample with electromagnetic radiation can be analyzed. For example, it is possible to use the absorption or emission characteristics of a sample as a function of wavelength for analysis. The concretely usable spectroscopic investigation methods are known to the person skilled in the art. Examples are infrared spectroscopy (IR), NIR, Raman spectroscopy, terahertz spectroscopy, nuclear magnetic resonance spectroscopy, etc.

The presented method is particularly suitable for the investigation of multicomponent systems. In the context of the invention, multicomponent systems are in particular substance mixtures which do not consist only of a chemical compound. This is especially true for mixtures or dispersion with multiple compounds. In addition, chemical reactions of one or more starting materials to one or more products by means of the presented method are examined.

In the method, the steps according to the invention are run through several times. This excludes in particular that the inventive method performs the steps only once. In this respect, the steps are carried out at least twice. The indicated sequence is preferably carried out at least more than five times and less than 500 times, more preferably more than ten times and less than 300 times.

In step a), at least one partial volume region of the multicomponent system is irradiated with electromagnetic radiation. The electromagnetic radiation studied includes the entire range of electromagnetic waves as well as those of the particle beams, e.g. Electron beams. Usually radio waves, microwaves, terahertz radiation, infrared radiation, visible light, UV radiation, X-radiation and / or gamma radiation can be used. The radiation is conducted either through the entire or only through a portion of the sample.

The measurement of the intensity of the electromagnetic radiation after passage through the multi-component system for a detection period t in step b) is usually carried out by means of a detector, wherein the detector is selected as a function of the electromagnetic radiation used. Useful detectors for the individual electromagnetic radiation used are known in the art. The intensity measurement of the electromagnetic radiation after passage can take place in the same wavelength range as the electromagnetic radiation used. However, it is also possible, for example in the context of emission spectroscopy, to detect and analyze a wavelength range other than the radiated wavelength range.

The detection period t is the time in seconds in which the wave-dependent intensity detector is enabled. The incident radiation is integrated over this detection period t and intensities are obtained as a function of the wavelength.

In step c), the signal-to-noise ratio (SNR) is calculated for at least one component of the multicomponent system on the basis of the measured intensities from step b). Depending on the number of components of the multicomponent system to be investigated, the intensity of the component-specific signals and the intensity of the background of the measurement are determined either for one or more components. In the simplest case, the signal-to-noise ratio can be determined by a simple quotient. However, it is also possible to use more complex calculations in which, in particular, the individual contributions to the measured background are taken into account. For example, the electronic detector noise or scattering contributions can be treated separately by impurities in the multicomponent system.

In step d), the SNR calculated in step c) is compared with a predefined target SNR SNR (target), the detection time interval t for the next measurement in step b) being varied as a function of the result. The detection period t is therefore not constant over the course of the entire measuring time but is adjusted as a function of the currently available statistics. In the case of sufficient statistics, the current SNR is greater than the predetermined value, the detection time can be shortened for the next measurement without fear that a statistically un-usable measurement is obtained. If the signal-to-noise ratio is below the specified value, a statistically insignificant measurement has been obtained and the detection period must be extended for the next measurement. The sampling rate of the measurement thus follows the course of the reaction and adapts automatically and quickly to strong concentration fluctuations of individual components. In particular, measurement series are not in accordance with the invention, in which the detection period t between individual measurements is kept constant.

Within a preferred embodiment of the method, the SNR of the at least one component can be calculated by a chemometric method which divides the measured intensities into a payload and a background signal, the SNR being the quotient of the payload to background signal (payload: background signal ). A simple calculation of the SNR over the quotient of the intensities has proven to be suitable and robust for industrial analysis. This in contrast to significantly more complex methods, which still break down after additional and different contributions to the radiation intensity.

In a further embodiment of the method, the useful signal can be determined based on the integrated intensity over a wavelength range at which the at least one component interacts with the electromagnetic radiation and the background signal based on the integrated intensity over a wavelength range at which the at least one component does not interfere with the electromagnetic radiation interacts. For many multicomponent systems, spectral ranges can be found in which the intensity can be assigned to only one of the components present in the system. Furthermore, it is also possible to identify spectral regions in which no component-specific signals are present. In order to obtain as meaningful statistics as possible, it is expedient to integrate the intensities occurring in these areas and to determine the SNR using a quotient of the integrated intensities. This method can contribute to more robust statistics.

In a further characteristic of the method, the determination of the wavelength ranges of the useful and the background signal can take place via reference measurements before the actual measurement. Usually, the component-specific signals and the signals of the background result in different wavelength ranges of the recorded spectrum. An assignment of the relevant spectral ranges for the individual contributions can be obtained on the basis of theoretical calculations or by reference measurements. In particular, the latter method can help to further increase the statistics of the measurement, since contributions can also be detected by the measurement environment. Also, by reference measurements better quantitative assignments are possible.

In a further aspect of the method, the variation of the detection period can be effected by a factor a, the factor a resulting in: $$\text{a}={\left(\frac{SNR\left(Ziel\right)}{SNR}\right)}^2$$

A change in the detection time t over the factor a, which results quadratically from the quotient of the target signal-to-noise ratio and the determined signal-to-noise ratio, has proven to be particularly robust and suitable, a significantly higher number of statistically significant measuring points for a series of measurements. By this adjustment factor, both a reduction of the measuring time can be obtained quickly, which in turn leads to the higher sampling rate, as well as a rapid increase of the detection time t for those cases, in which the SNR is below the default.

According to a further, preferred embodiment of the method, the mean value of at least two SNRs of different components can be used for comparison in step d), the two components being in an educt-product relationship with one another. For the spectroscopic investigation of multicomponent systems, it had proved particularly suitable to carry out the proposed methods in parallel with two different components. In particular, it is favorable if one of the components is an educt and the other component is a product prepared from the educt. Accordingly, a statistically significant signal is obtained for both components. Thus, reaction kinetics based on signal changes of only one component can be checked.

In a further embodiment of the method, the SNR of at least two different components of the multicomponent mixture can be determined, wherein the components are in a reactant-product relationship and a sum spectrum from the total measured spectra is used to determine the factor a until the SNR of the individual components above the target SNR. When using two SNRs for further mathematical calculation of the SNRs, it may be useful to base the spectra recorded so far on an average, even in cases where the signals of a component do not have the required target SNR. This can be the case, for example, if there is still very little product at the start of the reaction but correspondingly much starting material. In order to improve the statistics, it may be helpful in these cases not to first consider the spectra in which both components have a sufficient SNR, but to calculate the current product SNR using an average of the product-specific spectral ranges of the entire, previously recorded spectra. This can contribute to an improvement of the statistics and a faster reduction of the detection time t.

Within a preferred embodiment of the method, measurements in which none of the components have a signal above the predetermined target SNR can be excluded from further data processing. For further evaluation of the spectroscopic signals, it may be useful to subject the spectra to further mathematical operations such as averaging. In this case, it makes sense to reject the spectra in which none of the components of the multicomponent system has an SNR as defined above. This can contribute to improved statistics of the averaged spectra.

1. i) eine Strahlungsquelle für elektromagnetische Strahlung;
2. ii) ein für elektromagnetische Strahlung durchlässiges Probenvolumen;
3. iii) einen Detektor zur Bestimmung der Intensität elektromagnetischer Strahlung nach Wechselwirkung der Strahlung mit dem Mehrkomponentensystem;
4. iv) eine Auswerteeinheit zur Verarbeitung wellenlängenabhängiger Intensitätswerte; und
5. v) eine Steuereinheit zur Steuerung der Spektrometer-Einstellungen

wobei die Steuereinheit derart ausgestaltet ist, dass diese als Funktion der Ergebnisse der Auswerteeinheit automatisch und ohne Probenwechsel zwischen einzelnen Messungen zumindest die Detektionszeitspanne t variieren kann. Dieser Spektrometeraufbau hat sich als besonders geeignet erwiesen, komplexe Prozessabläufe in der chemischen Industrie verlässlich und mit hohen Abtastraten zu untersuchen. Auf diese Art und Weise können auch sehr schnelle Änderungen in der Zusammensetzung des Mehrkomponentensystems erkannt werden. Dies kann zu einem besseren Verständnis des gesamten Reaktionsablaufes und zu einer besseren Steuerung der Verfahrensführung beitragen.Furthermore, according to the invention, a spectrometer for the continuous, time-resolved, spectroscopic analysis of a multicomponent system, which comprises at least:

1. i) a radiation source for electromagnetic radiation;
2. ii) a sample volume permeable to electromagnetic radiation;
3. iii) a detector for determining the intensity of electromagnetic radiation after interaction of the radiation with the multicomponent system;
4. iv) an evaluation unit for processing wavelength-dependent intensity values; and
5. v) a control unit for controlling the spectrometer settings

wherein the control unit is designed such that it can vary as a function of the results of the evaluation automatically and without sample change between individual measurements at least the detection time t. This spectrometer design has proven to be particularly suitable for reliably investigating complex process sequences in the chemical industry at high sampling rates. In this way, even very rapid changes in the composition of the multicomponent system can be detected. This can contribute to a better understanding of the overall reaction process and to better process control.

In a preferred embodiment of the spectrometer, the spectrometer may be an online spectrometer. The invention proposed spectrometer is particularly suitable for use in industrial environments and in particular directly at the site of chemical implementation. Due to the increased sampling rate, changes in the process flow can be recorded very quickly, and thus the chance is increased to be able to intervene in the system in the event of deviations in the process flow. In this way, closer process monitoring is achieved and potential reject rates are reduced.

• 1 den Ablauf des erfindungsgemäßen Verfahrens und die
• 2 eine bevorzugte Ausgestaltung des Auswertungsschrittes.

Some embodiments of embodiments according to the invention are shown in the following examples. They show schematically the:

• 1 the course of the method according to the invention and the
• 2 a preferred embodiment of the evaluation step.

The 1 schematically shows a possible sequence of the method according to the invention. Before the start of the reaction, a detection time t is set. The definition of the initial detection time t (0) can be based on expert knowledge. With this detection time, the first spectrum is recorded. The recorded spectrum is subjected to an evaluation and at least one component-specific SNR is calculated. The calculation of the SNR is based on a spectral model or chemometric methods. Depending on the result of a comparison between the current SNR (SNR _i ) and the minimum predetermined SNR (SNR _i (min)), two cases can be distinguished. If the current SNR is greater than the default, the detection time t is reduced for the subsequent step. If the current SNR is smaller than the default, the detection time t is increased for the subsequent measurement.

The 2 shows a special embodiment of the method according to the invention in which the evaluation is based on an SNR value range. As part of the evaluation, it is queried whether the current SNR lies within the specified range. If the current SNR lies above the predefined range, a data correction can be carried out and the spectrum can be included in a spectra averaging. For the next measurements, the detection time t is reduced. If the current SNR is within the range, the detection time t is kept constant for the next measurement. If the current SNR is below the default range, the spectrum is marked, for example, as not suitable for an evaluation, and the detection time t is increased for the next measurements of the series of measurements.

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• WO 2007/050602 A2 [0005]

Claims (10)

Method for continuous, time-resolved, spectroscopic analysis of a multicomponent system several times comprising the steps: a) irradiation of at least one partial volume range of the multicomponent system with electromagnetic radiation; b) measuring the intensity of the electromagnetic radiation after passage through the multicomponent system for a detection period t, c) calculating the signal-to-noise ratio (SNR) for at least one component of the multicomponent system based on the measured intensities from step b), characterized in that a further method step d) a comparison of the calculated in step c) SNR with a predetermined target SNR SNR (target) made and depending on the result, the detection time t for the next measurement in step b) is varied. Method according to Claim 1 , wherein the calculation of the SNR of the at least one component via a chemometric method which divides the measured intensities into a payload and a background signal, wherein the SNR from the quotient of the payload to background signal (payload: background signal) results. Method according to Claim 2 wherein the useful signal is determined based on the integrated intensity over a wavelength range at which the at least one component interacts with the electromagnetic radiation and the background signal based on the integrated intensity over a wavelength range at which the at least one component does not interact with the electromagnetic radiation. Method according to Claim 3 , wherein the determination of the wavelength ranges of the useful and the background signal via reference measurements carried out before the actual measurement. Method according to one of the preceding claims, wherein the variation of the detection period takes place over a factor a and the factor a results to $$\text{a}={\left(\frac{SNR\left(Ziel\right)}{SNR}\right)}^2$$

Method according to one of the preceding claims, wherein for comparison in step d) the mean value of at least two SNRs of different components is used, wherein the two components are in a reactant-product relationship to one another. Method according to one of Claims 5 or 6 wherein the SNR of at least two different components of the multicomponent mixture are determined, wherein the components are in a reactant-product relationship and a sum spectrum from the total measured spectra is used to determine the factor a until the SNR of the individual components is above the target SNR lie. Method according to one of the preceding claims, wherein measurements in which none of the components have a signal above the predetermined target SNR are excluded from the further data processing. Spectrometer for continuous, time-resolved, spectroscopic analysis of a multi-component system comprising at least: i) a radiation source for electromagnetic radiation; ii) a sample volume permeable to electromagnetic radiation; iii) a detector for determining the intensity of electromagnetic radiation after interaction of the radiation with the multicomponent system; iv) an evaluation unit for processing wavelength-dependent intensity values; and v) a control unit for controlling the spectrometer settings characterized in that the control unit is designed such that it can vary as a function of the results of the evaluation automatically and without sample change between individual measurements at least the detection time t. Spectrometer after Claim 9 , wherein the spectrometer is an online spectrometer.

Images