EP1561089A1

EP1561089A1 - Highly sensitive spectrometer assembly comprising an entry slit array and a detector array

Info

Publication number: EP1561089A1
Application number: EP03785504A
Authority: EP
Inventors: Andreas Wuttig; Rainer Riesenberg; Christian Schachtzabel; Eckard Kopatzki; Jürgen Popp
Original assignee: CS CLEAN SYSTEMS AG; Templatec-Softwareent Wicklung Christian Schacht Zabel Gbr; Institut fuer Physikalische Hochtechnologie eV
Current assignee: CS CLEAN SYSTEMS AG; Templatec-Softwareentwicklung Christian Schachtzab; Institut fur Photonische Technologien EV
Priority date: 2002-11-11
Filing date: 2003-11-11
Publication date: 2005-08-10
Also published as: WO2004044536A1; DE10253058A1

Abstract

The invention relates to a highly sensitive spectrometer assembly, in which a slit array consisting of parallel individual slits is reproduced on a corresponding detector array with the aid of an optical unit that disperses light in a horizontal direction, splitting it into wavelengths, whereby signals from said detector array are fed to an evaluation device. Characteristic parameters can be determined in a highly precise manner by using fixed slit patterns. To achieve this, the slit array comprises at least one fixed, application-related pattern containing at least two slits that are offset in the dispersion direction by specific values, the spectral images of said slits overlapping on the detector array to form a cumulative spectrum, from which the evaluation device directly extracts the parameters that are to be determined for the application.

Description

HIGHLY SENSITIVE SPECTROMETER ARRANGEMENT WITH INLET SPLIT ARRAY AND DETECTOR ARRAY

description

5 The invention relates to a highly sensitive spectrometer arrangement according to the preamble of the claims.

Spectrometers with multi-slit arrangements are known, for example, from DE 198 13 558 C2, DE 198 15 079 AI, DE 198 15 080 Cl.

10 The basic goal of the known multi-slit arrangements is the measurement of one spectrum with increased resolution or several spectra with normal resolution. In the first case, a high-resolution spectrum is obtained by combining several normal-resolution spectra images from the different columns

15 originate, generated, in the second case different columns are offered light of different spectral distribution and the spectrum of each individual column is determined independently of the other individual columns. The two possibilities have two things in common: First, to carry out all measurements, either switching is necessary,

20 or the spectra images of the individual columns must not overlap each other, since in both cases the spectra image caused by each individual column must first be determined. Second, at the end of the measurement process, one or more spectra are obtained that are completely resolved. In other words: it results

25 vectors of intensity values, each of which belongs to a fixed wavelength or to a small, coherent spectral interval.

The object of the present invention is now to determine the parameters of a spectrum which are characteristic of a specific application, such as e.g. the strength or intensity of one or more basic spectra present in this spectrum or e.g. also determine the width or position of a contained spectral line with high accuracy.

35 This object is achieved according to the invention by the characterizing features of the first claim. The solution is improved by the features of the subclaims, which are at least partially application-related. The invention is based on the following considerations:

When measuring spectra using spectrometers equipped with a grating or a similar dispersion element, the light to be examined must first pass through a gap at the entrance of the spectrometer. The narrower this gap, the better, apart from diffraction or imaging-limited conditions, the better the resolution and thus the more detailed the measured spectrum. On the other hand, in a large number of applications it is not possible to couple any amount of the actually available light into the gap, since the second principle of thermodynamics prohibits concentration in a diffuse form of light beyond a certain level. This means that the amount of light that can be used is limited by the available slit area multiplied by the aperture ratio of the spectrometer. In order to increase the gap area, the gap and the detector elements can be made as long as possible and, for example, a beam former based on optical fibers can be used, which illuminates the entire gap. However, this procedure is limited on the one hand by the imaging errors increasing with increasing slit length and on the other hand by the available detector length. Furthermore, the beam shaping optics is a not to be neglected cost factor and also an additional optical element, which due to possible fluorescence effects in the fibers, strong damping effects in certain wavelength ranges and last but not least the minimum diameter of the fibers, which should correspond to the slit width, is not for suitable for all applications. Another way of increasing the incidence of light in the spectrometer is to widen the entrance slit; however, this necessarily leads to a reduction in the resolution and possibly to a loss of important information. A grating spectrometer with a single slit can measure any spectrum it is offered, regardless of the application, according to its resolution data in this respect a universal spectral sensor. The information contained in the spectrum can be extracted objectively or subjectively from the measured spectral data of the spectrometer using an evaluation unit. However, it is often not necessary to be able to measure any spectra, since only individual parameters, such as the position or width of a spectral line or certain basic spectra in a possibly varying composition, have to be determined anyway. An important example of this is the monitoring of a room for the presence of harmful gases, for example carbon monoxide, chlorine and the like. For such special tasks, it is not absolutely necessary to first determine a universally usable spectrum in order to then determine the desired parameters in an independent step. The invention makes use of this fact in order to optimize the light throughput of a grating spectrometer, possibly at the expense of unimportant signal components of the spectrum offered, without suppressing important signal components. Expressed more generally, important components of the spectrum are amplified by a higher light throughput and unimportant ones are weakened or even extinguished, as a result of which it is generally no longer possible to determine an individual slit spectrum on the basis of the measured data. For the respective application, however, there may be considerable profit. In the simplest case, the application of the invention results in only a single, scalar value, which is extracted by the evaluation device from the intensity distribution applied to the detector array. The use of the slit array enables a higher light throughput, a higher resolution and thus a higher accuracy than a single slit. The spectra images caused by the columns of the slit array generally overlap almost completely and give rise to a corresponding sum signal in the detector array. The use of at least one fixed slit pattern does not require switching during the measurement. Instead, a pattern is selected which particularly well fulfills the task, for example the distinction between two different signals. This is exactly the case when the spectra images resulting from both signals are on Detector array differ as far as possible in terms of the mean square deviation on the individual detector elements, so the sum of the squares of the pixel-wise signal differences becomes maximum. As a secondary condition, it must also be assumed that both signals have the same energy, ie that the summed signal values of the individual detector pixels result in the same value for all possible spectra. Otherwise, it is possible to distinguish the two signals very well only on the basis of their intensity, and the optimal split pattern is therefore a maximum opening. Since the intensity is generally not known or should even be determined, it is not possible in practice to distinguish between different signals based on the intensity, and components other than the constant components of the signals must be used. In the basic version of a suitable slit pattern, identically designed individual slits are arranged in a fixed grid in the slit array level, with certain grid spaces remaining free depending on the task, that is to say no slit is preserved. Alternatively, these positions can also be referred to as closed columns and those with columns accordingly as open columns. A pattern that maximizes the square deviation can now be determined, for example, by a full search across all possible column arrays - ie all possible combinations of open and closed columns. It is also possible to determine a suitable pattern by means of a heuristic, incomplete search if the number of different possible patterns becomes too large, since it is not necessary for the invention to use exactly the best pattern. A further increase in performance of the spectra arrangement according to the invention is possible if a shift in the slit positions relative to the associated raster positions is permitted when searching for a suitable pattern. Studies in this regard have shown that in this way both optimal patterns can be found and patterns which differ from the optimal pattern only by a very small reduction in the gain in accuracy. If more than two signals are to be differentiated, the pattern can be optimized in the same way for a maximum minimum distance of the images of the signals to be differentiated. If, on the other hand, a certain property of a individual signals, for example their half-width or center position are determined as precisely as possible, the first derivative of the spectra image is to be maximized according to the parameter of the property to be determined to the maximum square-mean deviation from the zero line. The sum of the squares of the deviations of two signals with different parameters is thus maximized again, these signals differing only infinitesimally.

For the determination of non-linear parameters, the criteria for the production of optimal samples may have to be modified somewhat, but the procedure remains the same. It is also possible to dynamically adapt the pattern to a changing task. The spectra image created on the detector array corresponds in a first approximation to the spectra image of an infinitely narrow single slit, folded with the transmission function caused by the slit array, that is to say the slit pattern. Depending on the pattern, some Fourier components of the spectrum generated by the slit array are greatly weakened or even canceled, while others are exaggerated. The latter components are optimally those in which the different signals to be detected differ most. Due to the loss of some signal components, it is usually no longer possible to calculate the spectrum originally present at the spectrometer input (slit array), or due to the strong suppression of some components it is considerably less accurate than when measuring with only one slit. As an exception, however, there are also patterns from several columns in which such a back calculation is possible at least without losses compared to a single gap measurement. A calculation back to the adjacent spectrum with a profit compared to a single slit measurement is, however, easily possible if measurements are carried out simultaneously or in succession with at least two different patterns; because then it is possible to suppress different Fourier components of the spectrum for each pattern, but overall to achieve a profit for each Fourier component. This also applies taking into account the fact that the existing signal energy has to be divided into the different patterns in terms of time or space. As already mentioned, the increase in the number of slits compared to the grating spectrometer with a single slit serves to increase the light throughput. On. Maximum sensitivity is achieved if all available light is concentrated on as few gaps as possible, provided that the slit pattern is optimized in the above sense. A fiber-optic beam former can ensure that only open gaps in the slit array are actually illuminated, so that ideally all the available light is used. If fiber-optic beam shaping is not used, make sure that the open gaps are illuminated with the highest possible intensity; because there is no point in expanding the available light in order to be able to illuminate an even larger number of gaps. As an extension of the arrangements with columns of the same width, arrangements can be mentioned which have different gap widths within a gap array and / or whose columns are furthermore arranged at unequal distances from one another. Such arrangements make it possible, for example, to design the gap widths themselves to be substantially larger than would normally be permitted for the resolution to be achieved. Since gaps of different widths cause a loss of different high-frequency signal components responsible for the resolution, a combination of different gap widths can ensure that none of these signal components is suppressed or lost, so that in principle the same result as when using significantly narrower, equally wide gaps can be achieved. Added to this is the fact that narrow gaps lead to increased diffraction losses. By replacing the narrow gaps with combinations of wider gaps with different widths, the diffraction losses can now be minimized and the sensitivity can be several times higher than when using narrower gaps.

The invention is explained below with reference to the schematic drawing. Show it: 1 shows a spectrometer arrangement according to the invention,

2 shows an individual slit and the slit arrays according to the invention corresponding to it, FIG. 3 summation diagrams for the intensity as a function of the number of slits,

Spectral images of three columns, FIG. 5 shows a slit array according to the invention with a slit pattern, FIG. 6 shows the gain achieved with the slit array according to FIG. 5,

7 shows a gap array according to the invention with two gap patterns, FIG. 8 shows the gain achieved with the gap array according to FIG. 7 and FIG. 9 shows a gap array with gaps of different widths.

A spectrometer arrangement 10 shown in FIG. 1 has a primary or secondary light source 11, the light 12 of which is to be examined spectrally. The light 10 enters a monochromator 14 through a slit array 13, which essentially has a dispersing element 15 and generates a spectral image of the slit array 13 on a detector array 16. In accordance with the spectral image, the detector array 16 outputs measurement signals 17 to camera electronics 18, which in turn outputs image data 19 to an evaluation device 20. In the opposite direction, the camera electronics 18 and the detector array 16 receive control information from this evaluation device. The evaluation device 20 evaluates the data and information obtained and displays the extracted parameters, stores them or forwards them to a monitoring device (not shown). The gaps in the gap array 13 can be combined in one or more gap patterns arranged one above the other and parallel to one another; they can have the same or different widths. In a slit pattern, the slits can be arranged in a grid with the same or changing grid constant. The grid constants of two gap patterns arranged on a gap array can also be different. Basically, as many spectra are generated overlapping on the detector array 16 as there are gaps in each slit pattern of a slit array 13 are. The gap array 13 can be replaced in accordance with the respective measurement task.

With the spectrometer arrangement according to the invention, the widths w _l5 w ₂ and the height ratio are, for example, in a spectrum S applied to the slit array 13 the peaks p _l5 ρ ₂ from the evaluation device 20 as a result 21 Wj _. = 1.8 delivered.

2 shows a series of uniform gap arrangements 131 to 135. These are gaps which each have dimensions of 15 μm in width and up to 2 mm in length. A single slit, which is used for comparison with a conventional spectrometer, and slit arrays with 2, 4, 10 and 48 columns are shown in detail. In the array with 48 columns, the outer columns are adapted to the circular illumination of the spectrometer input and are therefore somewhat shorter than the inner column. Using the slit arrays with 4 or 10 columns and the single slit, the sum signals shown in FIG. 3 were measured on a dye spectrum (Congo red). The wavelength specification only refers to the measurement with the single slit. By using the slit arrays with four or 10 columns, the sensitivity when determining the intensity of the broad background signal could be about four or ten times and the sensitivity when determining the amplitude of the narrow peak at about 650 nm could be about two- or three times higher than the measurement with a single slit.

4 shows three diagrams a, b and c, in which the intensity I is plotted over a path x in the plane of the detector array. a contains the intensity distribution generated by a single slit, b shows intensity distributions with a slit pattern m which has three single slits, the intensity distributions 22, 23, 24 being shown offset from one another. The sum spectrum 25 can be seen in c, as actually results from the spectral imaging of the three individual columns in the plane of the detector array. The gap array 13 according to FIG. 5 has a gap pattern 26 with 32 grid positions, 19 of which are occupied by gaps. The gap array with the pattern 10101110110001001011011110001111 (1 = gap, 0 = no gap) is used for universal detection with increased throughput. The slit array maximizes the spectrometer throughput, but at the same time does not suppress any signal component, so that it is also possible to reconstruct the spectral image of a spectrometer with a single slit from the measurement data of the multislit spectrometer. The associated diagram in FIG. 6 shows the gain in sensitivity over the spatial frequency, ie for each individual Fourier component of any spectrum to be measured. It can be seen that the gain for all Fourier components is greater than one. The minimum of the gain is a factor of 1.18, the maximum is 19 for the direct component of the measured signal (spatial frequency 0), corresponding to the number of open columns in the split pattern. An average profit of 3.62 is achieved.

With the arrangement according to the invention, it is possible to precisely determine the width of a Gaussian spectral line, the approximate width of which is known. Such a width determination is usually carried out by adapting a synthetic Gauss line to the measured spectrum. If a multi-slit arrangement were used, a Gauss line of variable width multiplied according to the slit pattern would have to be adjusted accordingly. To determine the optimal gap array, the first derivative of the Gaussian line based on the width parameter for the mean or most likely line width is used as the quadratic difference signal. The resulting gap array is then particularly well suited for determining line widths around this underlying mean width value, but can also bring a noticeable gain even with widely differing widths. As already emphasized, the number of columns used depends on the bundling ability of the light to be measured. The gap patterns are given below as sequences of zeros and ones, with a zero for a closed gap and a one stands for an open gap. A profit is given for each slit array or slit pattern. This specifies the factor by which the detector noise in the case of the multi-slit arrangement can be higher than that with a single slit in order to be able to determine the sought parameter with the same accuracy. The gain in accuracy with regard to the parameter generally depends non-linearly on the noise and is therefore not so easy to specify. For example, it is possible that a line width can still be determined to 10% for a certain detector noise, but that the line width can no longer be determined for a noise that is five times as high.

A slit pattern for the optimization of the measurement of the line width of an approximately 2 pixel wide Gauss line has the following appearance with a pattern width of 128 columns:

1111000011111000011111000011111000001111000011111000011111

0000111110000111110000111110000111100000111100001111100001

111100001111

Profit: 13.7311 (70 open positions)

Optimization for a 0.5 pixel wide Gauss line: Pattern width: 32 columns in the following distribution: 10101010101010100101010101010101 Profit: 5.08047 (16 open positions) The above-optimized gap patterns consist of a series of groups of open columns with an approximately proportional number of individual columns to the line width. The gap groups are separated by gaps of approximately the same extent as that of the gap groups.

If the possible line width spans a larger area, a slit pattern is sought which brings maximum advantage for any line width in a certain area in the spectral imaging and evaluation. The slit pattern is optimized so that the minimum of the profit, that is, the profit with the worst-working line width, becomes maximum. It results With this procedure, exactly the gap pattern that would be optimal for a line width determination with the smallest possible line width. For example, if the possible line widths were between 0.5 and four pixels and the width of the slit pattern was 32, the optimal slit pattern for a medium line width of 0.5 pixels wide would be 10101010101010100101010101010101 and the profit would be at least 5,08047 for all line widths.

Universal detection with increased throughput is also possible with two-dimensional slit arrays (arrays with two slit patterns). 7 is an example of one out of two one-dimensional

Individual patterns existing two-dimensional slit pattern shown. The

Sequences of the two patterns are

10011110110001101001110101011011 or 11100101010001010011111001001111. Assuming that the available light now has to be divided between the two individual patterns, i.e. only half the intensity is achieved per pattern, there is still a profit with a minimum value of 1.39, reached a maximum of 18.51 and an average of 3.60 (Fig. 8).

FIG. 9 is an example of how narrow slits can be replaced by a combination of wider slits with different slit widths and thus minimizing diffraction losses. A two-dimensional gap array with gaps of single and double gap width is shown.

All features shown in the description, the following claims and the drawing can be essential to the invention both individually and in any combination with one another. Bezusszeichenliste

10 spectrometer arrangement

11 light source

12 light

13 gap array

14 monochromator

15 Dispersing element

16 detector array

17 measurement signals

18 camera electronics

19 image data

20 evaluation device

21 result

22, 23, 24 intensity distributions

25 sum spectrum

26 split pattern

131 to 135 gap arrangements

S spectrum

Claims

claims

1. Highly sensitive spectrometer arrangement in which a slit array consisting of parallel individual slits with the aid of an optical unit dispersing in the width direction of the slits underneath

Splitting into wavelengths is imaged on a corresponding detector array, the signals of which are fed to an evaluation device, characterized in that the gap array has at least one fixed, application-related gap pattern with at least two columns offset in the direction of dispersion by specific amounts, the spectra images of which on the detector array form a sum spectrum overlap, from which the evaluation device directly extracts the parameters to be determined for the application.

2. Highly sensitive spectrometer arrangement according to claim 1, characterized in that the gaps are of equal width and are arranged at equal distances from one another.

3. Highly sensitive spectrometer arrangement according to claim 2, characterized in that the distances between individual pairs of adjacent columns to achieve an application-specific

Increased sensitivity equal to a multiple of one

Grid dimension or a minimum gap distance.

4. Highly sensitive spectrometer arrangement according to at least one of the preceding claims, characterized by a beam shaping optics which directs a maximum proportion of the light serving the slit illumination onto the column of the slit array.

5. Highly sensitive spectrometer arrangement according to at least one of the preceding claims, characterized in that the

Gap array consists of at least two patterns arranged at right angles to the direction of dispersion, one above the other, preferably different, and the gap array Corresponding detector array is assigned, which can separately detect the sum signals generated by the individual patterns.

6. Highly sensitive spectrometer arrangement according to claim 1 or 5, characterized in that those patterns are used which allow both a direct extraction of the parameters and a clear calculation of the spectra.

7. Highly sensitive spectrometer arrangement according to at least one of claims 1 and 4 to 6, characterized in that the

Columns have different widths.

8. Highly sensitive spectrometer arrangement according to claim 1 or 5, characterized in that the slit array is designed to be interchangeable and the pattern depending on the particular

Task is selectable.

9. Highly sensitive spectrometer arrangement according to at least one of claims 1 and 4 to 8, characterized in that the columns are arranged at uneven distances from one another.