Title of the invention: A spectrometer
The invention relates to a spectrometer and to the determination of properties of samples from their spectra. In its broadest aspects it may be applied for any kind of determinations by spectral analysis. It is however primarily intended for apparatus used for the determination of the contents of fat, protein, lactose, urea, etc in milk.
Prior art A dual beam spectrometer is known from US 5,652,654 showing a beam splitter and a chopper for sample and reference beams. All beams are detected by one single detector. A multichannel system for monitoring a plurality of samples is known from US 5,002,392. The use of optical fibres in spectrometers is known e.g. from US 5,450,194. US 5,020,909 discloses an "In-line" spectral reference method for spectrometers, comprising (a) scanning with a single-beam spectrometer (b) rotating a wheel in synchronization and in an optical path with the single-beam spectrometer, the wheel having open segments alternating with segments comprising reference material, (c) producing alternate spectral scans of spectrometer and sample, and of spectrometer and sample and the reference material, (d) comparing said spectral scans in order to extract the spectrum of the reference material, and (e) using said extracted spectrum for real-time, dynamic spectral compensation to result in absolute correct sample spectrum.
When using spectrometers to determine or quantify the content of specific components and properties of a sample it is essential to know the wavelength axis of the recorded spectrum, i.e. to know the wavelength corresponding to a detected absorbance, and to know this wavelength precisely, i.e. with a predetermined precision/accuracy. This is typically obtained by using expensive control means for mechanical guidance of the movement of a grating. Despite the expensive guidance means the instrument has to be checked by performing reference measurements in various manners, e.g. as disclosed in US 5,020,909. Typically measurements on unknown materials will be followed up by periodically performing a measurement on a known reference material, such as an etalon in order to check the instrument performance. Typically, such reference measurement may be every second measurement or, e.g. every tenth measurement.
The purpose of the present invention to make a simple spectrometer which inherently is wavelength standardised, i.e. the wavelength or wave number in a spectrum recorded by the new instrument may always be precisely derived from the recordings. The spectral range may comprise NIR, MID-IR, Visual and/or UV light.
Summary of the invention
The present invention concerns a spectrometer for the analysis of an unknown object, comprising means for spectral scanning, i.e. providing a monochromatic beam scanning a specified waveband, means for directing the beam in at least two directions, wherein the directing means separates the monochromatic beam into at least two separate beams (19 and 23), a first beam (19) being arranged to be transmitted and/or reflected by the unknown object (20), and a second beam (23) being arranged to be transmitted and/or reflected by a reference medium (24), and the spectrometer is characterised by means for substantially simultaneously recording the spectral patterns representing the spectra of the unknown object (20) and the reference medium (24), wherein the reference medium shows a characteristic spectrum having remarkable maxima and/or minima at known fixed frequencies (or equivalent wave numbers or wavelengths) which are inherently determined by its properties, such as chemical, physical and/or mechanical properties, the simultaneously recordings of sample and reference spectra being performed during each single scan, means for correlating the substantially simultaneously recorded spectral patterns, in order to determine the wave numbers and/or wavelenghts or frequencies in the spectrum of the unknown object from the substantially simultaneously recorded spectrum of the reference object.
By these means it is possible to determine the wave numbers in a recorded spectrum of an unknown sample simply by simultaneously recording a corresponding spectrum of a known object having an adequate spectrum incorporating easy recognisable peaks and valleys.
In a presently preferred embodiment of the new spectrometer the recording means comprises detectors means (22, 26, 40), for substantially simultaneously detecting the at least two transmitted and/or reflected beams (19, 23), data storage means (60), and data processing means (50), the detector means (22, 26, 40) being connected to the data storage means (60) for storing spectral patterns, said data storage means storing the substantially simultaneously recorded spectral patterns of the unknown object as well as of the reference medium, and wherein the stored data are processed in the data processing means (50)
The big advantage is that this spectrometer does not need expensive mechanical guidance in order to present a linear characteristic, such as a smoothly varying wavelength versus time or versus angular position. And it does not need any regular adjustments.
In this specification "recorded spectral patterns" means any collection of information such as data representing an almost or substantially continuous monotone list of values representing a spectrum, not necessarily a linear spectrum, most likely a very non-linear spectrum, i.e. a varying number of measurements per time unit and/or per wavelength or wave number unit.
The means for providing a monochromatic beam scanning a specified optic range may be any optical means able to diffract or refract the light, such as a grating, a holographic grating, or a prism.
The reference medium is preferably a polystyrene film. The spectrum of Polystyrene has a number of characteristic peaks and minima in the infrared range. Other possible media are a Mylar film and a fluid such as a monostyren in a cuvette.
As a further reference medium an etalon, such as a Fabry Perot etalon may be useful. Such an etalon will provide a number of equally spaced peaks on a wave number-axis.
The spectrometer according to the invention is connected to calculation means arranged to locate and/or relate any recorded sample measurement to a specific point or defined section of the wave number-axis.
Such calculations are based on the correspondingly recorded reference measurements, which are recorded simultaneously with the sample measurements. Absorbances measured at the same time have exactly the same wavelength.
The unknown object may be a flow cuvette comprising a fluid passage, e.g. a MID-IR cuvette for analysis of dairy products such as milk, the measurement being performed either during a continued flow, or in periods in which the flow is stopped. The stopped flow is preferred in the case of dairy products.
An ideal spectrometer would provide a spectrum of an unknown material using an exact linear and known frequency or wave number axis as well as a linear and known absorption axis. Unfortunately this is pure theory, and never embodied in reality. Generally the wave number axis is non-linear and showing considerable variations. Figure 5 illustrates how an irregular speed control, i.e. angular velocity of the grating may imply a non-linear wavelength axis. Even in the case of spectrometers applying expensive control means wavelength shifts may occur, e.g. due to a change in temperature. The present invention relates to a spectrometer using very simple means for moving the scanning means. Accordingly the wavelength axis can be expected to be highly non-linear.
Accordingly it is a problem to identify the exact wave number in a measured spectrum. Such knowledge is crucial for the accuracy of a spectrum-based determination of the content of specific components in an unknown sample.
The present invention provides a cheap solution to this problem.
Brief description of the drawings:
Figure 1 shows a schematic diagram of a simple spectrometer according to the invention.
Figure 2 shows an explanatory arrangement of data buffers for the spectrometer in Figure 1 , and three scans. Figure 3 shows a schematic diagram of a further embodiment of a spectrometer according to the invention.
Figure 4 shows a explanatory arrangement of data buffers for the spectrometer in Figure 3.
Figure 5 illustrates the movement of a grating.
Figure 6 a spectrum of polystyrene. Figure 7 a schematic spectrum of an Fabri Perot etalon.
Figure 8 a section of a spectrum of an unknown sample.
Figure 9 shows a four branched embodiment of a spectrometer according to the invention.
Detailed description of preferred embodiments:
The arrangement:
Figure 1 shows as an example a schematic diagram of a first embodiment of a spectrometer according to the invention. The schematic spectrometer of Figure 1 comprises a light source 12, a slit 16, a rotary grating 14, a second slit 16, beam dividing and/or directing means 18 (e.g. a beam splitter or a beam chopper, or split optic fibres / a bunch of optic fibres split in at least two cores) providing at least two beams 19, 23, a sample compartment, cuvette, flow cell or holder 20 and a first detector 22, a reference object, e.g. a polystyrene film 24, and a second detector 26.
The light source 12 may be any kind of light source providing a desired spectral range, i.e. a range being useful for the intended analysis. The rotary grating may be replaced by any other scanning means, e.g. diffracting or refracting means, e.g. a prism.
The beam dividing and directing means may be a (number of) semi-reflecting mirror(s) or a beam chopper alternatively directing the beam into two, three or four different directions (chopping very fast), or a bunch of optic fibres having a common input and two, three or four separate output. In an advantageous beam divider the majority of the power,( i.e. intensity) of the light is directed towards the unknown sample. Only minor fractions are used for the reference purpose.
In a preferred embodiment the sample cuvette 20 is arranged to receive a fluid sample between windows which are transparent to the selected waveband. The reference object 24 is a material having a spectrum including easy recognisable peaks or valleys in the selected waveband, e.g. a polystyrene film. The optional reference object 28 shown in Figure 4 is an etalon or empty
cuvette having two parallel windows providing a plurality of resonance peaks and/or valleys useful for determining the wavelength-axis for a large range.
The detectors are connected to electronic driver circuits, converters and/or interface circuits 40, as well as electronic registration means, e.g. data buffers (60) and data processing equipment, such as a computer 50 (which may be of conventional art) arranged to store and process data representing the intensity of the light beams transmitted through the sample and the reference object(s), respectively.
Preferably the data are stored in data buffers as shown schematically in Figure 2 and 4. Figure 2 specifically shows 3 scans of the same sample and reference medium. Typically the three scans differ in that the same wavelength does not occur at the same location in the three scans. This is due to the use of very simple scanning means, causing an uncertainty of the actually provided frequency or wavelength.
Operation:
A presently preferred embodiment of the new spectrometer is operated in the following way: The light source 12 emits light in a specified waveband. The light hits the rotary grating 14 and the slit 16. The beam on the right side of slit 16 is a monochromatic light beam 17 which is split into two (or at least two) or preferable three beams by the beam splitter/divider 18 . The two beams 19, 23 having exactly/(or almost in case of a beam chopper) the same wavelength hit the sample 20 and the reference object 24, respectively.
In a preferred embodiment shown in Figures 3-4 the light from the grating 14 enters a bunch of optical fibres 118. The bunch is divided into three smaller fibre bunches or three single fibres, carrying three beams 119, 121 , 123 directed to hit the sample 20, and two reference objects 24, 28, respectively.
The detectors 22, 26, 30 and adjacent electronic registration equipment (not shown in Figure 3) measure the intensity of the light beams transmitted through the sample and the reference object(s), respectively. The measurements are converted into digital numbers which are stored in data buffers. This is explained in further details below.
During a scan of a specified waveband range a plurality of intensities are recorded as a plurality of numbers stored in a plurality of locations in a data buffer, cf. Figure 2 and 4 in which each box represents a location, containing a number representing the intensity of light reaching the detector. The wavelength of this light is so far unknown (the known apparatus often use a co-
ordination between the angle of the moving grating and the wavelength, but this is either expensive or poor).
According to the present invention at least two scans are recorded simultaneously (or almost simultaneously in the case of a beam chopper), preferably using at least two detectors (22, 26). In a first preferred embodiment the beam splitter 18 provides at least two continuous or chopped beams 19, 23, one beam passing through the sample, the other(s) through the reference medium (media). The recorded spectra are stored in data buffers as indicated in Figure 2 and 4.
One of the reference media must be a known medium having easy recognisable peaks at well- defined wave numbers, preferably polystyrene. Polystyrene is a good medium because the spectrum shows several peaks which are easy to identify as specific wave numbers. For spectral analysis by which the desired wave numbers are close to the peaks and/or valleys of the polystyrene spectrum it might be possible to use only the polystyrene as a single reference medium. Figure 6 shows a spectrum of polystyrene.
Due to the spectrum measured by the second detector, and thanks to characteristic patterns in the spectrum of polystyrene, it will be possible to recognize specific wave numbers in the measured spectrum of polystyrene.
This is illustrated in Figure 2. Figure 2 shows schematically data buffers storing data including peak values measured in 3 scans. D1 indicates the data buffer storing the sample spectrum. D2 indicates the data buffer storing the polystyrene spectrum. Four distinct peaks of known wave numbers in the polystyrene spectrum are indicated by a hatching of the related location in the data buffer and are marked by v*-, v2, v3 and v4. As shown in Figure 2 they do not occur at the same locations in the three scans. Due to the distinct values in the reference scan it is however possible to locate the peaks in every scan. Each peak corresponds to a specific wave number. It is known that the first peak is located at a wave number v-,. Accordingly the corresponding number "X representing the intensity of the signal detected by detector 1 , can be interpreted as a measurement of either the absorption (or absorbance in), transmittance through, or reflectance by the sample at the wave number v, - dependend on the measurement arrangement. The wave numbers v2 , v3 , v4 are identified in the same way.
By suitable interpolations it might even be possible to assign specific values to all points on the wave number axis.
However, to ensure a broader spectral knowledge it is preferred to include at least one further reference, preferably a Fabry Perot element or etalon. Such element will due to internal reflection
provide a contiguous spectrum of equally spaced resonance peaks, if measured on a ideal linear frequency axis. In this way the accuracy of the values assigned to the wave number axis may be improved. Figure 7 shows schematically a spectrum of an etalon. An empty cuvette having substantially parallel windows may be used as well. In Figure 4 D3 shows schematically the locations in the data buffer assigned to Detector 3 detecting the spectrum of an etalon. The schematic diagram is intended to indicate that the data buffer D3 includes a plurality of locations (shown as hatched) containing registrations of spectral peaks or valleys providing a possibility to obtain an exact knowledge of the corresponding wave numbers when combined with the peaks and/or valleys of the polystyrene spectrum.
Accordingly the knowledge of the two reference spectres allows for allocating a great number of (or even all) the boxes/locations in the data buffer representing the recorded sample spectrum to a specific substantially true wave number. A necessary condition for using this method is that the number of recorded data in each scan is big.
The following is a simplified example showing the actions which may lead to the desired knowledge of the frequency or wave number of the spectral axis: The computer recognises a number of peaks belonging to the polystyrene spectrum, e.g. v^ v2, v3, and v4. One of the repetitive peaks in D3 is substantially coincident with v-*. Another peak in D3 is substantially coincident with v3. The mutual distance is 7 Δv.
From this Δv can be calculated as Δv = (v3 - v^ / 7. Accordingly the wave numbers of all of the hatched sections in D3 may be determined as e.g. . v.- + n* Δv, where n is an integer.
When using chopping of the beam the chopping frequency must be much bigger than the scan frequency. It is essentially to have a great number of data for each scan, and essentially that these data are divided into at least two, (preferably three), almost equally big amounts representing sample data, reference data of polystyrene, and preferably (optionally) reference data for Fabry Perot element, respectively.
In a further advantageous embodiment shown in Figure 9 a fourth branch has been added to the embodiment of Figure 3. The fourth branch includes a zero sample 32, (such as Zero cuvette containing a zero fluid), a detector D4 and electronic signal processing means, but no data buffer for evaluation of wave number. The information is applied for the calculation of the data which are stored in the data buffers D1 , D2, ...Dn.
They are calculated as the absorbance: - log10 (d dz).
Preferably this calculation is performed instantly, i.e. there is no need for storing the zero values in a data buffer. In this way a double beam absorbance spectrum is acquired.
Special features:
The light beam preferably emerged from a e.g. vertical slit, the upper and lower sections being deflected towards reference media and/or a zero medium, and the centre section proceeding towards the unknown sample.
Figure 5 illustrates that the movement of the grating can be fairly random. In the present invention there is no need for exact control of this movement. The wave number of the light will or may vary, generally in a non-linear way. As it appears from Figures 2 and 4 an exact knowledge of specific wave numbers are established by use of a suitable reference medium according to the present invention. The scanning means may include a variable bandpass filter.
In a further embodiment the beam splitting is provided by chopping the direction of the beam, e.g. by a toggling mirror. In this embodiment it is essential that the chopping frequency is high compared to the scanning frequency.
In an advantageous embodiment the monocromatic beam is collected in a bunch of optic fibres which is split into e.g. four separate fibre bunches.
The source may comprise a single light source covering the desired waveband or a plurality of light sources each of the light sources covering .a specific portion of the desired waveband. The scanning means may comprise an Acoustic Opto Tunable Filter (AOTF).
In the MID-IR-range the light beam is preferably chopped by a rotary wing or a tilting/toggling mirror. The beam chopping is preferred in order to provide for adequate detection of the light beams.
Further, the chopping, e.g. by a rotary wing, may be applied in advantageous manner to identify a co-ordination in time between the plurality of (or the at least two) storage means. In this way any differences in time delay in the means for recording the spectra (detectors, drivers, converters, interface circuits and data buffers or similar storage means) may be compensated or taken into consideration when performing the analysis of the recorded values and the associated calculations leading to determinations of specified contents.
For the MID-IR range the detection means may be pyro-electric detectors such as LiTa03. For visual light Si detectors may be applied. In case of NIR the preferred detectors are made of PbS or PbSe.
The spectrometer preferably includes means for thermostating the sample cuvette.
Calibration and standardisation:
Typically such spectral information is applied to determine specific properties of the sample such as the content of fat in a milk sample. Specific calibration software are applied for executing such a determination. When providing and using such calibration software it is essential to use standardised equipment. By standardised equipment is meant that the spectrometric instrument measuring the new unknown sample is standardised in such a way that it provides measuring results which are identical to the results which would be obtained using the spectrometric instrument originally used when elaborating the calibrations.
In principle calibrations may be provided individually for any spectrometer. However to fully utilise industrially manufactured spectrometers it is desirable that all instruments of the same type originating from the same manufacturer can use the same calibrations for determining e.g. the content of fat in milk.
The standard calibration provided for the determination of e.g. fat, will be based on measurements corresponding to a specific set of wave numbers or frequencies. According to the present invention the measurement results recorded for the unknown sample may be shifted by interpolations to values corresponding to the "standardised" wave numbers or frequencies, which were used for the standard calibration.
A standardisation of the frequency/wave number axis may imply an interpolation of all recorded sample values to be shifted to predefined "standardised" wave numbers or frequencies- corresponding to the supplied calibrations, which allow for an accurate determination of the content of a specific component such as fat, protein, lactose etc. in milk.
Alternatively, the calibration constants can be transformed to comply with the actual instrument (i.e. the actual Δv of the etalon)
Preferably, the measurements are standardised to a plurality of predetermined desired standardised wave number points. Such standardisation may be accomplished by applying a plurality of identified measurement points below a desired standardised wave number point, as well as a plurality of identified measurement points above the desired standardised wave number point for calculating, by interpolation, e.g. cubic interpolation, the corresponding value
(absorption or reflection) at the desired standardised wave number point. Figure 8 illustrates such interpolation. For the sake of clarity only a small section of a spectrum is shown. The actually measured points are indicated by small circles "o" and the interpolated points at the desired "standardised" wave numbers are marked by small squares, "□". All relevant portions of the spectrum (i.e. all portions to be applied for the purpose of the spectral analysis) should be handled in the same way, in order to determine the absorbances at a number of specified standardised wave numbers.
When the absorbances at the specified standardised wave numbers have been calculated by interpolations the standard calibration can be applied for determining the contents of the desired components.
In the above are described a few embodiments of a spectrometer, which may be characterised as a post-determined wave number spectrometer i.e. a simple spectrometer, in which the wave numbers are determined after the scan.
The terms and expressions employed in the above description are used as descriptive terms and not as limiting terms. There is no intention in the use of such terms and expressions of excluding any equivalents of the features shown and described or portions thereof. The described embodiments may be varied in several ways without diverting from the scope of protection as defined by the following patent claims.