GB2119086A

GB2119086A - Reduction of measuring errors in spectrophotometers

Info

Publication number: GB2119086A
Application number: GB08309636A
Authority: GB
Inventors: Wolfgang Witte
Original assignee: Bodenseewerk Perkin Elmer and Co GmbH
Current assignee: PE Manufacturing GmbH
Priority date: 1982-04-24
Filing date: 1983-04-08
Publication date: 1983-11-09
Also published as: FR2525767A1; JPS593223A; DE3215367A1; GB8309636D0

Abstract

Measuring errors due to stray light, in particular the effects at each measured wavelength lambda m of stray light from other wavelength bands, are reduced in a spectrograph 22 having a diode array 24 as a detector and an entrance slit on which a combined beam 20 comprising a reference beam 16 and a sample beam 14 passing through a sample 18 is incident. Proportion factors Sim, measured as instrument constants, are stored in a memory 28. These factors measure the proportions by which the deviation of the transmission Tmi occurring in a wavelength range DELTA lambda i from the transmission Tm at a measured wavelength lambda m affects, by stray light, the measurement of the transmission Tm. The spectrum is detected and the detector signals for the wavelengths corresponding to the different diodes are stored in a memory 26 for both the reference and sample phases. A computer unit 30 forms a stray light signal and processes the stored signals to form corrected transmission signals which are stored in a memory 32 and indicated on a display 34. <IMAGE>

Description

SPECIFICATION Reduction of measuring errors in spectrophotometers This invention relates generally to the reduction of measuring errors due to stray light in a spectrophotometer.

Generally speaking, the light measured by a spectrophotometer may be either a transmittance or a reflectance. In conventional spectrophotometers, if such a measured quantity is measured at a certain "measuring" wavelength Xm light of other wavelengths, referred to as "stray light" also falls on the exit slit and the detector. This stray light is light which has not passed along the normal optical path through the instrument but reaches the exit slit along other paths, e.g. due to reflections from optical elements or due to scattering. Such stray light can result in falsification of the measurement.

Opticai measures are known for reducing this stray light. A double monochromator, for example, in which the exit slit of a pre-monochromator forms the entrance slit of a second main monochromator is one such measure. Thereby light from the wavelength range being measured reaches the entrance slit of the main monochromator. Any stray light occurring at the exit slit is further reduced by the main monochromator. The optical and mechanical expense involved with such a double monochromator, however, is considerable.

Further, spectrographs are known which simultaneously detect an extended spectral range in their image plane, for example, by means of a detector array. In such spectrographs, the reduction of the stray light presents particular problems as it is not possible to filter out any partial ranges of the spectrum.

The present invention is based on the principle of reducing measuring errors due to stray light in spectrophotometers by suitable signal processing rather than optically. This principle is put into effect, in accordance with the invention, by an arrangement comprising a memory in which proportion signals are memorized as measured instrument constants of the spectrophotometer, which proportion signals correspond to proportion factors (siren) with which a photometric measured quantity (Tmi) occurring in a wavelength range (aha) affects, by stray light, the measurement of these measured quantities (Tm) at a measured wavelength (arm), means for detecting a spectrum to be measured and for memorizing signals corresponding to the values of the measured quantity (Tmi) in the wavelength ranges (aha), means for weighting the memorized proportion factors (Sjm) in accordance with the values of the measured quantity (Tmi) in the associated wavelength ranges (aha), means for summing the proportions (sjm(Tmj-Tm)) thus weighted for generating a stray light signal (ARm) and means for correcting the spectrophotometric measuring signal (Tm) obtained at the measured wavelength (Xm) for the stray light signal (AT) to generate a corrected measuring signal (Tkm).

The proportion factors with which a photometric measured quantity occurring in a wavelength range AB affects the measurement of this measured quantity at a measured wavelength Xm as a result of stray light, can be determined experimentally. These proportion factors are stored in a memory as instrument constants. It can be assumed that the stray light component is usually small in comparison with the useful light component. First a spectrum to be measured is detected. The proportion factors sjm are then weighted in accordance with the values of the measured quantity measured in the associated wavelength ranges AB and the products thus obtained are added.A stray light signal for the measured wavelength Bm results, for which the measured signal obtained at this measured wavelength Xm is corrected. Thus, each wavelength of the spectrum examined may be measured in the manner described as "measured wavelength".

An arrangement in accordance with the invention will now be described with reference to the accompanying drawing which is a block diagram showing the arrangement in conjunction with a spectrophotometer designed as a spectrograph.

If a measured wavelength Bml which may have any value, is considered the useful light comes substantially from a wavelength range of width 2army around the measured wavelength, Sml AXml being the spectral bandwidth of the spectrophotometer. The useful light is distributed in this wavelength range in accordance with the slit function of the spectrophotometer. For purposes of this explanation, it is assumed that the spectrophotometer is a double beam instrument. Thus the useful light defined provides, at the detector, a signal Sm during the sample phase and a signal Smo during the reference phase.The signals 5m and Smo respectively of the detector are each proportional to the radiation density of the light source, to the transmittance of the spectrophotometer and to the sensitivity of the detector for the wavelength kms The sample transmission T measured by a spectrophotometer without stray light is then

In practice, however, stray light affects this value. To determine the effect of stray light, a wavelength interval ahi is considered at a wavelength k1 which is different from fm Initially, the wavelength range ABj is assumed to be so small that the sample transmission within the range Akj, is effectively constant.This sample transmission can be designated Tj.

The stray light within the wavelength range nhl generates a signal S during the reference phase, i.e.

when only the reference light beam impinges upon the detector. The sample light beam then generates a signal S1T1 atthedetector.

If the entire spectral range is assumed to be continuously covered by such wavelength ranges ABj (i = 1, 2 .. n) with the range 2AXm around the measured wavelength fm spaced out, the sample transmission Tm at the measured wavelength Bm, effectiveiy measured from the signal proportion, will be

T being the true value of the sample transmission at the measured wavelength Bma If, however, instead of the absolute signals Si, the signals

referenced to the signal Smo are introduced, the equation (2) is transformed into

The transmission Twill equal unity if both the sample and the reference light beams are unobstructed i.e. if there is no sample in the sample light beam.

In the same way, all transmissions are equal to unity.

This yields

The spectrophotometer does not refer the sample signal to the ideal reference signal Smo, but to the actually measured reference signal,

affected by stray light, or, expressed in relative quantities, nottothe reference value "1" buttothe reference value

A quantity S may be introduced to take this into account, where:

Firstly, equation (4) may be written in the following form:

Together with 5 according to equation (6), this equation becomes:

Thus, the transmission error, T, due to stray light is

This formula explains some substantial features of the transmission error, AT, due to stray light.

For example, the transmission error AT = 0, when the sample light beam is blocked, i.e. when both T = 0 and all Tj's = 0.

Further, with an unobstructed sample light beam, i.e. when there is no sample in the sample light beam, and T = 1 and all T1,s = 1, the transmission error is also AT = 0.

At medium transmissions, however, the transmission error due to stray light may be positive or negative or may even be zero depending upon whether the transmissions Ti, with the weighting given by sj, are larger or smaller than the transmission Tat the measured wavelength .

The transmission error AT due to stray light cannot be determined according to equation (9), as the true transmissions T and Tj are not known. If, instead, the measured values Tm and Tm are substituted, this yields

Equation (10) is used in the spectrophotometer described hereinbelow to determine the transmission error Tm for each measured wavelength. This transmission error Tm is subtracted from the measured transmission value Tm so that a corrected transmission value (11) Tkm = TmATm is obtained.

Turning now to the drawing, a light source 10 emits a light beam 12 which is split into a sample light beam 14 and a reference light beam 16 in any conventional manner. The sample light beam 14 passes through a sample 18. Subsequently, sample light beam 14 and reference light beam 16 are re-combined to form a light beam 20. The light beam 20 impinges on the entrance slit of the spectrograph 22 proper. Preferably, the spectrograph 22 includes a diode array 24 as a detector. The spectrograph 22 generates a spectrum in the plane of the diode array 24. The diodes of the diode array 24 each detect a wavelength range around an associated mean wavelength. In one embodiment, for example, the wavelength range detected by the diode array 24 can extend from 400 to 800 nm.

During the reference phase, the detector signals for the wavelengths Am allocated to the different diodes are supplied to a memory 26 and are stored therein. These signals correspond to the denominator in equation (2) and are designated by Smo. The detector signals Sm corresponding to the numerator in equation (2) are stored during the subsequent sample phase. The proportions sjm for the different wavelength ranges Xj are stored, i.e. memorized, in a memory 28, which proportions contribute stray light from the wavelength range fj to the transmission error ATm at the measured wavelength Xmw These proportions sjml or percentages, are instrument constants of the spectrograph, which instrument constants are experimentally determined and stored in a way to be described.

A computer unit 30 receives the stored detector signals 8m*o and 5m* for each measured wavelength Xm and forms therefrom the measured transmissions Tm and Tm respectively. Then the transmission error ATm due to stray light is formed for each measured wavelength Xml in accordance with equation (10). The transmission signal Tm is then corrected for this transmission error in accordance with equation (11). The thus corrected values Tk of the transmission are provided to a memory 32 and indicated on a display 34.

The use of a diode array 24 offers the advantage that it supplies the spectrum in the required form, namely in the form of many discrete individual values. It is preferred, for accurately estimating the calculating and memory efforts, that the diode array 24 have five hundred and twelve diodes and be responsive to a spectral range from 200 to 712 nm.

In this manner, approximately five hundred proportions 5im are determinable for each measured wavelength. Therefore, approximately two hundred and fifty thousand individual values, altogether, are determined and memorized. Consequently, five hundred multiplications and five hundred additions are carried out at each measured wavelength Sme Thus, approximately two hundred and fifty thousand multiplications and the same number of additions result altogether. These numbers reach the amount of more than one million with a diode array having one thousand and twenty four diodes. This requires a considerable and nearly prohibitive expense. In practice, however, it is not necessary to determine such a great number of proportion values sjm.

The proportions become very small if the selected wavelength ranges Akj are very small. For this reason, it is useful in practice therefore to select fairly large wavelength ranges hAj. Such a selection is acceptable since in practice, the "stray light function" i.e. sml as function of the wavelength X, does not exhibit distinctive maxima or minima or steep gradients. Consequently, very fine spectral sub-division of the proportion factors 5m is not required.

Therefore, the selected wavelength ranges AXj may be considerably larger than the spectral distance between two diodes. The wavelength ranges Akj may be selected, for example, so that each corresponds to a spectral range of twenty five diodes of the diode array. Then only twenty, instead of five hundred proportion factors 5jm, are obtained for each measured wavelength Bms Furthermore, the same proportion factors Sjm may be provided for several adjacent measured wavelengths Bm which correspond to twenty five adjacent diodes of the diode array 24. In this embodiment, it is necessary to differentiate between only twenty measured wavelengths Bm instead of five hundred, and only four hundred proportion factors sjm are meinorized.

Furthermore, there are extended spectral ranges for each measured wavelength Am in general, the contribution of which spectral ranges to the stray light is negligibly small. If, on average, approximately half of the entire spectral range is ignored, the memory effort is reduced to approximately two hundred proportion factors 51m, and, correspondingly, the calculation effort is reduced. The determination of the proportion factors sjm can be effected in the way described hereinbelow, by way of example, for the measured wavelength fm = 400 nm.

In this example, coloured glass cut-off filters are used which are commercially available from Messrs Schott u. Gen., Mainz, West Germany.

First of all, a filter GG 435 is interposed in the sample light beam 14. This filter substantially completely blocks the radiation at 400 nm but transmits most of the radiation of larger wavelengths. A transmission T of 0 should be measured at the diode of the diode array 24 allocated to the wavelength Bm = 400 nm. Actually, however, a transmission Tmo is measured due to the stray light of longer wavelength.

Next, a filter GG 455 is placed in the common path of rays 12 or 20 of sample and reference light beams 14 and 16 respectively. This filter GG 455 additionally absorbs the wavelengths between approximately 435 nm and 455 nm beside the wavelengths already absorbed by the filter GG 435. Now the transmission Tml is measured at the diode for the measured wavelength Bm = 400 nm. Therefrom, a proportion factor Sml = Tmo - Tm is obtained with the associated wavelength range, AXj extending from 435 nm to 455 nm.

Now the filter GG 455 is replaced by the filter GG 475. This filter GG 475 blocks the rays of the wavelength range k2 from 455 to 475 nm in addition to the previously blocked rays. The transmission Tm2 is measured at the diode for the measured wavelength Xm = 400 nm. This gives the proportion factor 5m2 = Tm - Tm2.

This filtering is continued. Subsequently, the filters GG 495, GG 515... are inserted, until finally with the filter RG 9 only stray light of more than approximately 735 nm is detected.

In this way, fourteen different proportion factors Sim are obtained using the coloured glass cut-off filters available in series from Schott and Gen. If adjacent proportion factors sim differ only slightly, they can be combined as a single value by summation.

It can be assumed that nine proportion factors Sim for Xm = 400 are finally obtained. For example, Aka = 435..... 455 nm, k2 = 455..... 495 nm etc. As in the example described each diode covers a wavelength range of 1 nm width, the wavelength ranges are preferably selected as AX1 = 434.5..... 454.5 nm, AX2 = 454.5 nm ... 494.5 nm, etc.

For correcting the sample spectrum, the arithmetical mean T, is formed of the ten measured transmission values associated with the wavelengths 435 nm, 436 nm ... 454 nm. Correspondingly, the arithmetical mean T2 is formed of the twenty measured transmission value at the wavelengths 455 nm, 456 nm ... 494 nm etc.

Finally, AT = s1T1 + s2T2 + .... +.... sgTg is formed.

This value AT is subtracted from the transmission value Tm measured at the measured wavelength Xm = 400 nm according to equation (11). Therefrom the corrected transmission value Tk at the measured wavelength Xm = 400 is obtained.

This can be accomplished analogously for the other measured wavelengths or groups of measured wavelengths.

The electronic stray light adjustment has been described herein with reference to the use of a double beam instrument. It will be understood that it is also correspondingly applicable to single beam instruments and that the arrangement described is exemplary only.

Claims

1. An arrangement for use with a spectrophotometer for reducing measuring errors due to stray light in the arrangement comprising a memory in which proportion signals are memorized as measured instrument constants of the spectrophotometer, which proportion signals correspond to proportion factors (sum) with which a photometric measured quantity (Tmj) occurring in a wavelength range (AXi) affects, by stray light, the measurement of these measured quantities (Tm) art a measured wavelength (AXm), means for detecting a spectrum to be measured and for memorizing signals corresponding to the values of the measured quantity (Tmi) in the wavelength ranges (aha), means for weighting the memorized proportion factors (sum) in accordance with the values of the measured quantity (Tmi) in the associated wavelength ranges (at1), means for summing the proportions (sjm(Tmj - Tm)) thus weighted for generating a stray light signal (arm) and means for correcting the spectrophotometric measuring signal (Tm) obtained at the measured wavelength (m) for the stray light signal (AT) to generate a corrected measuring signal (Tkm).

2. An arrangement as claimed in claim 1 wherein the photometric measured quantity is transmission or reflectance.

3. An arrangement as claimed in claim 2, wherein the proportion signals correspond to the proportion factors (sum) with which the deviations of the photometric measured quantity (Tmj) occurring in a wavelength range (aha) from the value of these measured quantities (Tm) at a measured wavelength (cm), affect by stray light the measurement of this measured quantity (Tm) at the measured wavelength (m) and the means for weighting are formed by means for multiplying these proportion factors (sum) by the differences (Tmi - Tm) of the measured quantity (Tmi) in the associated wavelength ranges (ski) from the value of the measured quantity (Tm) at the measured wavelength (cm).

4. An arrangement as claimed in any one of claims 1 to 3 for use with a spectrophotometer in the form of a spectrograph having a diode array detector.