Characterisation & Use of Array Spectrometers

Transcription

1 Characterisation & Use of Array Spectrometers Mike Shaw, Optical Technologies & Scientific Computing Team, National Physical Laboratory, Teddington Middlesex, UK 1

2 Overview Basic design and features of array spectrometers. Advantages and disadvantages of array spectrometers over scanning systems. Why are NPL using an array spectrometer? Some key performance characteristics and how to assess them. In system (heterochromatic) stray light: what causes it and how to measure it. Stray light reduction using blocking filters. Future work and conclusions. 2

3 Basic optical layout of an array spectrometer Entrance slit Collimating mirror Diffraction grating Detector array Focussing mirror 3

4 Features of array Spectrometers Input optics: Optical fibre Optical layout Transmissive and reflective diffraction gratings Lenses and mirrors Internal baffles and size of housing Detector array Different photosensitive elements depending upon wavelength range Different no. of array elements and overall size of array Electronics 4

5 Why use an array spectrometer? Usually less accurate than scanning monochromator based instruments. Offer three principal benefits over scanning systems: Speed (multiple wavelengths measured simultaneously). Size (usually much smaller and lighter than scanning systems). Cost (typically much cheaper). 5

6 Why are NPL using an array spectrometer? To develop a new capability to measure the spectral output of light sources. Need to use an array spectrometer to carry out measurements in a reasonable amount of time. Spectral radiant intensity distribution, I(λ) Total spectral radiant flux, Φ( λ) = I( λ) dω 4π 6

7 Some key performance characteristics Wavelength accuracy Spectral resolution Linearity Responsivity Stray light There are also many other important performance parameters including those relating to the detector array itself such as uniformity, well capacity, noise, etc. 7

8 1. Wavelength accuracy Measured using emission lamps or laser lines. Accuracy depends upon calibration method and type of function used to fit pixel no. to wavelength. Output of a Mercury Neon Emission Lamp Measured Using an Array Spectrometer Signal (counts)

9 2. Spectral resolution Resolution depends upon the size of the spectrometer entrance slit & pixel bandwidth dispersion at focal plane x pixel width. May vary across the detector array. Output of a 532 nm Laser Measured Using an Array Spectrometer Signal (counts)

10 3. Linearity Best measured using double aperture method or via comparison with a detector of known linearity (e.g. high quality silicon photodiode or trap). Although single element photodiodes may be extremely linear the same does not necessary hold for detector arrays made using the same photosensitive material. Need to assess linearity with respect to irradiance and exposure time and also look at wavelength dependence. Typical non-linearity errors for silicon CCDs are O(10-3 ) or less for a 2:1 step in irradiance but much larger errors have been observed for other devices. 10

11 4. Responsivity Depends upon many factors including: type of detector array, size of entrance slit, grating efficiency, coupling into spectrograph. Spectral variation in responsivity also effects stray light errors. Responsivity calibration of a linear detector array 11

12 5. In System (heterochromatic) stray light Often the dominant source of uncertainty in measurements made using compact array spectrometers. Scattering, and interreflections from the optics, housing and detector array and inadequate blocking of higher diffracted orders cause rays to strike parts of the detector array corresponding to different wavelengths. 12

13 Stray light signal observed using a laser line Dark Corrected measured signal (normalised to max) 1.E+00 1.E-01 1.E-02 1.E-03 1.E-04 1.E-05 1.E Pixel no. Measured Spectra Ideal spectra Background due to heterochromatic stray light These results could be used to state that stray light rejection is < 10-5 some distance away from the centre wavelength of the laser line. 13

14 Stray light errors are source dependent AE_high-pressure_sodium_lamp.html ojects.html Spectral Total Flux of a Fluorescent Lamp Spectral Total Flux of a High Pressure Sodium Lamp Relative SPD of Four LEDs Spectral Total Flux of a Tungsten Halogen Lamp E Spectral Total Flux (arb. units) Spectral Total Flux (arb. units) Spectral Total Flux (arb. units) 1.00E E E E E+02 LED1 LED2 LED3 LED4 Spectral Total Flux (arb. units) E Stray light errors tend to be most critical when measuring a broadband spectrum with an intensity varying over several orders of magnitude, e.g. a quartz tungsten lamp. 14

15 Spectral Total Flux of a Tungsten Halogen Lamp Stray Light Errors for an incandescent source Spectral Total Flux (arb. units) Relatively low spectral flux at shorter visible and UV wavelengths Relatively high spectral flux at longer visible and NIR wavelengths 15

16 Stray Light Errors for an incandescent source Spectral Total Flux of a Tungsten Halogen Lamp Spectral Total Flux (arb. units) W avelength (nm ) Small fraction of radiation inside the spectrometer is measured as heterochromatic stray light 16

17 Measuring stray light signals from a broadband light source Fibre input to spectrometer Background corrected Signal Measured from a Quartz Tungsten Lamp 1.E+07 Measurement of lamp signal, V lamp (λ) Background corrected signal (counts) 1.E+06 1.E+05 1.E+04 1.E

18 Measuring stray light signals from a broadband light source Measurement through cut on filter, V filter (λ) Fibre input to spectrometer Background corrected Signal Measured from a Quartz Tungsten Lamp Through a GG435 cut on Filter 1.E+07 Transmittance (%) Nominal transmittance of GG435 (3mm thickness) cut on filter 1.E E-01 1.E-02 1.E-03 1.E-04 Background corrected signal (counts) 1.E+06 1.E+05 1.E+04 1.E

19 Measuring stray light signals from a broadband light source Shutter to block light source from spectrometer field of view Fibre input to spectrometer Background Signal 1.E+07 Measurement of background signal, V bg (λ) signal (counts) 1.E+06 1.E+05 1.E+04 1.E

20 Analysis of Stray Light Data For an ideal spectrometer: T filter ( λ) = V V filter lamp ( λ) V ( λ) V bg bg ( λ) ( λ) Stray light signals cause deviations from ideal behaviour and indicate erroneously high filter transmittance at wavelengths shorter than the cut on. Transmittance (%) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Transmittance of GG435 glass filter (3mm thickness) Measured using array spectrometer Nominal Stray light error of > 90%! 20

21 Analysis of stray light data By making some simplifying assumptions, the fractional stray light error when measuring unfiltered source can be approximated as N SL /N Total Estimated stray light signal / total measured signal (%) Signal (counts) Estimated Fractional Stray Light Error in Measurement of the Output of a Quartz Tungsten 100% 80% 60% 40% 20% 0% Lamp using an Array Spectrometer Dark Corrected Signal Measured From a Quartz Tungsten Lamp Through Two Cut on Filters Using an Array Spectrometer Dark corrected signal measured through GG435 Dark corrected signal measured through RG Stray light signal level N SL 21

22 Comparison of Different Array Detectors The cut on filter method provides a way to compare the performance of different array spectrometers for measuring the spectral irradiance from a broadband light source. Transmittance of GG435 measured using a quartz Tungsten lamp and different array detectors 100% Measured transmittance (%) 10% spectrometer A Spectrometer B Spectrometer C Spectrometer E Spectrometer F Spectrometer G Spectrometer H GG435 Nominal 1%

23 How to handle stray light? Live with it Correct for it Reduce it Determine stray light contribution to measurement uncertainty. Spectral Total Flux of a Tungsten Halogen Lamp 0.03 Spectral Total Flux (arb. units) Minimise the effect of stray light by calibrating the detector under conditions as close as possible to those under which it will be used Magnitude of stray light errors is too large for many applications. 23

24 How to handle stray light? Live with it Correct for it Reduce it Input laser radiation at different wavelengths into the array spectrometer to determine amount scattered onto each pixel as a function of wavelength stray light contribution to detector responsivity. Characterise the stray light rejection of the instrument and then correct for it. Dark Corrected measured signal (normalised to max) 1.E+00 Spectrum of HeNe Laser Measured Using Array Spectrometer 1.E-01 1.E-02 1.E-03 1.E-04 1.E-05 1.E Pixel no. S. W. Brown, B. C. Johnson, M. E. Feinholz, M. A. Yarbrough, S. J. Flora, K. R. Lykke, and D. K. Clark, Stray light correction algorithm for spectrographs, Metrologia 40, S81-83 (2003). Y. Zong, S. W. Brown, B. C. Johnson, K. R. Lykke, and Y. Ohno, Simple spectral stray light correction method for array spectroradiometers, Applied Optics, Vol 45 No. 6, 20 Feb

25 How to handle stray light? Live with it Correct for it Reduce it Use additional baffles inside spectrometer to block interreflections difficult to implement and many detectors are sealed. Use stray light blocking filters to limit the wavelengths of light reaching the detector array 25

26 Stray Light Blocking Filters Reduce the bandwidth of radiation reaching the spectrometer using bandpass filters. Spectral Total Flux of a Tungsten Halogen Lamp 0.03 Measure the spectrum over a reduced wavelength range without influence from stray light caused by scattering of other wavelengths. Spectral Total Flux (arb. units) Different blocking filters incorporated into a filter wheel behind the spectrograph entrance slit. 26

27 Stray light blocking filters Target: to reduce the fractional stray light error in the measurement of the irradiance spectrum of a Quartz Tungsten lamp to < 1% over the range 300 nm to 800 nm. Use data from cut on filter measurements to estimate stray light level through theoretical filters. Stray light signal / total measured signal Transmittance (%) 100% 100.0% 90% 80% 70% 10.0% 60% 50% 40% 30% 1.0% 20% 10% Estimated Gaussian Fractional Transmittance Stray Light Profiles Errors Through of Four Four Theoretical Blocking Stray Light Filters Blocking Filters 0% 0.1% filter 1 filter 2 filter 3 filter 4 no filter Filter 1 Filter 2 Filter 3 Filter

28 Choice of stray light blocking filters Look at effect of filter FWHM and CWL on predicted stray light error Gaussian Transmittance Profiles of Four Theoretical Blocking Filters 100% 90% 80% Transmittance (%) 70% 60% 50% 40% 30% 20% 10% Filter 1 Filter 2 Filter 3 Filter 4 0% Theoretical filters with guassian transmittance 28

29 Real blocking filters Measured Transmittance of Four Real Blocking Filter Combinations 100% 90% 80% Transmittance (%) 70% 60% 50% 40% 30% 20% 10% T Filter 1 T Filter 2 T Filter 3 T Filter 4 0% Real filter combinations 29

30 Real blocking filters Estimated Fractional Stray Light Error Through Four Stray Light Blocking Filters 100.0% Stray light / total measured signal (%) 10.0% 1.0% No filter filter 1 filter 2 filter 3 filter 4 0.1% Estimated stray light signals using real filters 30

31 Stray light tests using blocking filters Transmittance of GG435 (3mm) Measured Without Blocking Filters 100% 90% 80% 70% Transmittance (%) 60% 50% 40% 30% 20% 10% 0% Nominal No blocking filter -10%

32 Stray light tests using blocking filters Significant reduction in 100% stray light signals 90% at short wavelengths. Transmittance (%) 80% 70% 60% 50% 40% 30% 20% Transmittance of GG435 (3mm) Measured With Blocking Filters Small discrepancy between nominal & measured filter transmittance at longer wavelengths. Filt1 Filt2 Filt3 10% 0% Filt4 Nominal -10% Increased noise at shorter wavelengths. 32

33 Limitations of using stray light blocking filters Increased measurement time. If using N blocking filters in a filter wheel then N different exposures are necessary + time to move filter wheel. Once filters have been selected and tested for suitability it may be possible to incorporate them onto the detector array allowing the spectrum to be acquired in a single exposure. Adding an additional element into the detector may result in additional interreflections and stray light. Slightly reduced detector sensitivity (not significant if filters are well chosen). Temperature effects (need to be aware of temperature sensitivity of filter transmittance). 33

34 Future work Development & increased automation of goniospectroradiometer facility. Validation of array spectroradiometer for measuring different source types. Intercomparison with existing integrating sphere facilities. 34

35 Conclusions Array spectrometers are extremely useful instruments, however they can suffer from significant errors if not well understood and characterised. In particular, stray light errors from array spectrometers may be several orders of magnitude larger than from traditional scanning double monochromator systems. NPL have modified an array spectrometer to incorporate a series of custom designed stray light blocking filters, with the effect that stray light errors have been reduced by 1-2 orders of magnitude for measurements of the irradiance from a quartz tungsten lamp. The array spectrometer has been integrated into a new goniospectroradiometric instrument for measurement of the radiant intensity distribution and total spectral flux of a range of different light sources. 35

36 Acknowledgements Thanks to colleagues in the optical technologies and scientific computing team at NPL and Teresa Goodman in particular for her help and advice. 36

37 Questions? Mike Shaw, Optical Technologies & Scientific Computing Team, National Physical Laboratory, Teddington, Middlesex, UK Tel