Infrared filters and dichroics for the advanced along-track scanning radiometer

Size: px

Start display at page:

Download "Infrared filters and dichroics for the advanced along-track scanning radiometer"

Gerard Phelps
6 years ago
Views:

1 Infrared filters and dichroics for the advanced along-track scanning radiometer Roger Hunneman and Gary Hawkins The design and manufacture of the band-defining filters and their associated dichroic beam splitter for the 11- and the 12- m infrared channels of the advanced along-track scanning radiometer are described. The filter requirements that have led to the choice of coating designs, coating materials, disposition of coatings, and effects of polarization are discussed. Overall spectral throughputs of the filter and dichroic interaction for the two channels are also presented Optical Society of America Key words: Infrared filters, dichroic beam splitter, polarization, spectral throughput. The authors are with the Infrared Multilayer Laboratory, Department of Cybernetics, The University of Reading, Whiteknights, Reading, Berkshire, RG6 2AY, England. Received 22 November 1995; revised manuscript received 3 May $ Optical Society of America 1. Introduction The advanced along-track scanning radiometer AATSR instrument is a seven-channel visible infrared radiometer that is due to be launched on the European Space Agency satellite, ENVISAT-1. The instrument is an improved version of the successful ATSR flown on the European Space Agency s first remote sensing satellite, ERS-1, in 1991 and its successor, ATSR-2, launched in April 1995 on ERS-2. The instrument views the same area of sea through both a near-vertical nadir atmospheric path and through an inclined path of different length some distance along the satellite track, permitting accurate determination of the atmospheric correction. The infrared channels described here are used for the measurement of global sea-surface temperatures 1 averaged over a 50 km s 50 km area with an accuracy of 0.25 C between adjacent 1 km 1 km areas. The instrument builds up images of the sea surface pixel by pixel by means of a scanning telescope that focuses radiation onto a field stop in the focal-plane assembly. 2 The thermal infrared channels that are the subject of this paper are centered at approximately 11 and 12 m. The disposition of the coatings in the 11- and 12- m channels is novel insofar that the dichroic beam splitter that splits the 11- and 12- m channels simultaneously defines both the entirety of the transmitted 11- m band and the short-wavelength side of the 12- m band. This technique was originally developed by us for the ATSR and ATSR-2 filters and beam splitter to minimize the effects of polarization from the beam splitter, as discussed below. It is published here for the first time. Unknown to the present authors, a tilted bandpass filter was used in the advanced veryhigh-resolution radiometer as a dichroic, 3 presumably to reduce the number of coated surfaces, there is no mention in this reference to this approach being adopted to reduce polarization sensitivity. Also, the designers of the advanced very-high-resolution radiometer did not use a further bandpass filter in the 11- m transmitted channel to prevent stray radiation reaching the detector, as has been done in this work. 2. AATSR Infrared Focal-Plane-Array Layout The AATSR s focal-plane assembly shown in Fig. 1 is similar to that of the ATSR, as described by Tinkler et al. 4 ; the whole of this baseplate assembly is mounted in a housing that provides thermal isolation and can be evacuated when sealed with a ZnSe window for ground testing. To minimize detector noise and increase sensitivity, the baseplate assembly, including all the optics, is cooled to a temperature of 90 K by a Stirling cycle cooler. Beam splitters BS1 and BS2 operate in conical illumination of half-angle 5.5 incident at 30. The filters are required to operate in a greater converging illumination of half-angle 22, with the center of the beam being normally incident. Spectral requirements specify that the filters should have a minimum of distortion to the bandpass profile because of the highly conical illumination and the beam splitters 5524 APPLIED OPTICS Vol. 35, No October 1996

2 beam splitter BS2. The earlier ATSR and ATSR-2 did not have this coating. Fig. 1. AATSR focal-plane baseplate assembly. FPA, focalplane array. should exhibit minimal sensitivity T s p 2% at a given wavelength to any polarization present from the incoming signal. The out-of-band radiation should be blocked to at least the 10 4 rejection level from the Ge substrate absorption edge to 20 m. 3. Disposition of Coatings for the 11- and 12- m Channels To provide optimal performance and compliance with the specification, considering the precision of spectral placement required between the 11- and 12- m filter profiles 0.8%, the use of a bandpass filter was chosen as the dichroic BS2 instead of the more normally used edge filter. This choice was made first because the spectral placement of bandpass filters is generally more precise than that of edge filters, and second for the reduction of polarization effects both these points are discussed below. The 11- m banddefining filter was placed on the front surface of the BS2 dichroic, transmitting this channel through the beam splitter and reflecting the 12- m band with the long-wave reflection zone of this filter. The banddefining filter for the reflected 12- m channel was placed on the 12- m detector package window. This filter by itself is able to block to 20 m; its rear surface carries a filter that supplies the continuous short-wavelength blocking. For the 11- m channel, a bandpass filter of slightly greater width than the band-defining filter on the beam splitter was deposited on the detector package window. This filter, together with the continuous short-wavelength blocking on its rear surface, ensures that the detector, like that of 12- m channel, receives only in-band radiation, any scattered out-of-band radiation being blocked. The 11- m band-defining filter deposited on the front surface of BS2 is of the same design and materials as those of the 12- m filter, ensuring the same bandwidth, but because it was placed at the shorter wavelength it needed some assistance with the blocking to 20 m. To achieve this, a long-wave blocking filter was deposited on the rear surface of 4. Choice of Coating Materials and Designs In view of the strongly converging illumination on the bandpass filters and the use of the tilted beam splitter in converging light, it was desirable to use materials with the highest values of refractive index. This reduces the sensitivity of the filter profiles to wavelength variations with angle of incidence, as the change in effective thickness of each layer is much less for high-index layers than for low-index layers in the multilayer. As our laboratory has considerable expertise in the use of PbTe n 5.8 at 90 K, this was used in combination with ZnS n 2.15 and ZnSe n Furthermore, the bandpass filters were designed with cavities of PbTe, thus ensuring the highest effective index n* n* H n L when used with ZnSe or ZnS for the best minimization of angle of incidence effects. All three bandpass filters are of the same design. The slightly wider bandwidth of the 11- m detector package window bandpass filter over that of the band-defining filter deposited on BS2 was achieved when the low-index material used in the common design was changed from ZnS to ZnSe, thus reducing the index contrast and hence increasing the bandwidth. The use of the PbTe and ZnSe as coating materials in filter designs also has the advantage of minimizing polarization effects when compared with other material combinations for the reasons discussed above. 5. Filter Design The generic design used for all these coatings was a 13-layer triple half-wave filter comprising three cavities, each of one half-wavelength optical thickness at 0 separated by two reflectors of three layers each. The outer reflectors were the Ge substrate in combination with one layer and, at the air end, a reflector and antireflection layer of three layers. The formal statement of design was sub LHHLHLHHLHLHHLHL 1.0. This filter design has a square profile with a HBW of 10%; it is a good example of the minimal number of layers needed to achieve an acceptable spectral performance by the use of the high-index contrast available with PbTe in combination with ZnS. The continuous blocking to the short-wave side of the passband is provided by a 41-layer multilayer comprising two subsidiary Herpin stacks plus a 17- layer Chebyshev stack; the whole of this stack and parts of the others are optimized for in-band flatness. Full advantage was taken of another unique property of PbTe, viz, the movement to long wavelength of its short-wave semiconductor absorption edge from the 3.5- m region to 5.8 m when cooled to 90 K, thus reducing the number of layers needed for blocking over that needed with other material combinations. This short-wave blocking system was deposited on the rear of both the 12- m band-defining filter and the 11- m stray rejection filter. A 13-layer PbTe ZnSe Chebyshev stack optimized for flatness in band was deposited on the rear face of the dichroic. 1 October 1996 Vol. 35, No. 28 APPLIED OPTICS 5525

3 Fig. 2. a Nonpolarizing high-pass edge filter design after Seeley 7, b nonpolarizing high-pass edge filter design after optimization of layer thicknesses, c optimized Chebyshev high-pass edge filter design for comparison. This stack not only provided the necessary blocking to 20 m for the 11- m channel but also considerably reinforced the short-wave PbTe absorption blocking in the system by virtue of its thicker layers. 6. Filter Manufacture The filters were deposited in a Balzers 510 vacuum deposition machine fitted with a cryopump. The deposition materials were evaporated from resistance-heated sources mounted on a rotating slipring assembly. This arrangement allows the deposition of uniform thickness layers while accurate temperature control of the stationary filters mounted on a heated block in the upper part of the chamber can be maintained. The monitor and the filter substrates were thermally clamped into the jigwork attached to the block by the use of lead annular washers, backing pieces, and disk springs. Optical thickness monitoring was employed throughout the manufacture of the filters with various fixed wavelengths in the region m. Fractional layers, where necessary in the blocking multilayers, were deposited by the use of a spreadsheet-implemented algorithm to determine optical thickness from reflectivity levels. A shutter that isolated the filter pieces from the centrally mounted monitor was used to ensure that the computed thickness was deposited on the filters. The deposition took place at 185 C in a residual vacuum of 10 6 Torr, with O 2 added to Torr to ensure optimum optical properties for the PbTe. 5,6 7. Polarization The requirement to minimize the effects of polarization from beam splitter BS2 results from the rotating scan mirror s introducing a small amount of polarization into the sea-surface reflectivity; this polarization will then rotate as the scan mirror moves. If one plane of polarization is transmitted through the focalplane assembly more than the other, the signal will contain variations in the mirror scanning frequency, leading to errors in the retrieved brightness temperature. Seeley 7 made a study of possible alternative nonpolarizing edge filters designs by using the above materials. The transmission spectra of these designs show minimal polarization at the edge, but significant and increasing polarization as one moves away from the edge into the high-transmission passband region Fig. 2 a. This problem with the prototype design can be easily overcome by the optimization of the layer thicknesses, which renders the design less polarizing across the passband without significantly changing the minimal polarization situation at the edge Fig. 2 b. Unfortunately these designs were unrealizable because of our inability to deposit the necessary ZnTe n 3.1 outer layer with adequate low loss. The more conventional design of a high-pass edge filter by the use of an optimised Chebyshev equal-ripple polynomial with PbTe and ZnSe layer materials was investigated for use as the beam splitter. This design allows the best fit for both the edge steepness and the minimum ripple in the passband at normal incidence, but, as would be expected, the two polarizations split at the edge when used in oblique illumination to an unacceptable degree for use in this case Fig. 2 c. Bandpass filters that utilize these coating materials exhibit only small amounts of polarization, showing a split between the two polarizations in the case under discussion here of 2% or less on the edges of the passband profile. Further, given the accuracy of our deposition-process thickness monitoring, there are difficulties in achieving the specified spectral place APPLIED OPTICS Vol. 35, No October 1996

4 Fig. 3. AATSR 11- m filter showing computed performance at normal incidence in parallel light and in a cone of 22 half-angle, showing shift and polarization split. Fig. 5. AATSR combined filter dichroic beam splitter BS2 system response. Spectral measurements include the effects of polarization. ment of an edge filter that contains fractional quarter-wave layer thicknesses, but the accurate placement of a bandpass filter with integral quarterwave layer thicknesses is relatively straightforward. Accuracy of spectral placement is typically 1.5% for edge filters, compared with 0.3% for bandpass filters. This, coupled with the intrinsically good angle of incidence performance and the small sensitivity of the H-spaced bandpass design to polarization, led us to select the bandpass rather than the edge filter approach for the design and implementation of BS2. For example, Fig. 3 shows the computed profile of the 11- m filter both at normal incidence in parallel illumination and when in conical illumination of halfangle 22. This demonstrates the minor nature of the effects of conical illumination on the filter profile and the small polarization under these circumstances. 8. Overall Optical Train Throughputs for the 11- and 12- m Channels The resulting optical throughputs by calculation and for comparison by measurement for both the s and the p polarizations of the manufactured AATSR flight beam splitter BS2 and filter combinations are shown in Figs. 4 and 5, respectively. Measurements are performed at 90 K at the correct angles of incidence in a spectrophotometer cone angle of f 5.6. As can be seen, there is reasonable agreement between the two cases, the main deviation being caused by the slope on the top of the 11- m filter, which was probably due to a small mismatch in cavity thicknesses. The overall 11- m channel performance is lower than that of the 12- m channel because of the product of a 87% BS2 band-defining filter transmission, and the 11- m detector window filter at 74% is less than the product of 87% from a nearly 100% BS2 reflectance and a 87% 12- m band-defining filter transmission. However, both channels are of the correct bandwidth and overlap at the specified wavelength. 9. Conclusions We have demonstrated the use of a bandpass filter for simultaneously defining a spectral band and performing the function of a dichroic beam splitter with a minimum of polarization and angle of incidence effects. The change in transmission for the two different polarization directions for each component is less than 2%, giving an estimated error in the retrieved brightness temperature of less than 0.01 K. By exploiting the small but significant difference in refractive index between ZnS and ZnSe, we have shown an efficient way of using the same design of bandpass filter to perform the three functions dichroic mirror, band definition, and stray light rejection crucial for the separation, defining, and spectral integrity of the 11- and the 12- m channels. Additionally it should be pointed out that the 12- m band-defining filter and the 11- m stray rejection filter can also perform the function of antireflected detector windows. Fig. 4. Filter dichroic beam splitter BS2 interaction. Calculation is inclusive of absorption and shows polarization effects assuming ideal spectral placement. References 1. T. Edwards, R. Browning, J. Delderfield, D. J. Lee, K. A. Lidiard, R. S. Milborrow, P. H. McPherson, S. C. Peskett, G. M. Toplis, H. S. Taylor, I. Mason, G. Mason, A. Smith, and S. Stringer, The along track scanning radiometer measurement of sea-surface temperature from ERS-1, J. Br. Interplanet. Soc. 43, J. Delderfield, D. T. Llewellyn-Jones, R. Bernard, Y. de Javel, E. J. Williamson, I. Mason, D. R. Pick, and I. J. Barton, The 1 October 1996 Vol. 35, No. 28 APPLIED OPTICS 5527

5 along track scanning radiometer ATSR for ERS-1, in Instrumentation for Optical Remote Sensing from Space, J. W. Lear, M. Monfils, S. L. Russak, and J. S. Seeley, eds., Proc. SPIE 589, AVHRR, advanced very high resolution radiometer, technical description, NASA contract NAS ITT Aerospace Optical Division, D. Tinkler, D. R. Pick, S. J. Stringer, and C. G. Woods, The design of the focal plane assembly for the along track scanning radiometer, in Instrumentation for Optical Remote Sensing from Space, J. W. Lear, M. Monfils, S. L. Russak, and J. S. Seeley, eds., Proc. SPIE 589, C. S. Evans, R. Hunneman, J. S. Seeley, and A. Whatley, Filters for the 2 band of CO 2 ; monitoring and control of layer deposition, Appl. Opt. 15, C. S. Evans, R. Hunneman, and J. S. Seeley, Optical thickness changes in freshly deposited layers of lead telluride, J. Phys. D 9, J. S. Seeley, Simple nonpolarizing high pass filter, Appl. Opt. 24, APPLIED OPTICS Vol. 35, No October 1996

Filters for Dual Band Infrared Imagers

Filters for Dual Band Infrared Imagers Thomas D. Rahmlow, Jr.* a, Jeanne E. Lazo-Wasem a, Scott Wilkinson b, and Flemming Tinker c a Rugate Technologies, Inc., 353 Christian Street, Oxford, CT 6478; b