Qualitative and quantitative analyses of the effect of unwanted stray radiation in optical systems are discussed. The basic power transfer equation is written as the product of three factors: the surface BRDF, the projected solid angle, and the power on the source objects. The minimization of any one of these factors along the most significant paths of scatter will reduce unwanted radiation reaching the image. BRDF measurements and their use in selecting suitable coatings are emphasized, along with the effect of altering the baffle configurations, edge scatter, and tolerancing requirements. Optical design aspects, such as stops, field stops, Lyot stops, and obscurations, are discussed in suppressing stray radiation for various systems. The performance and testing of scaled models is related to the expected system performance. The major software tools, GUERAP and APART stray radiation analysis programs, are discussed. The approach presented is to develop qualitative concepts that can be easily grasped by the optical and mechanical designers. The approach should help stimulate stray radiation rejection ideas that can be incorporated during initial design evaluations.

The presence of stray light is a continuing problem in the design and performance of optical systems. In well-designed systems most of it comes from the optical components themselves. By proper baffling it can be reduced in sensitive areas by orders of magnitude, but ultimately it is necessary to improve the components themselves. The dominant source of component scattering is from the optical surfaces. Bulk scattering can also occur in windows or lenses but it is typically one to two orders of magnitude lower than surface scat-tering. In the visible and ultraviolet regions of the spectrum scattering from surface microirregularities only a few nanometers in height dominates. At longer wavelengths micro-irregularity scattering drops exponentially and defect scattering from surface blemishes, scratches, dust, etc. becomes important. It is often nearly independent of wavelength, and forward scattering is typically somewhat larger than backscattering, as would be predicted from dipole scattering theory. Pitting of the surface by sand or rain erosion increases the importance of defect scattering. More complicated than the case of individual incoherent scattering centers is that of correlated scattering from a repetitive surface. Diamond-point turned optical surfaces are an example. Resonances can occur for such surfaces, particularly when they are overcoated with dielectric films. In the best cases, however, scattered light levels from diamond-turned optical surfaces are remarkably low, and the absence of polishing defects or scratches makes this technique very attractive for producing infrared optics. Scattered light resonances can also occur in the ultraviolet. If aluminum or silver coatings are used, optical excitation of surface plasmons may occur. If they decay radiatively, the result is scattered light. Dielectric overcoatings can enhance scattering in this case; increases of as much as an order of magnitude have been observed. These effects are reduced by using surfaces having very low rms microirregularities. We conclude that when ordering optical components for low scatter applications, one should include a specification of the rms height of microirregularities (often termed the rms roughness of the surface) in addition to or even instead of the traditional scratch/dig specification. The rms roughness specification is particularly important for optics to be used in the visible and ultraviolet regions of the spectrum. In the infrared region defect scattering is of primary importance, and it is here that the scratch/dig specification is most useful.

In this paper, three different surface theories are discussed that give the engineer wide flexibility in the types of surface reflection problems that can be analyzed. The first is a rigorous theory of surface reflection (based upon Maxwell's boundary conditions) that accurately treats reflection from flat surfaces with roughness that is small compared to wavelength. The second is a model similar to the Hopkins-Marechal approach to optical tolerancing that can be used to treat small angle scatter from various optical forms. The third model is one that correctly accounts for off-specular reflection peaks that sometimes occur on baffle type surfaces. The utility of these three models as well as their limitations to stray-light calculations will be discussed.

A scalar theory of surface scattering phenomena has been formulated by utilizing the same Fourier techniques that have proven so successful in the area of image formation. An analytical expression has been obtained for a surface transfer function which relates the surface micro-roughness to the scattered distribution of radiation from that surface. The existence of such a transfer function implies a shift-invariant scattering function which does not change shape with the angle of the incident beam. This is a rather significant development which has profound implications regarding the quantity of data required to completely characterize the scattering properties of a surface. This theory also provides a straight-forward solution to the inverse scattering problem (i.e., determining surface characteristics from scattered light measurements) and results in a simple method of predicting the wave length dependence of the scattered light distribution. Both theoretical and experimental results will be presented along with a discussion of the capabilities and limitations of this treatment of surface scatter phenomena.

Forward and backward scattering data on 20 infrared radiation transparent materials is presented. The samples include IRTRANs 1, 2, and 4 and other selected infrared radiation materials including Aℓ₂0₃, As₂S₃, CaF₂, CdTe, GaAs, Ge and PVT ZnSe. The measurements were made at wavelengths of 0.6328, 1.6, 3.39, and 10.6 μm . The angular distribution of the scattering was measured from 0. 5° to 70°.

In a typical infrared optical scanning system, the surface temperature of the detector is usually much less than the ambient. The narcissus effect occurs when the detector, through unintended reflections off internal lens surfaces, sees sources at temperatures other than the background ambient. These sources usually are reflections of the detector itself, hence the term narcissus. The resulting display shows abrupt changes in the background radiation as a function of field position. A technique that evaluates the narcissus effect in terms of narcissus equivalent temperature difference (NARCAT) has been developed and programmed for the computer. Computer results for several infrared optical scanning systems indicate that predicted and measured values of NARCAT agree well. The theory is based on the principle that radiance along any geometrical ray bundle does not vary. 1 The narcissus-induced variation of this radiance is found to be the arithmetic product of three factors, viz., (1) image-forming, (2) surface spectral reflec-tance, and (3) temperature difference. Therefore, to suppress the narcissus effect, one simply minimizes these factors. Examples illustrating both the computational approach and the suppression techniques are given.

The basic philosophy for the selection of aperture configurations, referred to as aperture shaping, for the azimuthal manipulation of the diffracted energy is formulated. Aperture shaping is implemented through the modification of the entrance aperture by the introduction of opaque obstructions of various forms. The basic "building blocks" for the construction of the aperture consist of triangles, rectangles, and annular rings. An analytical expression is derived for the point spread function with the limitations of Fraunhofer diffraction. Computer software has been developed and is operational for the detailed analysis of any aperture that may be so constructed. Three types of aperture configurations pertinent to the stray light rejection problem are examined. (1) An orthogonal supporting spider, (2) a venetian type baffle, and (3) a ring shaped baffle.

This paper presents the theory of operation and the operating characteristics of the GUERAP II computer program. In this program, a deterministic approach to the computation of the stray radiation terms is taken, rather than a statistical approach. Thus, each of the stray radiation terms arising from the phenomena of diffuse scattering from surfaces and diffraction occurring from edges is computed in a deterministic rigorous fashion. The program utilizes a generalized differential ray trace routine, which permits the tracing of rays through an optical system along with a measure of the ray weight, as determined by the area of the ray. The program also utilizes a scalar diffraction theory that permits the description of diffraction by the use of rays originating from edges. This method allows the rigorous handling of multiple diffraction sequences, such as those that occur in systems that utilize a Lyot stop for stray radiation reduction. The paper will contain a discussion of the overall structure and use of GUERAP II, and a description of the diffraction calculation methodology.

The Monte Carlo approach to the solution of radiant energy transfer among surfaces of a system represents radiant energy flux statistically as rays which can be traced through the system according to a set of probability distribution functions. Although the ray repre-sentation of the radiant energy is similar to that of a photon characterization, it is treated in this method as merely a statistical representation. Previous computer programs of this type have experienced considerable problems in achieving attenuation prediction accuracy, with acceptable confidence, in a reasonable computer run time. Traditionally, problems have arisen in the areas of excessive computer time requirements, poor statistical convergence, biased results, and related problems. Advanced Monte Carlo variance reduction techniques were applied in the development of the GUERAP III program which have alleviated problems in this area. The techniques of ray-splitting, expected value, and importance sampling have led to very significant reductions in the required GUERAP run times.

The APART code (Arizona's Paraxial Analysis of Radiation Transfer) is a deterministic stray radiation analysis program capable of yielding quantitative descriptions of systems along with insight into the scattering mechanism present. APART uses y-y geometrical optics to image primarily rotationally symmetrical sys-tems. APART provides a sectional power map of the internal surfaces of a system and identifies "critical" objects seen from the image. Vane structures are modeled by configuration factors. Once the geometrical configuration factor between the internal objects has been calculated and stored, nonstructural changes to the system can be analyzed without re-running the complete program.

The GUERAP III Monte Carlo simulation method traces the paths of statistical samples (rays) of energy flux as they interact with the surfaces of an optical system. To determine ray path interceptions, the surfaces are represented by analytical equations that define the surfaces and their boundaries. The result of each surface interaction is determined by probability distributions characterizing the optical properties of each surface. Property models include emission, absorption, reflection, and transmission (which includes refraction and diffraction). The semiempirical diffraction model assumes an angular probability density function of the distance of the ray from the edge of the aperture. The general surface equation and radiation model utilized permits analysis of complex visible and infrared systems with reflective and/or refractive elements. Although the program was written primarily for off-axis rejection analyses, it can also be used as a general on-axis ray tracing program for determination of such things as image quality and optical efficiency, including in a very natural way the effects of field of view, aberrations, obscurations, vignetting, diffraction, etc. In the off-axis mode the program is used for analySis of system sensitivity to scattering and diffraction, determination of the effectiveness of exterior and interior baffling mechanizations, and the identification of unwanted multipath ghost images.

The Viking Lander Cameras now operating on Mars are carefully designed to minimize stray light. Many details of the camera that affect stray light are discussed. A source of ghost images that appear to be in the sky is described.

One method of analyzing the effectiveness of a reimaging optical system in rejecting unwanted radiation from out-of-field sources is described. The rejection ratio, the parameter used to characterize rejection performance, is defined for point and extended sources. The manner in which the scattering rejection ratio is evaluated for each major scattering source by means of a succession of manual and computer operations is outlined. The diffraction process is described, along with the method of evaluating diffraction rejection ratio. The method of obtaining the rejection ratio for an extended source by summing all of the point source rejection ratios is formulated for the case where the earth is the extended source.

A large-scale optical system with a primary mirror, 82" diameter, is presently being constructed in order to make scattering measurements from infrared to visible wavelengths on large objects. The necessity of having a spherical primary mirror and finite field of view lead to a three-mirror design. The need for low mirror backscatter required separate transmitting and receiving optics with only the primary common to both systems. The background behind the target is in the field of view; control of the resulting backscatter is discussed. Scatter from the room walls and objects in the room is analyzed for various test conditions. BRDF measurements were made on coated and uncoated Cervit samples in an effort to validate the predicted primary mirror scattering.

This presentation suggests a model for flare in microdensitometers based on scattering at the air-glass interfaces of the instrument resulting in several quasi-lambertian sources directly in the sensor optical train. It is further shown that a sample trace is not only a convolution of the target with the projected sensor aperture but, to some extent, a convolution of the target with the irradiated area in the sample plane as well. Based on this analysis, it is suggested that flare can be corrected for provided its magnitude is held consant. This requires that, during a full scan, the target always remains effectively within the irradiated area. This presentation is extracted from a substantially larger paper to appear in "Optical Engineering" at a later date this year.

Light suppression in optical instruments has improved significantly in the last several years. These improvements can be attributed directly to three previous developments: optical computer programs, low reflecting surfaces and more efficient baffle edges. In 1968 a new electro-chemical process was developed, which with subsequent refinement, exhibited a reflectance of about 0.57 over the visible (Figure 1) and infrared spectrum (8-14 μm). Several years later a method was devised to fabricate baffle edges which when subjected to the electro-chemical process, produced edges 8 to 10 times more efficient (Figure 2) than the previous state-of-the-art. With refinement in the computer programs and experimentally obtained data on surfaces and edges, light shades were designed and tested which exhibited an improvement over previous light shades by an order of magnitude. This combination of computer program and physical properties improvements has enabled us to improve instruments performance significantly by reducing the largest source of noise, i.e., stray light.

The angular response of LWIR telescopes is measured to verify predicted off-axis rejection (OAR) performance. The data are also useful to guide any necessary modifications for performance improvement. The off-axis response must be 9 to 12 bels (orders of magnitude) below the on-axis maximum response to reject the influence of strong off-axis sources. Collimated lasers serve as the strong, highly directional test sources required in the characterization measurements. Carefully used optical attenuators permit the application of detectors which can normally cover only 4 or 5 bels. This paper describes the apparatus and methods employed in measuring the OAR of specific telescopes at wavelengths of 1. 06 and 10. 6 4m. Illustrative data are given for telescopes built by Honeywell under the sponsorship of Air Force Geophysical Laboratories.

Initial results are presented of a program designed to experimentally verify the APART (Analysis Pro-gram Arizona Radiation Trace) stray radiation analysis computer program. This computer program was developed by the University of Arizona under contract to the National Aeronautics and Space Administration (NASA) and in support of NASA's Space Telescope Project. New experimental techniques were developed to measure the stray radiation rejection performance of a fifty centimeter aperture telescope model. The resulting experimental data is compared to computed data generated by an analysis of the telescope model utilizing the APART stray radiation analysis computer program.

The presence of extraneous light sources has often represented a problem to the reli-able operation of stellar sensors. The Bendix Star Sensor Assembly (SSA) is designed to accurately detect faint stars (+4 visual magnitude) with the existence of bright objects (sun, moon, etc.) near the field of view. A measurement was needed of the SS's off-axis light attenuation when operating as close as 40° to the sun. This was accomplished through careful control of a 16 by 10 meter solar-thermal-vacuum chamber. Background scattered light was minimized and mapped enabling accurate photometric testing with the SSA under going simulated solar impingement. The test results showed that the SSA will provide a minimum system attenuation of 10¹³:1 when operating at 40° to the sun. This paper sum marizes the test concepts and techniques formulated and the problems resolved in proving that a) such a test was indeed feasible and would reliably represent an accurate measure of SSA performance and b) the SSA did, in fact, perform to its requirement.

Veiling glare results from the scattering of light by optical system components. In this paper we present some experimental data on veiling glare and a theoretical approach to the understanding of veiling glare. We also outline a simulation method that may be used to correct photometric measurements for glare.

The panel discussed termonlogy briefly. The name "Bidirectional Reflectance Function" (BDRF) is a tongue-twister, but seems to be well established and no one has a better suggestion, so it will stick. It was pointed out that BDRF is a quantity that depends on four angles, but is usually quoted from only a few data points, so a related term such as "spread function" or "average BDRF" might be better.

In the optical design of most sunshields, baffles are used to prevent stray light reflected by the interior surfaces of the sunshield from reaching the optical system entrance pupil directly. This improves the performance of the sunshield. However, it is usually impossible to eliminate all light reflected by the edges of the baffles themselves; thus these edges must be designed to reflect the minimum light possible. Martin Marietta's design for baffle edges, using the black surface reflected five to eight times less light than edges of true razor blades, the previous standard for baffle edges. The graph compares the light reflected from the baf-fle edges and from razor edges as a function of the angle of reflection. No baffle edges known to date have exceeded the performance of razor edges.