On-line industrial inspection of batch manufactured parts requires fast measurement techniques for surface finish quality. In order to develop the measurement basis for these techniques, a system has been built to determine surface roughness by measuring the angular distributions of scattered light. The system incorporates data gathered from the angular distribution instrument and traditional surface stylus instruments. These data are used both as input and as comparison data in order to test various mathematical models of optical scattering phenomena. The object is to develop a mathematical model that uses the angular distribution of scattered light to deduce surface roughness parameters such as Ra and surface wavelength. This paper describes the results of an experiment in which angular scattered data from surfaces with sinusoidal profiles was used to compute the surface R and wavelength. Stylus measurements of these parameters were made separately. A comparative table is given of the computed and measured values. Estimates of uncertainties are also given.

The development of high fluence laser diagnostics and other methods has resulted in a wealth of data on the surface and optical properties of diamond-turned metal mirrors. From this data, it has been determined that the diamond-turned surface exhibits unique proper-ties which are related empirically to the machining conditions. Among these are optical reflectance changes and variations in the laser damage thresholds of melt, slip, and selec-tive area damage. The plastic deformation of the surface during diamond machining results in a higher concentration of crystalline defects in the near-surface region that contrib-utes to the reflection process. This paper presents the results of an investigation into the nature of the deformation introduced by the diamond-machining process. In particular, the structure of the surface region is elucidated by high resolution transmission electron microscopy. In addition, the correlation of the various observed defects to the physical properties of the surface will be discussed.

The nature of surface errors on diamond-turned optics is frequently machine specific. The diamond tool may have a slightly irregular cutting edge. In addition to the obvious tooling grooves associated with the feed rate, the cutting operation may be beset with subtle oscillations or resonances that affect the quality of the surface. Such errors are typically quasiperiodic in nature and will be so treated in this study. Quasi-periodic surface errors will be modeled where the height of the irregularities is assumed to be a small fraction of the incident wavelength. First order perturbation scattering theory will be used to predict the effect of such errors on the performance of optical systems.

This paper reviews the development of the analysis of surface finish from the early attempts in the nineteen thirties to the present day. The development of parameters is put in context with the techniques available and it is demonstrated that characterization based on graphical and experimental convenience has influenced industrial practice. As the requirements of manufacture and functional performance have been stretched by advancing technology many industrialists have been forced to accept that existing practices are limited, and have sought other descriptions still based on well established techniques; but these existing techniques are limited, their suitability to inprocess measurement practically non existent. It is shown that recently attempts have been made to develop optical methods of assessing surface finish using traditional parameters such as Ra. This paper suggests that it may be time to look towards a new form of specification, more suited to assessment by optical transducers, and some methods of assessment are proposed. To support this view a simple low cost device is discussed which can be calibrated to give Ra but which also presents information in a more relevant empirical way which may be more valid than the existing parameter specification.

We define an objective method to classify defects based on their visibility : If a sample is illuminated against a black background, the observer sees only the defects. Now we replace the black background by an other with finite luminance. Some defects will become invisible, others will still be seen. If we change the illumination on the sample, the threshold between "seen and not seen" defects will be changed. With one standard defect and the la,' of gradation of the object luminance, we can classify all macroscopic surface defects.

Laser damage to optical surfaces, particularly coated surfaces, is typically initiated by material defects which couple strongly to the laser radiation. Knowledge of damage-related defect characteristics is therefore essential to optical material development efforts and to quality control. Such characteristics include individual defect failure levels, defect densities, and content of optically absorbing impurities. Defect failure distributions containing defect density information can be inferred from properly designed laser damage measurements. The data also permit distinction between different damage mechanisms, including nondefect-related mechanisms involving the substrate. Inclusions containing volatile, optically absorbing impurities can be detected by mass analysis of laser-desorbed vapors. These methods will be described, and examples of results will be presented.

Defects responsible for laser damage in visible-wavelength mirrors are observed at nondamaging intensities using a new video microscope system. Studies suggest that a defect scattering phenomenon combined with lag characteristics of video cameras makes this possible. Properties of the video-imaged light are described for multilayer dielectric coatings and diamond-turned metals.

The development of a formal standard test method for the surface microroughness characterization of optical components is outlined. The results of an interlaboratory comparison of the test method are presented to illustrate the advantages and limitations of this approach to measure effective rms surface roughness. The lab-to-lab variations in measured total integrated scatter (TIS), effective roughness, were found to be about +15%. The within-lab variations were about ±5%.

The increasing requirement for stringent standards relating to surface imperfections on certain optical components has highlighted the need for objective methods of measuring such defects. In recent years, several Standards Committees and many individual companies have developed their own series of standard defects which are compared visually with a defect observed on a test piece. This paper describes a comparator technique for measuring the visibility of a test defect relative to any chosen standard and discusses the relationship between the uncertainty of this measurement and the various design parameters such as lens aperture, focal setting, magnification and source coherence. The measured visibility of a selection of standard scratches as recommended in BS, DIN and MIL standards are recorded, together with an in-house comparator set provided by Kodak Ltd. The method enables any defect to be quantified in terms of a line-equivalent width and therefore provides means for the intercomparison of different standards.

The scratch standard. (MIL-0-13830A) is a cosmetic standard. that is effected by a visual comparison with a set of submasters that is in turn evaluated by comparison with a set of master standards. Both manufacture and certification of the submasters are somewhat unre-liable. In this paper, we show that the submasters can be classified. according to the relative power scattered at a relatively small angle. We have designed etched gratings with Which. to replace the submasters; these gratings have the appearance of scratches but diffract a broad peak between, 5 and 10 degrees off the axis of the incident beam. We have classified some prototypes both by comparison with the master standards and by a photo-electric measurement; agreement between. the two methods is good. We suggest that such gratings be used as the submasters and possibly that they be classified by a photoelectric rather than visual measurement.

The scratch/dig concept for specifying optical surface quality was first suggested by scientists at Eastman Kodak in 1944 and subsequently incorporated into a military specifi-cation (Mil Spec), the current version of which is MIL-0-13830A, revision L. The initial Kodak work suggested that the scratch width in microns equal the scratch number, but the scratch standards defined by the Mil Spec have a width in microns which is 0.1 times the scratch number, i.e., a number 60 scratch is 6 in ± 10% in width. Digs have an average width which is 10 times the dig number. When used as appearance standards -- the original intent -- the width is not critical. However, if the Mil Spec is used as a specification for laser optics, the width is important. When the surface damage mechanism is joule heating and the scratch or dig is absorbing, the damage threshold is directly related to defect width. The theoretical analysis and experimental evidence supporting this conclusion and the practice of using the Mil Spec for laser optics specification will be discussed.

This paper is concerned with the functional effects of various types of defect found on optical surfaces. For convenience, the defects may be divided into two main types - localised defects such as scratches and distributed or non-localised faults such as scatter from coatings and imperfect polish. These two types need to be considered separately but it is important to assess their relative effects in the image planes of optical systems so as to permit the writing of suitable standards and to avoid the waste of effort due to the inappropriate use of some types of specification. This paper will consider mainly refracting, image forming optical systems operating with incoherent illumination at visible wavelengths. Laser systems are a special case as are reticules and other surfaces lying very close to an image plane. For these systems, particularly stringent defect specifications are required. Cosmetic qualities are very important in the ophthalmic industry but these are not being dealt with here. The glare characteristics of optical systems are considered and the glare contributions from the various kinds of defect are related to the overall glare level. Some attention is given to the existing standards and it is concluded that there are sometimes weaknesses in the current methods of specification. Much more attention should be paid to the effects of the distributed faults, whereas the localised defects often receive far more attention than is warranted by their functional effect.

The purpose of this study is to give a practical understanding of the effect of abnormalities found on various types of thin film coatings on different types of substrates used at 10.6μm.

Modern semiconductor device technology imposes increasing demands on the surface perfection of polished and epitaxial silicon slices. These requirements along with the need for statistical process control call for improved objective, automated and quantifiable inspection techniques. An overview is given over various imperfections of polished and epitaxial silicon slices. Current prevailant inspection techniques are discussed with emphasis on their shortcomings regarding the above mentioned requirements.

The measurement of fine surfaces optical finishes and flaws is becoming more important because of a number of factors. One of these is the hunt for better quality of conformance another is the smoother surfaces required in present day applications such as found in the computer and video industries. Defects such as scratches, pits or cracks cannot only impair the cosmetic appearance of the object, they can actually cause premature failure as in fatigue or corrosion. These new measuring criteria have caused some real problems to instrument makers. In the case of defects the problem is that of spatial bandwidth; that is the problem of searching for a small scratch over a wide area. When measuring fine surfaces the problem is usually the signal to noise ratio of the instrument itself. In many instances the search for defects or the measurement of fine surfaces has been left to human judgement - a powerful if unpredictable measuring tool. This is becoming unsatisfactory because standards have sometimes been built into commercial evaluation of quality based upon the eye. This is rather unfortunate; it ties the hands of the instrument maker who for compatibility has to try to simulate the eye or use indirect measurements.

Comparison of ellipsometric and reflectance measurements provides information on the amount of Total Integrated Scatter (TIS) when the surface model is reasonably well known. This is possible because the ellipsometric data are not strongly affected by scattering whereas the measured reflectance is reduced below the "model based" value by an amount equal to the TIS. The ellipsometric data provide the information necessary to correctly fit the model parameters (n and k). This procedure is discussed and a combined ellipsometer/reflectometer is described. Results of measurements on a metal surface are presented.

Scanning electron microscopy (SEM) was used to identify four distinct laser damage mor-phologies in ZnSe/ThF₄ multilayer mirrors. There were three types of defect-initiated damage morphologies. Oblong-shaped damage sites oriented perpendicular to the electric field vector of the laser were associated with particulates on or near the surface of the ZnSe/ThF4 multilayers. Circular-shaped damage sites were initiated by particulates embed-ded beneath the top ZnSe layer. Selective laser damage at pinholes was identified as the third defectinitiated damage morphology. In addition to defect-initiated damage, stress-related damage was indicated by cracks near or within laser damage craters and erosion sites. Selective laser damage at nodular growth defects in SiH/Si0₂ multilayers was also observed using SEM. Samples with different numbers of nodules were prepared in-house using RF-diode, reactive sputtering. The low-defect mirror had the highest laser damage onset, and the mirror with the highest number of nodules had the lowest laser damage onset.

Photographs of the Hubble Space Telescope's primary mirror after cleaning in June 1984, have been analyzed. Prior to the cleaning, approximately 2.5% of the mirror aperture area was obscured by dust. Particle sizes ranged from 25 micrometers diameter up to about 250 micrometers. These particles are what settled out on the mirror while covered in a class 10K clean room over a period of 2.5 years. After the cleaning operation, the primary mirror had no more than 0.7% of its area obscured by particles of dust. These particles are predominantly less than 50 micrometers in diameter.

We outline the results of surface profile measurements which suggest that many solid surfaces might realistically be modelled as multi-scale or fractal in nature. Experimental results of scattering radiation from such a well characterised fractal surface are presented and show a strikingly different behaviour to that expected from smoothly varying single-scale surfaces.

This paper describes the possibility to determine the statistical caracteristics of a reflecting diffusor (tooth or dental restorations samples with different qualities of polish) by means of holographic correlation. A modified Vander Lugt method is used to measure the local roughness and the lateral mean dimensions of diffusing elements of the surface. This non-destructive method allows surfaces to be tested without preparation, and the quality of polish and finish of tooth or dental restoration to be checked in their actual aspect. A comparison with S.E.M. method is also presented.

The scattering of optical components has become a more and more critical issue in modern sensor systems. This is a result of improved performance capabilities on the part of detectors and of optical elements themselves, as well as an increased awareness of the effects of scatter on system performance. There are several programs that evaluate scatter in complicated optical trains, the best known of which are APART and GUERAP (Arizona Paraxial Analysis of Radiative Transfer and General Unwanted Energy Radiation Analysis Program). These programs require component scatter data as inputs for the analysis and even the beginning design approaches. These data have generally been generated with laboratory instruments that were essentially goniometric radiometers. The data have been difficult to obtain and usually obtained for small representative samples of the real surfaces. If, in the process of manufacture of the parts a measurement of the scatter was made, it was done with a relatively crude jerry-rigged apparatus. Sometimes the measurements made this way were quite accurate, but they were never easy.

A Schlieren microscope is described which can be used to measure profiles of optical quality surfaces. The intensities in the image field are measured with a photo-diode array. By measuring a reference sample and the object, both with and without knife it is possible to correct for different errors. Shape and roughness can be measured with accuracies of 10 nm and 1 nm respectively with a horizontal resolution of about 3μm. The system is fast and relatively insensitive to vibra-tion.

A long scan non-contacting optical surface microprofiling instrument is described in this paper. The instrument ^l, based on the principle of differential interference microscopy, measures the profile of a precision surface along a single scan line. The present hardware allows scan lengths of up to 100 millimeters. Data is acquired at a rate of 2 millimeters per second. Calculation of the surface profile from the raw data requires 2 minutes using a standard IBM-PC. The lateral resoluition of the instrument is diffraction limited at 2 micrometers. This paper presents surface profile data for several types of pecision surfaces. Data from a commercial quality optical flat demonstrates vertical resolution of less than 2.0 Angstroms and lateral resolution of approximately 2.0 micrometers. An aluminized surface relief zone plate with vertical surface deviations in the range of 12,000 Angstroms is profiled over a scan length of 3 millimeters. Irregularities approximately 100 Angstroms high are clearly resolved. A scan of a Gaussian diffuser shows the ability of the instrument to profile surfaces with large surface deviations. The RMS roughness of the diffuser is 2.0 micrometers. Data is also presented for a non-optical surface. Surface profile data for a burnished Winchester disk illustrates the use of the instrument to study the surface characteristics of components use for high density information storage.

Different possible interferometers for surface roughness measurements are described. Advantages and disadvantages of optical techniques compared to stylus instruments are discussed. Important features of phase-shifting techniques for surface roughness measurements are given. The WYKO NCP-1000M optical profiler is described along with techniques for measuring surfaces having an rms roughness on the order of 0.1 nm.

The quality of optical surfaces is generally evaluated by how much light (normally visible light) is scattered by the surface. Most optical glasses and many coating materials are completely opaque to ultraviolet light (253.7 nm). Ultraviolet light tends to scatter much more than visible light. Scatter measurements made with ultraviolet light are therefore very sensitive and the scatter from second surfaces and from the interior (bulk) of the optical material is eliminated by the opacity. A novel scattermeter that operates with ultraviolet light has been developed. The construction and operation of this scattermeter will be described. Cleaning soon becomes the limiting factor when measuring the surfaces with very low level of scatter. Sensitivity to repeated cleaning has been investigated. Different surfaces are compared and uniformity of surfaces is measured by mapping a surface area with an x-y stage. Polished glass surfaces generally have much higher scatter than natural glass surfaces (fire polished, drawn or floated surfaces). Very low scatter levels have been found on thin drawn glass.

A high vacuum instrument for angle-resolved scattering studies in the 30 nm - 633 nm wavelength region has been built. Scattering can be measured at nearly all angles, not only in the plane of incidence, but also out of the incident plane by mounting the detectors on a trolley on a semi-circular rotatable arm. The sample is mounted on an XYZ-translator which also allows the angle of incidence to be continuously varied from 0 to 900. Three inter-changable detectors, two PM-tubes and one channel electron multiplier cover the complete wavelength region of interest. The spherical geometry of the detection system assures that the same part of the detector surface is used at all detection positions. Because of its flexible detector and sample positioning system, the instrument also provides an easy-to-use reflectance and transmittance measurement system. The instrument can be adapted to any light source. At present we make use of a resonance light source, A 58.4 nm for the VUV and a HeNe-laser for the visible. Monochromatized mercury and deuterium radiation will soon be provided for the UV and visible regions.