Sellwell (Singapore) Optics Pte Ltd.

Surface Accuracies

Surface accuracy is a measure of how accurately the optical surface matches its designed shape. There are a variety of ways of defining surface accuracy and the errors to the surface shape. They are grouped into three categories based on their frequency across the surface of a part: form errors, waviness, and surface roughness.

Form error, or irregularity, is typically the most important and commonly specified surface specification for aspheres. This specification consists of low frequency or larger errors usually peaking one to three times across a part. Form errors are typically specified as peak-to-valley error in waves or fringes, but can also be specified as a linear deviation in microns or as an RMS deviation.

Waviness, or mid-spatial frequency error, describes ripple-like errors happening in a frequency of 5-100 instances across the part and is most often introduced when a surface is polished with small polishing tools. This rarely happens in whole-aperture polishing performed when making spherical optics. Because of this, waviness is typically ignored in spherical lenses but may need to be specified in aspheres. Waviness is most commonly defined as a slope error over a specific scan length. Waviness sensitivity is application-specific and many lenses are not sensitive to it, therefore it is important to only specify a waviness tolerance if the tolerance will affect your application. When adding additional lens requirements, costs can increase due to added testing.

Surface roughness, or high-frequency error, is a measure of smoothness, or the quality of the polish on an optic’s surface. Surface roughness can affect scatter and the ability to withstand high laser power on the surface. To define surface roughness, it is important to describe both the amplitude and the frequency range of interest as the choice of test equipment may filter out high frequencies. Analyzing surface roughness requires very special testing and can be time-consuming; therefore it is best to only specify surface roughness when necessary.

Radius

Radius error, a specific subset of form errors, is a constant change in radius across the lens. It is the most common and also generally the easiest error for a system to tolerate since it is typically corrected by adjusting the focus position. Radius error can be defined as a percentage change in radius from the design radius (or vertex radius), a linear change in radius, or as fringes of power. Manufacturing costs for a lens can be reduced by allowing a looser radius tolerance.

Metrology

Proper metrology is necessary to ensure an asphere meets all required tolerances. The two most common measurement techniques for surface accuracy or figure error are interferometry and profilometry.

Interferometry measures the difference between a reference wavefront and the wavefront reflected off a surface or transmitted through an optic. Testing an aspheric wavefront is much more difficult than testing spherical wavefront due to the difficulty of generating an aspheric reference wavefront compared to the common practice of generating a spherical wavefront. It is possible to test an asphere using a spherical reference wavefront if the asphere has deviations from a sphere smaller than the dynamic range of the interferometer, but this is seldom the case for aspheres.

Null interferometry is a branch of interferometry that is either done by using null lenses or computer-generated holograms. A null lens is a spherical lens, or an assembly of spherical lenses, designed to have an amount of spherical aberration equal to the departure from a sphere of the nominal aspheric surface. The amount of interference observed shows the deviation between the real aspheric surface and the nominal surface. Computer-generated holograms (CGH) use holography to generate the desired wavefront when a special plate is placed in the reference path of the interferometer. Null interferometry is expensive and time-consuming to set up because it must be carefully calibrated for the shape of the specific aspheric surface being tested, but afterward it can be used to quickly and accurately test many identical aspheric surfaces.

Stitching interferometry is a branch of interferometry in which a small section of the aspheric surface is tested with a spherical wavefront. If the deviation from a sphere is smaller than the dynamic range of the interferometer over a small area, a measurement can be made using a spherical wavefront over that region. Measurements from many small sections are stitched together to give a complete map of the surface. There are several methods of stitching and each varies by the sections they divide a surface into. All stitching methods have limitations on the shapes they can test and are restricted to surfaces without inflection points where the local radius of curvature goes from a positive to a negative radius. Stitching interferometry has a faster set up time than null interferometry, but the test time per part is more than that of null interferometry because multiple sections of the lens must be tested.

Profilometry measures the change in height of a lens by moving a probe across the surface. This is usually done in spirals or slices across the surface, building a cross section or a surface map of the height. Slices are normally quicker to measure but do not provide full surface information. Profilometry is more simple and flexible than interferometry, but is not as accurate. The limitations on shapes a profilometer can test is typically only limited by the slope of the part, while features such as inflection points do not limit the profilometer. Set up time is typically short for a profilometer, but scan time can vary depending on the number of scans or area scanned.