INTERFEROMETRY: Asphere interferometry powers precision lens manufacturing
Semiconductor, consumer electronics, and aerospace systems all depend on aspherical optics to increase product performance and value.
A large-aperture, low-uncertainty laser Fizeau interferometer with highly expanded dynamic range enables accurate qualification of high-performance aspheric optics.
Semiconductor, consumer electronics, and aerospace systems all depend on aspherical optics to increase product performance and value. Aspheres provide an extra “degree of freedom” to optical designers-not only do they improve optical performance but aspheres also help decrease overall system cost and part count, reduce weight, and increase the system light-transmission efficiency. Lens diameters range from submillimeter up to several hundred millimeters, and accuracy requirements span from microns to subnanometers.
The primary limitation to the broad application of aspheres has been the ability to qualify them to the required level of accuracy. Although different manufacturing methods, such as computer-controlled polishing, small-tool polishing, ion-beam polishing, diamond turning, and precision molding, have existed for many years, the success of these fabrication techniques often depends on the capabilities of metrology equipment.
Asphere process-control challenges
For adequate manufacturing process control, aspheres present a unique set of metrology challenges.
Tight manufacturing tolerances for precision aspheres are the primary hurdle for available metrology solutions. Low measurement-uncertainty requirements that are preferably three to five times smaller than the tolerance of the measured part are a tremendous obstacle when considering lithography optics requiring measurement uncertainties of less than 0.1 nm root-mean-squared (rms). In addition, metrology solutions must be able to span an 800 µm asphere departure from a best-fitting sphere to accommodate 90% of the manufactured aspheric lenses.
Complex data processing is often an attribute of asphere manufacturing. To measure asphere form and waviness and to detect surface defects, three-dimensional (3-D) surface data requires the collection of individual x, y (position), and h (phase) data with a density that can exceed 200,000 data points. This measurement grid must overlay the lens without distortion to minimize errors that would require compensation or correction. Asphere manufacturing also requires a rapid total average cycle time (TACT) that must be faster than the production process, which is typically five to ten minutes for aspheres smaller than 80 mm in diameter manufactured using a small-tool polishing process.
Finally, the often high-volume production requirements for aspherical optics must be nondamaging to the optical surface and allow for in-process quality control. Because manufacturers are often producing a number of different optics of varying size and shape, a flexible metrology system that can measure multiple parts without major tooling changes is preferred .
Today the available metrology systems fall into three classes: touch probe, null compensators, or stitching systems. Touch-probe systems contact the optical surface and build a map like a coordinate-measuring machine. Null compensators are custom lenses or computer-generated holograms (CGH) that transform the spherical wave of a laser Fizeau interferometer into an aspheric wavefront that matches the test asphere. Stitching systems acquire small patches of the asphere surface with a laser Fizeau interferometer and tie them together into a full surface map. Unfortunately, none of these systems meets the requirement for on-line process control.
Touch-probe systems are very flexible and measure a broad range of parts. However, they can only perform low-density two-dimensional (2-D) line scans with very slow TACT (3-D data maps can take between 20 and 60 minutes) and have the potential to cause damage when the probe contacts the optical surface.
Null lenses are inflexible and offer limited accuracy depending on the null-lens quality and alignment. Null compensators also require weeks to manufacture and offer no flexibility for different optical sizes and configurations.
Stitching systems have a slow TACT and are currently limited to departures from the best-fit sphere of less than 80 µm, one-tenth of what is required. Certainly, none of these systems meets the measurement uncertainty requirements of aspherical optics as demanding as lithographic projection lenses, for example.
A new asphere metrology system
Driven by custom product-development programs, Zygo has developed a novel noncontact laser Fizeau interferometer for asphere measurement. The metrology tool combines two Zygo core technical competencies: laser Fizeau interferometry and displacement-measuring interferometry.1, 2 This approach uses standard interferometer components, including a laser Fizeau mainframe, transmission lenses, motion stages, and a displacement-measuring interferometer (DMI)-a DMI measures linear position of motion stages to nanometer resolution. The result is a new class of noncontact asphere metrology systems. These systems are fast, accurate, and produce a high-data-density surface map.
Laser Fizeau interferometers are traditionally used to measure spheres. An interferometer compares a reference surface to a test surface where the reference and test surfaces form an optical cavity. In the cavity the light returns into the interferometer along the same path it exited the system (its common path). In a common-path system, all optical path differences in the interferometer are zero except in the cavity where the measurement is performed-a critical condition for measurements with low uncertainty. When high data density, low measurement uncertainty, and high-speed measurement are requirements, interferometers are the instrument of choice for spherical optical metrology.
However, Fizeau interferometer performance is compromised for aspheric surfaces. When testing an asphere the interferometer is common-path only over a localized zone and measurement uncertainty is increased over the rest of the surface. Because the interferometer is no longer common-path over most of the surface, retrace errors are introduced. Stitching-based systems must cope with these retrace errors, limiting their measurement uncertainty. In the worst case with high surface slopes the light will not even re-enter the interferometer. Under high slope conditions it is impossible to measure the full asphere at one time, much less achieve a low measurement uncertainty. But by teaming a displacement-measurement interferometer to the laser Fizeau interferometer, it is possible to remove this limitation.
Zygo focused on axially symmetric aspheres-the largest group of aspheric optics produced by volume. As the aspheric optic is scanned along the optical axis, the ring-shaped region producing interference fringes will move from center to edge (see Fig. 1). Thus, the common-path region with low measurement uncertainty sweeps out the whole surface.
The radial position of the rings has a precise relationship to the design equation and the asphere’s position along the optical axis of the interferometer. This relationship and proprietary analysis unifies the standard phase-shifting measurements at the separate rings as if they were measured at the same time, in the common-path condition, with all the metrology benefits of a laser Fizeau interferometer.
In addition, the data is not stitched. In a stitching system, phase relationships between the measured areas are estimated from overlapping regions. Metrology errors easily propagate region to region in a stitching system. In this system, however, the phase data h is known at each x, y position-solely based on interferometric distance measurements-and in that sense, the asphere laser Fizeau interferometer is an absolute test.
The asphere measurement is the difference between the design asphere surface and the actual surface-the same result expected from standard laser Fizeau interferometry. The result is an interferometric, high-data-density x, y, and h map with form, waviness, and defect detection in one noncontact measurement (see Fig. 2).
FIGURE 2. The asphere interferometry system produces a high-data-density, three-dimensional map of x, y (location), and h (phase) coordinates that allow form, waviness, and defect detection in one noncontact measurement.
The measurement uncertainty is primarily governed by the same parameters of interest to a standard laser Fizeau interferometer. For standard measurement uncertainties, normal metrology environments and reference surface calibrations are typically acceptable. But for the lowest measurement uncertainties, such as in lithography optics, tight temperature control, low-pressure environment, and precise reference-surface calibration must be maintained. Regarding TACT, the speed of measurement depends on the number of asphere zones measured and ranges from three to ten minutes.
Close teamwork between Zygo research and development staff in Germany and the United States, as well as contributions by strategic partners in Europe, the United States, and Japan, has allowed the development of these asphere interferometry systems for custom programs (see Fig. 3). This new technology promises to become a general production metrology tool that will find its way, over time, into commercial product manufacturing.
1. M. Kuechel, “Scanning interferometer for aspheric surfaces and wavefronts,” U.S. Patents 6781700, 6972849, and 6879402.
2. M. Kuechel, OSA Optical Fabrication & Testing meeting, Rochester, NY (October 2006).