Requirements on wafer flatness, like most semiconductor specifications, are becoming increasingly tight, with greater accuracy and resolution needed for measurements. In addition to traditional peak-to-valley surface deviation and root-mean- square roughness measurements, it is desirable to measure the flatness of silicon wafers over a small area, or site flatness. This involves dividing the wafer into many sub- regions and calculating the surface statistics for these smaller regions in addition to the overall wafer statistics. Veeco Metrology has developed a high-resolution phase-shifting laser Fizeau interferometer for site flatness testing. The system is designed with 40 mm X 40 mm square field and a 1000 X 1000 pixel CCD camera. Features as small as 100 micrometer may be measured by the system with high resolution, repeatability, and accuracy. A motorized stage allows any region of the wafer to be measured by the system such that problem areas do not escape measurement. This paper discusses the overall system design and presents data from the wafer flatness tester developed by Veeco. Data on lateral resolution, vertical repeatability and accuracy are presented. In addition, the site flatness statistics of a silicon wafer measured by the instrument are given.
WYKO Corporation is currently designing and manufacturing specialized phase shifting interferometers to aid in the qualification of large optics for the U.S. NIF program. The interferometers will be used to qualify homogeneity of raw material and provide in-process inspection information and final inspection qualification data. The 24' systems will be the largest commercially available Fizeau phase shifting interferometers ever manufactured. Systems will be produced using traditional CCD cameras as well as megapixel CCD camera for applications requiring higher lateral resolution. Mechanical and optical design considerations include vibration and distortion control of critical optical elements, polarization control of the laser source, imaging system design, and optical transfer function optimization. We also address effects in the test cavity arising from measuring transmitted and reflected wavefronts of optics mounted at Brewster's angle.
The goal of this 24 inch phase shifting Fizeau interferometer design is to measure the wavefront of an optical window at Brewster's angle. There are two important requirements: a small wavefront slope error and a high optical resolution. To test the sample in transmission, each pencil of light returned from the RF must go back through the window from which it was previously transmitted. Therefore, the slope of the wavefront transmitted through the TF has to be less than a few arc seconds, especially for a long cavity length. For example, for a 2 meter round trip, a 5 arc second slope causes the beam to deviate 0.05 mm. For a 431-mm sample imaged onto a 1000 pixel array, a 0.05 mm displacement corresponds to a 0.116 pixel, which is negligible. However, when a 100 mm sub-aperture is imaged, a 0.05 mm displacement is significant. A shorter round-trip distance can effectively reduce the displacement. The deviations due to a 5-arc second wavefront slope is 0.12 pixel for 2-meter round trip in the full aperture and 0.10 pixel for 0.4-meter round trip in the sub-aperture imaging. Because the phase of the optical window is to be measured and not the amplitude or intensity, the MTF is ont suitable for evaluating the interferometer's resolution. A phase object was measured to determine the system transfer function. The fidelity of the measurement is required to be within 60 percent amplitude for a specified spatial frequency range. For example,for a sinusoidal phase object with a phase undulation of 0.01 wave p-v, the measured result should not be less than 0.006 wave. From theory, a phase object with a smaller phase undulation can be imaged with good fidelity. Because the wavefront slope and optical resolution requirements are very tight, to ensure the interferometer meets these requirements, theoretical errors were thoroughly analyzed and the design implementation was carefully studied.
WYKO Corporation is designing and building 24' (61 cm) aperture phase shifting interferometers to aid in the manufacture and qualification of optics for the NIF (U.S. National Ignition Facility). The first interferometer is scheduled for delivery in early 1997. The 24' systems will be the largest commercially available phase shifting interferometers, and will use a megapixel CCD camera to give high lateral resolution. Some of the NIF optics will be tested at Brewster's angle, and that condition places unusual design requirements on the interferometer. The main effect of testing a planar optical element in transmission at Brewster's angle is that there is a large separation between areas of the element and the return flat, so that optical propagation effects become important. We describe our design of a large aperture phase shifting interferometer, and how it will be used to test the NIF optics.
This interferometer is designed to measure 75x40 cm laser glass. The entire surface is measured
at a Brewsters angle, 56.57°, with an s-polarization beam. The reflected beam is retro-reflected
by a highly reflective mirror. Thus, a 75x40 cm surface can be tested with a 60-cm aperture.
The most troublesome problem is the ghost reflection from the rear surface of a flat while the
front surface is being measured. After the second surface is polished, both surfaces are reflective
and their beams can interfere. However, the second surface of a flat is to be polished to ensure
the transmitted wavefront quality, not the quality of the surface itself. Therefore, the second
surface does not need to be measured directly. To avoid reflection from both surfaces, the laser
is switched to a p-polarization after the first surface is measured while the flat is still at a
Brewster's angle. Thus, the transmitted wavefront is not affected by the reflection.
We believe that a 60-cm clear aperture, Fizeau phase-shifting interferometer is the most practical
and accurate instrument for testing 75x40 cm optical flats. In this paper, we briefly summarize
the important design factors, and show in theory that the design can meet the required
Refractive microlens arrays with lens speed F/1 to F/5 are fabricated by the PROM (photoresist refractive optics by melting) technique. Optimal PROM fabrication parameters are determined from interferometric measurements of the optical quality. Elements in a hexagonal PROM microlens array, composed of 10,000 200 micrometers diameter F/2 lenses on 205 micrometers centers, exhibit less than one-tenth wave deviation from sphere. Four planes of these F/2 microlens arrays, each containing 10,000 lenslets have been assembled into an afocal imaging system.
The microlens array reported in this paper is for enhanced optical coupling of near IR photons into the edges of the Rockwell solid-state photomultiplier (SSPM). Here the edge-illumination is necessary to achieve the required quantum efficiency of near IR photons in the SSPM, a device which is capable of photon counting. SSPM detection requires a method of highly efficient optical coupling to the gain medium to concentrate light with a 100 fill factor where lenslets are centered with a precise 0.2 micrometers accuracy. The microlens array is designed for a center wavelength of 1.3 micrometers , 1975 micrometers pixel dimension and a speed of f/4, which results in the array being nearly diffraction limited. The smallest feature size is 0.9 micrometers for the 8-phase level devices. From this design, we have successfully fabricated an 8-phase- level SSPM microlens array which demonstrates a 0.2 micrometers alignment accuracy among all three mask levels. SEM studies of the microlens show a high-quality surface finish and near vertical sidewalls. Optical characterization demonstrates that the microlens array is diffraction- limited at the design speed and design wavelength, with diffraction efficiency higher than 84.
A goal of our work with binary optics is to identify the underlying efficiency limits of the technology. To aid in our understanding, we have designed a mask set that incorporates ten different 200-micrometers square lenslet designs with speeds ranging from f/0.8 to f/61 as single elements and as 10 X 10 arrays. The f/6 lenslet of the set is also used in 3.2 cm diameter arrays for an afocal imaging relay application. To measure efficiency reliably, we have developed a dedicated apparatus: the `EtaMeter.'' Based on a dual-beam, single- detector, self-referencing approach, the system offers low-noise performance, long-term stability, and excellent repeatability. EtaMeter measures relative efficiency with 0.001 precision and, when calibrated, gives absolute efficiency measurements accurate to 0.005. We have measured the optical efficiency of several devices and compared the results to benchmark calculations, concentrating on the effects of layer-to-layer alignment accuracy for 8-phase-level devices. For an f/4.5 microlens we find a distortion-induced excess loss of about 1.5/0.1 micrometers misregistration. For an f/4.5 microlens with overlay registration better than 100 nm, we achieve an absolute efficiency of 0.85, corresponding to 96 of the prediction.