Non-contact, 3D optical interferometric profilers provide detailed topography measurements of super-smooth surfaces
such as hard disk substrates and super-polished optics. However, the contribution of the interferometer system to the
measurement can be significant for surfaces with an RMS roughness of one Angstrom and below. Special care must be
taken to minimize random noise as well as to remove the systematic instrument error from the measured data. While the
random noise can be addressed by low-noise design and averaging of measurements, the systematic instrument error is
more difficult to eliminate.
In this paper an interferometer configuration is presented that eliminates the mid to higher spatial frequencies from the
reference beam. This configuration is called a virtual-reference interferometer, since there is no physical surface in focus
conjugate to the test surface. This provides very smooth systematic instrument errors with essentially no contribution in
the mid to high spatial frequencies of surface waviness and roughness. The virtual-reference interferometer has a midsize
measurement area of 20x20 mm, is fully compensated for white light, extended source illumination, and enables
data acquisition for both phase shifting and coherence scanning modes. Current performance data show a residual
systematic tool waviness error of < 0.2 Angstrom RMS, with potential for improvement. Efficient stitching of subaperture
measurements accommodates high resolution roughness and waviness maps of test surfaces up to 150 mm x100
Despite the introduction of phase-shifting interferometers in the 1980's, high volume catalog and camera production
lenses are still inspected using qualitative visual fringe inspection methods. This error-prone inspection technique
requires the human operator to quickly judge whether the lens “passes” or “fails” based on the appearance of the fringes.
Although this method is sufficient for optics with < 0.25 wave of surface figure irregularity, it is not sensitive enough to
properly inspect surfaces in the increasingly common 0.1 wave regime. Furthermore, as visual fringe inspection is not
quantitative, it does not produce the statistical surface measurement data that is necessary to monitor and optimize
production polish process yields.
To overcome these disadvantages, we have developed a robust, quantitative lens surface inspection instrument. A
compact, user-friendly, and economical 60 mm aperture Fizeau interferometer directly addresses production optical test
applications, providing rapid vibration-robust optical surface measurements of P-V irregularity, RMS irregularity,
astigmatism magnitude, and power via a simple touch-screen interface. Pass/Fail criteria are applied to these values,
enabling accurate and repeatable sorting of production optics based on these quantitative values and eliminating human
interpretive error. Batch statistics are also displayed and stored for customer inspection reports and rapid polish process
feedback. This paper will also describe how next-generation Fizeau interferometers serve as part of a total optical
production process improvement strategy.
A conventional phase shifting interferometer is capable of measuring opaque surfaces with sub-nanometer precision.
However, it cannot be used to measure an object with multiple parallel reflective surfaces such as a transparent plate, a
glass disk, or an Extreme Ultraviolet Lithography (EUVL) mask blank. This is because the plane parallel reflective
surfaces generate multiple interferograms that are superimposed in the recording plane of the interferometer. Although
every individual interferogram is associated with phase information that is related to the height or thickness, the
conventional interferometer is not able to differentiate one surface from another. To measure these surfaces, we have
developed a method that integrates a Fizeau interferometer with a tunable light source and a weighted least-square
technique. The tunable light source controls the wavelength during the data acquisition process, producing phase shift
speeds that are proportional to the optical path difference (OPD). The weighted least-square signal processing
technique separates each surface from the others in an optimal manner. Thus the desired information, such as the front
surface height, back surface height, and relative optical thickness of a plane-parallel transparent glass plate are extracted
without multi-surface fringe print-through artifacts. In this paper we will present the method and demonstrate its
performance. The demonstrated surface height accuracy for EUVL mask blank substrates is 5 nm and the RMS
repeatability is <0.01 nm.
Phase shifting interferometry (PSI) is a highly accurate method for measuring the nanometer-scale relative surface height
of a semi-reflective test surface. PSI is effectively used in conjunction with Fizeau interferometers for optical testing,
hard disk inspection, and semiconductor wafer flatness. However, commonly-used PSI algorithms are unable to produce
an accurate phase measurement if more than one reflective surface is present in the Fizeau interferometer test cavity.
Examples of test parts that fall into this category include lithography mask blanks and their protective pellicles, and
plane parallel optical beam splitters. The plane parallel surfaces of these parts generate multiple interferograms that are
superimposed in the recording plane of the Fizeau interferometer. When using wavelength shifting in PSI the phase
shifting speed of each interferogram is proportional to the optical path difference (OPD) between the two reflective
surfaces. The proposed method is able to differentiate each underlying interferogram from each other in an optimal
manner. In this paper, we present a method for simultaneously measuring the multiple test surfaces of all underlying
interferograms from these superimposed interferograms through the use of a weighted least-square fitting technique.
The theoretical analysis of weighted least-square technique and the measurement results will be described in this paper.
Wafer shape and thickness variation are important parameters in the IC manufacturing process. The thickness variation,
also called flatness, enters the depth-of-focus budget of microlithography, and also affects film thickness uniformity in
the CMP processing. The shape mainly affects wafer handling, and may also require some depth-of-focus if the wafer
shape is not perfectly flattened by chucking. In the progression of technology nodes to smaller feature sizes, and hence
smaller depth-of-focus of the lithography tool, the requirement for the PV-flatness over stepper exposure sites is
becoming progressively tighter, and has reached 45nm for the next technology node of 45nm half pitch. Consequently, in
order to be gauge-capable the flatness metrology tool needs to provide a measurement precision of the order of 1nm.
Future technology nodes will require wafers with even better flatness and metrology tools with better measurement
precision. For the last several years the common capacitive tools for wafer dimensional metrology have been replaced by
interferometric tools with higher sensitivity and resolution. In the interferometric tools the front and back surface figure
of the wafer is measured simultaneously while the wafer is held vertically in its intrinsic shape. The thickness variation
and shape are then calculated from these single-sided maps. The wafer shape, and hence each wafer surface figure, can
be tens of microns, necessitating a huge dynamic range of the interferometer when considering the 1nm measurement
precision. Furthermore, wafers are very flexible, and hence very prone to vibrations as well as bending. This presentation
addresses these special requirements of interferometric wafer measurements, and discusses the system configuration and
measurement performance of WaferSightTM, KLA-Tencor's interferometric dimensional metrology tool for 300mm
wafers for current and future technology nodes.
The precision metrology of patterned wafer is increasingly demanded by the semiconductor device manufacturers. The
most common methods include scanning probe microscopy (SPM) techniques such as stylus profilometry and Atomic
Force Microscopy (AFM). These methods acquire data by contacting the surface over a sequence of one-dimensional
scans. While high lateral resolution can be achieved in this way, such processes are time-consuming and can have the
potential to deform the surface under test. An alternative non-contact interferometric method is presented here. The
method uses the white-light interferometry (WLI) to provide wafer topography quickly in a direct three-dimensional
format. The improved measurement throughput suggests that it is feasible to use this method for production monitoring.
Most commercial interferometers with WLI are capable of measuring opaque surfaces with sub-nanometer precision.
The described method extends this capability to determine the top surface topography of structured surfaces in the
presence of varying phase shifts on reflection. The phase shift on reflection may be due to the material properties of bulk
surfaces, single or multi-layer film stacks on a substrate, or other micro-structures on the wafer. Furthermore, this
method simultaneously or separately provides additional parameters of the test piece e.g. layer thickness and/or material
refractive index for film stacks, or line width and structure depth of micro-structures. The measurement results on
various types of the wafer surfaces will be presented in this paper.
Continuing improvements in the fabrication of super-smooth spherical and aspheric optical surfaces for applications such as extreme-ultraviolet lithography have given rise to the need to characterize the surface roughness of these optics to the sub-Angstrom level. Phase-shifting interference microscopes are well suited to acquire high-precision, three-dimensional surface structure rapidly, without contact, and over a wide range of spatial frequencies. This paper describes a new phase-shifting interference microscope by ADE Phase Shift, together with exemplary measurements of Angstrom-level surface roughness. The microscope was designed to measure large diameter (up to 500 mm) convex, concave and aspheric optics. To access all areas of the test surface, samples are placed symmetrically at the center of an R-theta stage. The phase-shifting microscope head pivots in the direction of the translation stage travel by ±20 degrees to match the major component of surface slope, while a second tilt control nulls minor tilt in the orthogonal direction. A 1K by 1K high-resolution, digital camera reduces random noise in the system to below 0.15 Angstrom rms for 16 averaged maps of nulled interferograms. When testing non-planar surfaces, the interference fringes cannot be nulled out, and special care has to be taken to minimize phase-dependent errors, commonly known as fringe print-through. A print-through level of less than 0.2 Angstrom rms has been achieved with a careful opto-mechanical system design in combination with an advanced phase algorithm whose filtering action suppresses higher-order harmonics caused by phase-shifter and detector non-linearity, as well as vibrations.
Temporal phase shifting interferometers require a stable environment during the data acquisition, so that well controlled phase steps can be introduced between successively acquired interferograms. In contrast, single-frame interferometers need to acquire only one interferogram to provide a phase map with very good precision at high spatial resolution. Thus these interferometers are well suited for the interferometric testing of large optics with long radius of curvature for which vibration isolation is difficult, e.g. testing astronomical telescope mirrors in a test tower, or testing space optics inside a cryogenic vacuum chamber. This paper describes the Instantaneous Phase Interferometer (IPI) by ADE Phase Shift, together with measurement results at NASA. The IPI consists of a polarization Twyman-Green interferometer operating at 632.8nm, with single-frame phase acquisition based on a spatial carrier technique. The spatial carrier fringes are generated by introducing large amount of tilt between the test beam and the reference beam. The phase information of the optical surface under test is encoded in the straightness of the interference fringes, which can be detected in a single frame with spatial sampling of 1000x1000 pixels. Measurements taken at the NASA Marshall Space Flight Center in support of the characterization of developmental optics for the Next Generation Space Telescope are presented. Such tests consist of a mirror placed inside a cryogenic vacuum chamber, with the IPI placed outside the test chamber without any additional vibration isolation.
The analysis of phase detection algorithms in the Fourier domain provides information about the overall performance characteristics for complex, non-ideal signals encountered in real-world instruments. Thus phase algorithms can be optimized and adapted to the particular application at hand. After a review of the basic Fourier description of phase algorithms, several cases are discussed demonstrating this ability. These cases include phase shift error, non-sinusoidal.
A novel instrument is described for the optical, non-contact measurement of the waviness and figure component of the surface texture of flat surface. Here the spatial frequency range for waviness is typically chosen from 1.25/mm to 0.05/mm, whereas the global figure error contains the lower spatial frequencies. The special requirements on the dynamic range, the spatial resolution, and the signal-to-noise ratio of the measurement are discussed. The presented instrument consists of a white-light, extended-source, phase-shifting Mach-Zehnder interferometer. The special design employing low temporal and spatial coherence avoids coherent speckle noise on the measured surface maps while providing good spatial resolution. Thus in the waviness frequency band the modulation transfer function exceeds 0.75, an the RMS- precision is 0.1nm over the measurement area of 100mm in diameter. Measurement examples of typical applications, e.g. substrates for hard disks and flat panel displays, are shown.
An important parameter in the production of flat panel displays is the waviness, or micro-corrugation, of the substrate surface. It describes surface deviations in the mid spatial frequency range between roughness and global shape. Typically, the waviness has to be determined to an accuracy in the 5 nm-range. A technique is presented to measure the waviness optically in a non-contact fashion along a profile 14' long. Thus large panel substrates can be measured. Special issues to consider are: (1) suppression of the light reflected from the back surface of the panel substrate, (2) large required dynamic range due to the overall shape of the panel of more than 100 micrometers , (3) the flexibility of the panel substrates requiring well designed fixturing to avoid bending and pick-up of vibrations. The described technique is based on an extension of a phase- shifting white-light Mach-Zehnder interferometer which has been used successfully in the characterization of thin glass disk substrates.
An infra-red interferonieter for testing optical elements and assemblies in their operational wavelength band is described.
The interferoineter is a phaseshifting, LUPI-style Twyman-Green interferometer operating at a wavelength of 3.392pm.
The system is designed for optical testing in a vacuum chamber. For that purpose the interferometer is contained in a
hermetic, vacuum-compatible enclosure, and all internal alignment functions are remotely controlled. The light source is
an infrared HeNe-laser at 3392 p m. The camera sensing the interferograms consists of a PtSi CCD-array with 256x256
pixels. For operating the system in a vibration environment, short exposure pulses are generated from the laser light. In
addition, a specialized phase.shifting data acquisition and reduction algorithm is employed which senses the random phase
shifts and adapts the phase algorithm accordingly. A vacuum compatible diverger (f/2.5) and beam expander (250mm beam
diameter) allow for a wide variety of optical testing configurations. Measurement results and system performance
parameters are presented.
A custom-built phase shifting interferometer is described for measuring the wavefront quality of Ultratech Stepper 1X projection lenses over a field of 25 mm X 50 mm using wavelengths of 363.8 nm (for i-line systems) and 442.0 nm (for gh-systems). Over an N.A. of 0.42 the wavefront is determined at 240 X 240 points with a calibrated accuracy of (lambda) /20 peak-to-valley (at (lambda) equals 363.8 nm). The interferometer is an unequal path Twyman-Green interferometer with an Argon-laser and a HeCd-laser as light sources. The interferometer test system is integrated on a 10 ft X 4 ft vibration isolation table together with an xy-stage to position the projection lens at different field points. The safe handling of the projection lens is supported by a load/unload mechanism on the table. The measurement wavefront maps at various field points of the tested projection lens are used to optimize the state of alignment of the projection lens.
Using highly coherent laser sources in interferometry often leads to speckles in the interferograms. These speckles constitute a noise on the fringe phase and, hence, lead to a reduction of wavefront measurement precision. They arise from the light scattered by random imperfections of the optical surfaces. A new technique was developed at Carl Zeiss to reduce the effects of speckles in the laser interferometer DIRECT 100 by a virtual reduction of the spatial coherence regarding the speckle contrast. In the technique presented here the direction of the illuminating light beam in the interferometer is modulated while averaging wavefronts (not intensities) with the real-time wavefront averaging capability of DIRECT 100, resulting in a virtually larger extent of the light source. The fringe contrast is independent of this beam modulation, whereas the speckle contrast in the accumulated wavefront is determined by the virtual extent of the light source. Thus, speckle effects not only from the imaging part of the optical train but also from the illuminating part are reduced.
Recently, measurements with slope sensitive optical tests, such as shearing interferometry, Shack-Hartmann test, or Ronchi test, have become of great interest. The primary outcome of these measurements are the slope or difference data of the wavefront under test. Therefore, a numerical reconstruction procedure is necessary to integrate these difference data in order to obtain the desired wavefront map. This paper describes a numerically efficient reconstruction technique for orthogonal difference data as they are obtained, e.g., in lateral shearing interferometers. The reconstruction process consists of two steps. The integration is carried out as a filtering operation in the spatial frequency domain. Since it requires two full rectangular arrays of valid x- and y-difference data, an additional step is necessary to accommodate for general pupil shapes, e.g., circular pupil with central obscuration. This additional step consists of synthetically generating difference data at the previously invalid data points. The reconstruction is unbiased and has a very slow error propagation, i.e., the variance of the noise on the reconstructed wavefronts is approximately equal to the variance of the difference noise.
Surface deviations of spherical mirrors from a best fitting, mathematically ideal sphere were measured to an absolute precision of 0.25 nm rms. Because of the long radius of curvature, a Hindle-type arrangement was used as interferometric setup, resulting in a test arm length of about 1.4 m. A special calibration procedure was implemented to eliminate systematic, setup-dependent errors. A very fast data acquisition technique was combined with real-time wavefront averaging to eliminate the effects of random errors, such as wavefront variations due to the turbulent atmosphere in the beam path. For the evaluation of one mirror surface, all in all 400,000 individual wavefront measurements at 400 x 400 points were combined, requiring an overall measurement time of only one to two days.
The new Zeiss interferometer works with a complex software package which can be run in two different
modes : a "workshop mode" for simple pushbutton operation of standard measurement sequences, and a
"master mode" well suited for the use in the laboratory with changing requirements for measurement and
evaluation methods. Macros of arbitrary functional sequences can be created in the master mode and
associated with function keys for the workshop mode. A description of the implemented features concerning
data acquisition and display, control of the instrument as well as data evaluation and manipulation is given.
We will report on a new interferometer developed at Carl Zeiss, which has real-time measuring
capability with instant visualization of results, is nearly insensitive to vibrations, has a variable fringe spacing
from one lambda to lambda/1O (lambda represents the wavelength of the light used in the interferometric
test), and can give lambda/100 accuracy through a simple calibration procedure. It can be handled with the
same ease and in just the same way as conventional interferometers.