Optics: measuring and testing – By dispersed light spectroscopy – Utilizing a spectrometer
Reexamination Certificate
1999-03-29
2001-11-27
Evans, F. L (Department: 2877)
Optics: measuring and testing
By dispersed light spectroscopy
Utilizing a spectrometer
C356S369000, C250S372000
Reexamination Certificate
active
06323946
ABSTRACT:
BACKGROUND OF THE INVENTION
This invention relates in general to optical measurement instruments, and in particular, to an optical spectroscopic measurement system including a spherical mirror that focuses radiation in an off-axis configuration and refractive elements.
Ellipsometers are used to measure thickness and optical constants of thin films as well as the optical constants of bulk materials. They function by directing a beam of light on the sample at a high angle of incidence and analyzing the effect of the sample on the polarization of the reflected or transmitted beam. Ellipsometers used to measure patterned samples, such as integrated circuits, must be able to measure within small features (often 50 microns wide or less) surrounded by a completely different material or film stack. Even an extremely small amount of light falling on the surrounding features and collected by the detector can cause errors in the measurements. Thus the optical systems focusing light onto the sample and collecting light from the sample must be designed to minimize the radiation falling on or detected from areas outside the smallest measurable feature, and this condition must be achieved over the entire range of wavelengths used by the instrument. It is also necessary for the optical system to have a minimal effect on the polarization of the light. In addition the optics must not physically interfere with the flat sample which is often very large.
If the wavelength range is relatively narrow, refractive microscope objectives work adequately to focus a small spot onto the sample. When the wavelength range is large, refractive objectives exhibit too much chromatic aberration. Reflecting objectives (such as the Schwarzchild design) using a concave and convex mirror are well known and have zero chromatic aberration. However if any significant demagnification is required to produce a small spot and high numerical aperture (NA), then the angles of incidence on the internal mirrors are too high and the polarization is altered. De-magnification is required because the light must first pass through a polarizing prism which can only handle a small numerical aperture beam. This must then be converted to a larger NA beam by the focusing optics to produce a small spot.
Theoretically, the ideal reflective optical element to focus a collimated beam (such as from a laser light source) onto the sample would be an off-axis paraboloidal mirror, and the ideal shape to focus a small source (such as a fiber optic end) onto the sample would be an off-axis ellipsoidal mirror. An off-axis ellipsoidal mirror has been used before in a spectroscopic ellipsometer, such as in U.S. Pat. No. 5,608,526. Such an aspheric mirror can provide a wide range of possible demagnifications with a small angle of incidence on its surface. It has no chromatic aberration and (theoretically at least) has no other aberration. These mirrors can also be combined with some types of low or zero power refractive elements with minimal impact on the optical performance. Examples of such elements are compensators or waveplates (frequently used in ellipsometry to deliberately change the polarization state), windows (used to control air currents or to enclose the sample in a vacuum or gas), apodizing filters (such as ones described in U.S. patent application Ser. No. 08/835,533, filed on Apr. 8, 1997), other optical filters (useful for calibration and diagnostics) or low power lenses (potentially useful for adjusting focus or magnification). Generally, the maximum tolerable thickness or power of these elements is inversely dependent on the NA of the beam.
The main problem with these aspheric mirrors is that they are made by a single-point diamond turning process that leaves a surface with a figure error composed of a multitude of grooves and ridges covering a broad range of spatial frequencies. Each spatial frequency component diffracts light at a characteristic angle increasing the stray light outside the desired small spot on the sample. This characteristic error produced by diamond turning is one of the main factors limiting the minimum box size within which accurate ellipsometric measurements can be made. Aspheric mirrors made with conventional polishing would perform better, but are very expensive given the aperture and focal lengths needed by this system. Techniques also exist for replicating aspheric mirrors in a thin layer of epoxy on glass, but the performance and durability of these mirrors has not yet been proven in this application.
To eliminate the diamond turning grooves, it would therefore be desirable to replace the aspheric mirror with an optical system composed of only conventionally polished spherical surfaces. The optical system still must meet all the requirements discussed above. If the aspheric mirror is replaced with a tilted spherical mirror, the spherical aberration, astigmatism and coma would be much too great. There are many designs that exist to correct the aberrations from a tilted spherical mirror. But none can be adapted to work adequately for ellipsometry.
U.S. Pat. Nos. 4,208,585, 4,196,961, and 3,598,468 use a tilted glass plate either before or after the tilted spherical mirror. This approach suffers from many drawbacks when applied to ellipsometry systems. Among these are space constraints, inadequate aberration correction, and excessive chromatic aberration when used in the ultraviolet (UV) and visible wavelengths.
U.S. Pat. Nos. 4,230,394 and 4,588,269 use the approach of having two spherical mirrors tilted in orthogonal planes. This design does not correct the aberrations adequately for ellipsometry purposes especially considering the approximate 18:1 demagnification required for many ellipsometry systems.
U.S. Pat. No. 4,226,501 uses a relay system composed of 4 spherical mirrors to correct the aberrations from a tilted mirror. While the incident angle on each mirror is relatively small, the combined effect of four mirrors would have too large of an effect on the polarization of the light.
U.S. Pat. No. 5,168,386 uses a single positive meniscus lens before or after the tilted spherical mirror, or the lens can be placed where both incident and reflected rays are intercepted by it. This design works well for narrow wavelength ranges, but has too much chromatic aberration for a wide wavelength range, such as one including visible and ultraviolet (UV) wavelengths.
U.S. Pat. No. 4,135,820 uses a faceted beam combiner. The corners of the facets would scatter too much light into adjacent patterns for ellipsometry purposes, and the angle of incidence on the facets is too large.
A tilted spherical mirror can also be thought of as an off-axis section of a much larger axially symmetric spherical mirror (the axis defined by the object and image points). For conjugate ratios other than 1:1, the axially symmetric spherical mirror will exhibit pure spherical aberration. An off-axis segment will produce aberration that seems to include components of astigmatism and coma, and a tilted spherical mirror is often referred to as having such aberrations. However, for the purposes of lens design, it is more useful to consider the aberration of an off-axis segment as being a subset of the spherical aberration of the larger, axially symmetric mirror.
A design that corrects well over the entire aperture of a low F# system could be adapted for use in ellipsometry by cutting out off-axis segments of all the mirrors and lenses. There are many such catadioptric designs that use lenses to correct a spherical mirror. Classic examples include the Schmidt camera and Maksutov telescope (see Kingslake, “Lens Design Fundamentals” chap. 14, Academic Press, Inc., San Diego, Calif., 1978). The chief problem with adapting most of these designs is that they have too much chromatic aberration when used from the deep UV through near infrared (IR). The classic way to correct for chromatic aberration in lens is to make it using 2 elements made of glasses having different dispersions. Chromatic aberration in the UV is a particular problem because th
Evans F. L
Kla-Tencor Corporation
Skjerven Morrill & MacPherson LLP
LandOfFree
Spectroscopic measurement system using curved mirror does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Spectroscopic measurement system using curved mirror, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Spectroscopic measurement system using curved mirror will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2579198