Radiant energy – Inspection of solids or liquids by charged particles – Methods
Reexamination Certificate
2000-08-30
2002-10-29
Berman, Jack (Department: 2881)
Radiant energy
Inspection of solids or liquids by charged particles
Methods
C250S306000, C250S252100
Reexamination Certificate
active
06472662
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to methods and apparatus for determining the critical dimension of workpieces. 2. Description of Related Art
An occasional problem with a conventional automated Critical Dimension Scanning Electron Microscope (CD SEM) measurement is poor correlation thereof with subsequent electrical measurements. This problem can be due to feature positions to be measured, for example, the foot of a photoresist line being obscured by an overhanging structure. Other examples are T-topping, undercutting, and negative angle or recursive sidewall.
Currently available standard top/down CD SEM systems prevent the SEM electron beam (SEM-beam) from tilting relative to the sample for several reasons. However, at least one CD SEM provider is developing a system that can quickly and automatically tilt the beam by several degrees and acquire secondary electron waveforms or images from scanning the same structure at various tilt angles.
Such technology can advance the core capability of the CD SEM only if the additional information resulting from changing the angle of deflection of the scanning SEM-beam can be used quickly and in an automated fashion to improve the accuracy of measurement. Problems needing to be solved include positional alignment of the waveforms, separating various contributors to the effective edge width of a tilted structure, and finally, synthesizing the information into a critical dimension measurement.
There has been considerable effort directed at extracting three-dimensional information from two images acquired at different angles of view (stereoscopic imaging), as in robotic vision.
These methods use the phenomena of shadowing and parallax to calculate the relative coordinates (including height) of identifiable features in two or more images. Unfortunately, on the scale of interest for CD metrology (nanometers) and for the primary structures of interest (straight lines or spaces), there are few dependable identifiable features. More seriously, the SEM-beam interaction with the structure is very different from the interaction physics of these other applications. Successful sidewall metrology needs to account for the finite size of the SEM-beam and the interaction volume within the structure material.
An example of scatterometry is found in the area of semiconductor manufacturing metrology. In the approach recently commercialized by Biorad, a defocussed laser beam scatters off of a periodic array of structures on the wafer (target) and the zeroth order diffracted beam intensity is measured for two polarizations of light. Data is collected as a function of the incident angle. The resulting waveform is compared with simulations. The ability and resources for calculating the electromagnetic response for model structures is crucial to this approach. Other variations on this approach include using higher order diffracted beams or multiple wavelengths of light. None of these methods deals with images or waveforms acquired by scanning focused SEM-beams or the very different interaction physics of an SEM-beam with matter.
One noteworthy approach to improving the accuracy of top/down CD SEM metrology is the work of the Spectel Corporation. System responses, based on the use of an approximate simulation of the SEM-beam interaction with model structures, produce a database of waveforms. The best match to the actual waveform is used to interpret the measurement. That is similar in concept to the commercialized scatterometry approach. Possibly, this approach can be applied to the tilted SEM-beam CD SEM system. However, the overhead in calculation resources is significant and the accuracy of the modeling, especially in the presence of sample charging, is highly questionable.
Beam tilting is the same thing as beam deflection that are used for column alignment as exemplified by U.S. Pat. No. 6,066,849 of Masnaghetti et al. for “Scanning Electron Beam Microscope” which applies an x tilt voltage and a y tilt voltage but as described at Col. 11, lines 41-53 , it is employed as follows:
“The upper quadrupole . . . is configured to align the beam after a particular gun lens voltage is selected. In other words, the beam may have to be moved such that it is realigned with respect to the aperture. This realignment is accomplished by supplying an X and Y tilt voltage from the multiplexer control system . . . and the beam may be realigned with respect to the aperture by setting the X and Y tilt voltage values that are supplied to the upper quadrupole . . . ”
See U.S. Pat. No. 5,969,273 of Archie et al. “Method and Apparatus for Critical Dimension and Tool Resolution Determination Using Edge Width” describes measuring hump width to obtain SEM resolution information. See U.S. Pat. No. 6,025,600 of Archie et al. “Method for Astigmatism Correction in Charged Particle Beam Systems”; and U.S. Pat. No. 5,869,833 of Richardson et al. “Electron Beam Dose Control for Scanning Electron Microscopy and Critical Dimension Measurement Instruments”.
A common prior art algorithm is to declare the outer extremal slope location for each edge of a feature on a sample to be the location of the edge and therefore to report the distance between these locations.
SUMMARY OF THE INVENTION
Glossary
DESL
=
Distance between Extremal Slope Locations in
SEM data for a single edge of a feature
Extremal Slope
=
Maximum or minimum slope or rate of change
of a function
Height
=
Height of structural edge of a feature on a
sample
Interaction volume
=
Extent within a sample of excited electron
activity due to the electron beam of the
microscope
K
0
=
constant determined during calibration
K
=
constant determined during calibration
Threshold value
=
Empirically determined minimum of valid
DESL value
Waveform
=
One-dimensional digitized line scan
&thgr;
=
Relative angle between SEM-beam direction
and sidewall
The discovery of a simple relationship between structure properties and the Distance between Extremal Slope Locations (DESL) in SEM data as a function of electron beam tilt angle forms the basis for an automated methodology of obtaining such structural properties without the need for extraordinary alignment or 3D (three-dimensional) reconstruction techniques. Normally careful alignment of the data is required to extract three-dimensional information from multiple SEM images or waveforms (one-dimensional digitized line scans). In cases of interest, related to this invention, that alignment is on the nanometer scale. Today, it is not possible to acquire multiple SEM images or waveforms on the nanometer scale, after stage movement, with blind navigation. Use of pattern recognition can improve matters, if suitable pattern recognition targets that are required are available, which is generally not true.
This invention gets around the alignment problem by not requiring alignment. Instead, each waveform (either directly obtained from the SEM or extracted from a SEM image) can be analyzed to find the locations of extremal slopes for each structural edge of interest. The DESL value so determined should have a precision of a few nanometers. With calibration, the accuracy of the DESL value measured should be comparable to its precision.
In order to automate a fast determination of structure properties (height as well as, left and right sidewall angles), the method of this invention minimizes the actual number of measurements in real time by requiring that for each feature edge, there are two DESL values determined at different tilt angles that are larger than a threshold value, which is set at the time of calibration.
Variations of this include requiring only one DESL measurement greater than the threshold value if either sidewall angle or structure height is already know. Another variation is to use the height determined from one edge analysis in the analysis for the other edge. This then requires two DESL measurements above the threshold value for one side but only one DESL measurement above the threshold value
Berman Jack
Fernandez K
Jones II Graham S.
Schnurmann H. Daniel
LandOfFree
Automated method for determining several critical dimension... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Automated method for determining several critical dimension..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Automated method for determining several critical dimension... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2932209