New Techniques for Microscope Imaging

With dimensions shrinking to below 0.18 µm in semiconductor manufacturing and the increasing complexity of layers and metrology of wafers and masks, using optical techniques continues to be challenged. To meet these challenges, optical manufacturers are taking new approaches to allow optical imaging to keep pace.

The new approaches include deep UV, 3-D, confocal, scanning probes, and aerial imaging. These technologies take a different approach at developing an optical system to provide the optimal solution. Although they are unique, they all fully use an optical technology and develop it to the highest capability.

Ultra-High Resolution

Resolution is the ultimate goal in microscopy, and many techniques are used to improve the resolution of optical microscopes. The fundamental requirements to improve resolution are:

Reduce the illuminating wavelength.

Raise the numerical aperture (NA) of the objectives.

NA is a measure of the light-gathering capability of an objective. The theoretical limit is 1.0 if there is air between the objective and wafer. However, replacing the air interface with an immersion fluid can increase NA to as high as 1.4.

Typical broadband illumination operates at 500 nm, the wavelength used to calculate the theoretical resolution limit of optics or 0.3 µm. By reducing this wavelength to 365 nm and 248 nm, you can dramatically increase resolution to double what was previously possible or to 0.15 µm.

Materials that normally reflected, absorbed, or transmitted light in the visible spectral range may exhibit changes in the ultraviolet, producing very dramatic images. Finally, post-image processing can increase the image quality, resulting in resolution of lines and spaces down to an 80-nm pitch.

This level of image quality only is possible with a specifically designed microscope optimized for deep UV in the fundamental design. This includes the objectives, the most critical component of a UV microscope. By designing them for specific wavelengths, light throughput and optical corrections can be controlled completely. By combining this imaging capability with a high-performance navigation system, you now have a powerful optical microscope for semiconductor defect imaging and analysis.

Aerial Imaging

New lithography techniques, such as phase shift masks (PSM), optical proximity correction, and decreasing wavelengths into the deep ultraviolet (248 nm), push optical lithography to new performance levels. As a result, defect inspection, analysis, and repair of masks and reticles have increased in complexity.

Aerial image measurement/analysis is one method of analyzing defects and repairs on masks. This technique uses a specially designed microscope that simulates the illumination system of a lithographic stepper.

The mask is illuminated with the same wavelength as the stepper (248 nm and 365 nm), and the aerial image is captured on a high-resolution camera to simulate the actual mask print on a wafer. This saves time and money by eliminating the need to use the stepper to print on a wafer and analyze the results (Figure 1).

Then you can determine the results of the mask repair. With the use of customized and off-illumination apertures, various stepper configurations can be modeled. PSMs can be analyzed for proper phase shifting by focusing through the shifter and collecting the aerial image at each level. This provides a map of the phase shift through the mask.

The results can be fed to analysis software for process optimization. Finally, by interfacing with a defect detection system, defects can be easily found and analyzed.

Confocal and 3-D Spatial

Imaging defects in 3-D has been a long-desired feature of microscopy. As the human brain generally sees in three dimensions, visualizing defects in three dimensions would clearly provide benefits.

The design of optical systems for 3-D imaging has taken several approaches. Confocal imaging provided the first real useful results in 3-D imaging, and today laser and real-time confocal systems are available.

Laser-scanning confocal systems, using lasers as the illuminating sources, scan the laser beam over the object and recombine the image with software. In addition to confocal scanning and reconstruction, using different lasers provides flexibility in various failure analysis techniques; for example, an infrared, 1,164-nm laser for imaging through the backside of silicon and UV for high-resolution confocal imaging. Optical-beam-induced current evaluation, using the laser as the stimulation source, also is possible.

Real-time confocal systems, using a spinning disk with standard halogen or an ark lamp illumination, provide a real-time confocal with real color without the need for software. However, true color separation and representation can be achieved only if the optical system accounts for chromatic corrections made in the objectives (Figure 2). Otherwise, the inherent advantage of a real-time confocal system is lost in the objective corrections.

The chromat concept allows the chromatic distribution to be broader and evenly spread over the focus range, resulting in sharp color shifts with focus. In addition, the color shift can characterize and measure the Z distance.

For 3-D reconstruction, the systems slice an image into planes that can be manipulated in various ways to provide a useful composite image. One popular method reconstructs the image planes into a 3-D surface profile of the scanned region. Reconstruction does not add resolution or contrast to the final image because the quality only is as good as the initial optical data. Post-image processing, however, enhances the image and provides useful contrast and topographical information. In addition, video zooming results in a higher final magnification of the image than a microscope provides.

Either of these confocal systems has advantages. The laser scanning systems have more flexibility and control of the illumination, pinhole size and location, image zooming, and manipulation of the stored image. Real-time confocal systems are easy to use, particularly in a production environment. With tight integration into the microscope, a real-time confocal system provides another imaging alternative in the search for defects.

A newer and potentially more powerful 3-D technique, called 3-D spatial imaging, offers a true stereoscopic image at a high magnification and numerical aperture. This technology uses a switching LCD lamp shutter synchronized with special microscope eyepieces to allow you to recombine the image with your eyes.

The split-type shutter switches at 100 Hz/s, too fast for the brain to distinguish as separate images. As a result, the two images recombine in your brain into a single 3-D image.

This technique can be added to any standard microscope and offers an improvement in resolution of 15% over conventional microscopy and a depth-of-focus improvement of up to three times. Using high-quality polarization components also provides higher image contrast and proper color rendition.

Some applications require the capability to discriminate heights of surface particles as well as determine if a particle is a pit or a bump. Table 1 summarizes the differences between laser confocal scanning, real-time confocal, and 3-D spatial imaging.

Scanning Probes

Scanning probe imaging is a relatively new technique available for high-volume, semiconductor imaging. The technology is based on interpretation of atomic-force interactions between a small probe near the surface of the sample and the surface itself.

With the capability to image at extremely high resolution, down to the atomic level, the scanning probe microscope (SPM) provides a unique and very useful image of surfaces. These images find wide application in microelectronics, photomask metrology, material sciences, magnetic domain studies, and nanostructures.

Scanning probe systems operate in various modes, including magnetic force, contact, noncontact, lateral force, and potential contrast. These techniques vary by the type and shape of the probe tip and the analysis of the probe interaction and provide optical and nonoptical information about the material surface.

The technique is selected on the analysis desired for the material. The contact technique, for example, has extremely high-resolution surface-profile information; however, this can be slow and vibration sensitive. By contrast, noncontact has less resolution but can be faster.

There are inherent limitations to SPM systems. Low throughput comes with atomic resolution. In addition, the field of view is hard to find and typically small as a result. Since the site to be scanned generally is very small, it does not take long to generate an image once the site is found.

One approach that overcomes these limitations installs the SPM on an optical microscope. By combining these techniques, the optical microscope can find the region of interest (ROI) and switch to the SPM, positioning the scanner in the desired location very quickly. This also allows the scanning area to be minimized as the probe is placed directly over the ROI. You also have a single platform with a wide range of imaging solutions—brightfield, darkfield, differential interference contrast, confocal, florescence, and scanning probe—with no technique compromised by the others.

The reputation of SPMs as a laboratory tool is being dispelled. New tools are available with full wafer automation for production use and easy-to-use software interfaces. Although SPMs will not likely replace optical or electron microscopy, it has established itself as a valuable alternative imaging solution for many applications.

Summary

Existing optical systems generally have compromises built into them to allow a broad variety of uses; however, this ultimately limits the performance in any given specialized application. For this reason, future tools must be application specific, allowing optical technology to continue to provide solutions for the microelectronics industry.

Several new, application-specific, optical techniques available today help provide high-resolution imaging, analyze defects, perform metrology, and investigate surface characteristics. They differ because each is designed for a specific application and problem: aerial imaging for mask simulation, UV for ultra-high resolution imaging, confocal for resolution and contrast, and 3-D and scanning probe for surface characterization. Each technique meets a specific need; and so far, results have shown this approach will continue to use optics for inspection and metrology of submicron semiconductor technologies.

About the Author

John Fitch is the general manager of the Microelectronics Business Unit of Carl Zeiss. He has an E.E. degree from the University of Louisville and has been with Carl Zeiss for 11 years. Carl Zeiss, One Zeiss Dr., Thornwood, NY 10594, (914) 681-7397, [email protected].

Wafer Surface Objective Lens Spinning Disk Pinholes

Figure 2a.

Wafer Surface Objective Lens Chromat Spinning Disk Pinholes

Figure 2b

	Laser Confocal Scanning	Real-Time Confocal	3-D Spatial Imaging
Real Time	No	Yes	Yes
Real Color	No	Topography colors	Yes
PC Analysis	Optical sections scanned sequentially 3-D reconstruction with display of surfaces and profiles 0.30-µm lateral and axial resolution PC control	Single image section gives a large Z-range Parallel scanning No PC control	Extended depth of focus Z height determination by viewing No PC control
Features	Precise X, Y, Z measurements Single confocal pinhole	Color-depth function Chromat Nipkow disk