Special for SEMICON WEST '99 - July 12-16, 1999
NSMP IN 1999: METROLOGY INNOVATIONS ARE CENTRAL
TO SEMICONDUCTOR ADVANCES
Media Contact: Manufacturing innovations in the semiconductor industry—which lead to smaller and more complex integrated circuits and electronic devices—pose an obvious challenge: scientists and engineers must invent new ways to measure tiny dimensions and quantities, such as the widths of lines in minuscule circuits, ultrathin layers of insulation and the power outputs of lasers. Without such measurements, new products cannot be tested accurately and characterized for commercial distribution. To help American industry shoulder the research and development burdens associated with new measurement technologies, the National Institute of Standards and Technology launched the National Semiconductor Metrology Program in 1994. The program, managed by NIST’s Office of Microelectronics Programs, was designed to meet the most critical measurement needs identified by industry, including those listed in the Semiconductor Industry Association’s National Technology Roadmap for Semiconductors. The following recent developments at NIST represent a sample of the metrology program’s progress:
|
NIST Gathers Valuable Gas Data for Chip ProcessingIn the competitive struggle to increase the processing efficiency and the quality of semiconductor wafers, U.S. industry must have reliable data for the properties of numerous gases used in chip processing. That’s because mass flow controllers—which are critical to the processing of semiconductor wafers—must be calibrated differently for each of the more than 50 gases now in use. Complicating matters is the fact that many process gases are extremely dangerous to handle, making it impractical for the manufacturers of mass flow controllers to directly calibrate each controller for each type of gas. To get around this problem, industry calibrates mass flow controllers using benign gases, such as nitrogen, and then fine tunes the calibrations with models that are based on approximations of the properties of individual gases. More accurate data about the gases needing calibration would reduce the degree of guesswork needed for these models. Now NIST is developing a comprehensive, reliable database for process gases. Agency researchers are gathering the data by measuring the speed of sound as it travels through gases. The technique yields accurate information about the heat capacity and the equation of state, which is used to determine the density of a gas from measurements of its temperature and pressure. Thermal conductivity, viscosity and diffusion constants also are derived from specialized acoustic measurements. The work is being conducted by John Hurly and Michael Moldover in the Fluid Science Group of NIST’s Chemical Science and Technology Laboratory. Technical Information:
|
| Promising Strategy May Help Overcome Measurement Dilemma As the features of electronic circuits continue to shrink, it becomes increasingly difficult to accurately measure their widths because images of object edges are inherently fuzzy at extreme magnification. No microscope is perfect, so images always contain distortions at some level, and width measurements are particularly sensitive to these. With mathematical models of the instruments, the distortions can be corrected. But how accurate are the models? The answer can be found only by testing the models with a sample for which the exact width is known. Unfortunately, there are no such samples; the inherent image fuzziness makes them hard to come by. Now, researchers at NIST have taken a major step in solving that problem. Their solution is to make a sample that can be measured by different instruments that operate on completely different principles. Then, the image distortion is corrected with models developed at NIST (and in some cases, now commercially available). The better the results agree, the less likely that significant model errors exist. NIST scientists have successfully measured a silicon sample with three types of measurement techniques: scanning electron microscopy, atomic force microscopy and a method called electrical critical dimension (or ECD) measurement, which determines the width of a feature by analyzing its electrical resistance. The uncertainties for the first two instruments were a mere five and 13 nanometers, respectively. The electrical measurement technique yielded a higher uncertainty of 34 nanometers, possibly because the sample’s low electrical conductivity made it less than optimal for this type of measurement. The next step will be to measure a sample with higher conductivity so that the ECD can be better measured. Technical Information:
|
| NIST X-ray Detector Heads for Commercialization
NIST has granted co-exclusive licenses to EDAX Inc. of Mahwah, N.J., and NORAN Instruments Inc. of Middleton, Wisc., for commercialization of a revolutionary microcalorimeter-based X-ray detector with an energy resolution of two electron volts, some 50 times better than conventional semiconductor-based detectors. The new technology will be used in instruments for the characterization and analysis of materials by X-rays in semiconductor and other materials-intensive industries. The detector fits easily onto a commercially available scanning electron microscope and conveniently operates even though the sensor is cooled to near absolute zero. The vastly improved detector system will enable chemical analysis of particles that are difficult or impossible to study with current detectors. It permits the chemical analysis of tiny particles that contaminate silicon wafers during semiconductor fabrication. It also has been used to measure the shift in X-ray energy that occurs due to chemical bonding of one atom to another. In the meantime, the NIST research team that developed the technology is exploring other uses for it. One current project is evaluating the microcalorimeter’s role as the detector on a high-resolution mass spectrometer that might help speed up human gene sequencing. Technical Information:
|
| Microstrip Testing More Accurately Measures
Interconnect Performance As transistors get smaller and faster, interconnect
performance also must be improved. To meet this challenge, manufacturers are combining
low-dielectric constant (known as low-K) thin films with high-conductivity copper
interconnects. The speed of light in these low-K thin films approaches that in a vacuum
while the copper significantly reduces loss. Until recently, it has been difficult to
accurately characterize the performance of these A collaborative effort between two NIST groups—the Electromagnetic Properties of Materials Group and the Monolithic Microwave Integrated Circuit Program—and SEMATECH has developed microstrip test structures that thoroughly assess the dielectric properties of candidate low-K thin films and the conductivities of accompanying metals over a range of 50 megahertz to 40 gigahertz. The test structures are small printed interconnects comprised of the thin film and metal combinations to be characterized. Testing has already been completed on a new low-K dielectric and copper conductor system supplied by SEMATECH. Other systems from companies such as Dow Chemical and Texas Instruments are undergoing performance assessments at NIST's Boulder, Colo., laboratories. Technical Information: Michael Janezic
|
| NIST Leads in Key Measurements Needed for 21st Century Chips Scientists at NIST are the first to succeed in making critical measurements of materials that will be essential in manufacturing 21st century computer chips. Industry hopes to be producing the chips based on this new technology in high volume by
the year 2005. Their circuitry will be made using a laser with wavelengths so short that
the light does not penetrate air. This requires the semiconductor wafers to be
manufactured in an oxygen-free environment or in a vacuum, within a region of the spectrum
called the The intricate circuitry in chips is created using a photolithographic technique in which lasers and photosensitive chemicals are used to form features in semiconductor wafers. Current state-of-the-art chips are manufactured using a laser that emits wavelengths of 193 nanometers (billionths of a meter). Future progress will require the use of lasers that emit wavelengths of 157 nanometers. However, traditional lens and masking materials used in photolithography are not transparent at such small wavelengths. Therefore, more exotic materials must be used. But these materials can only be used if extremely accurate data about their optical characteristics are known. For example, a material’s index of refraction, or how much the material bends light, must be known for calcium fluoride, which will likely form integral lenses in equipment used to make future chips. Recently, researchers at NIST have been the first to succeed in making key measurements by using a specially modified spectrometer that operates in the vacuum ultraviolet and is capable of the ultra-high accuracy needed for the job. NIST, which is working with scientists at the Massachusetts Institute of Technology’s Lincoln Laboratories, is currently the only source for the data. The NIST researchers recently presented their results at the International Society for Optical Engineering’s 24th International Symposium of Microlithography. Technical Information:
|
| SRMs Improve Resistivity and Sheet Resistance Testing for
Silicon Technology The resistivity (electrical resistance of a conductor per unit volume) of silicon wafers and the sheet resistance (electrical resistance of a conductor per unit length) of deposited thin films are vital concerns to the multi-billion dollar semiconductor industry. NIST recently has released seven new Standard Reference Materials that will enable manufacturers to calibrate resistivity and sheet resistance test instruments to 0.3 percent or better, with 95 percent confidence. This is at least a fivefold improvement in the uncertainty of certified value, compared to previous resistivity SRMs. The improved uncertainty significantly surpasses the accuracy and precision requirements specified for silicon resistivity measurement capability set forth at a previous SEMATECH Workshop on Silicon Materials for Mega-IC applications. The SRMs are 100 millimeters in diameter, approximately 625 micrometers thick, and are intended for the calibration, or performance verification, of four-point probes and eddy current testers. In addition to greatly reduced uncertainty, the new reference standards have better uniformity of resistivity and thickness, and a larger characterized area than previous NIST resistivity SRMs. Unlike previous SRMs that were certified for value only at the wafer center, the new standards provide certified measurements at the center and on circles of 10-millimeter and 20-millimeter diameters for better compatibility with automated resistivity/heat resistance uniformity mapping instruments. The seven SRMs are: SRM 2541 at 0.01 ohm centimeter, SRM 2542 at 0.1 ohm centimeter, SRM 2543 at 1 ohm centimeter, SRM 2544 at 10 ohm centimeter, SRM 2545 at 25 ohm centimeter, SRM 2546 at 100 ohm centimeter and SRM 2547 at 200 ohm centimeter. To purchase the SRMs at a cost of $731 each, contact the Standard Reference Materials Program, Bldg. 202, Rm. 204, NIST, Gaithersburg, Md. 20899-0001, (301) 975-6776, fax: (301) 948-3730, srminfo@enh.nist.gov. Technical Information:
|
|
US
Department of Commerce July 1999 |
