SEMICONDUCTOR MEASUREMENT TECHNOLOGY PROGRESS REPORT July 1 to September 30, 1976 Abstract: This progress report describes NBS activities directed toward the development of meth- Key Words: Auger electron spectroscopy; capacitance-voltage methods, dew-point sensing; draggingstylus probe; electrical properties; electronics; four-probe method; hermeticity; hole mobility; infrared reflectance; ion implantation; ion microprobe mass analysis; line-width measurements; nuclear-track technique; particle-impact noise detection; photolithography; photovoltaic method; p-n junction; powerdevice grade silicon; resistivity; resistivity variations; resistors, sheet; safe operating area, transistor; scanning acoustic microscope; second breakdown; semiconductor materials; semiconductor process control; silicon; silicon dioxide; silicon-on-sapphire; spreading resistance; test patterns; test structures; thermally stimulated current and capacitance; thermal properties, transistor; thermal response; thyristors; transistors, power; ultraviolet reflectance; x-ray photoelectron spectroscopy. This is a report to the sponsors of the Semi- The work of the Program is divided into a num- The report of each task includes a narrative description of progress made during this reporting period. Additional information concerning the material reported may be obtained directly from individual staff mem bers identified with the task in the report. Organizational locations and telephone numbers for Program staff members are given in Appendix A. Background material on the Program and in- Technical services in areas of competence are 2. HIGHLIGHTS Highlights of progress in the various technical task areas of the Program are listed in this section. Particularly significant accomplishments during this reporting period included: (1) completion of the systematic study of the effects of surface preparation and probe material on the empirical calibration between specimen resistivity and spreading resistance of n- and p-type silicon; (2) initial evaluation of the nucleartrack technique for quantitative determination of trace amounts of boron in silicon; (3) development of procedures for using an optical research microscope to make accurate measurements of the width of a clear line in a completely or nearly opaque background or of an opaque line in a clear background down to 0.5 μm; (4) design of a compact cross-bridge test structure for electrical measurement of line width and sheet resistance in minimum line-width geometries; (5) completion of the initial phase of the study of the particle-impact noise detection test for screening devices for the presence of loose particles in the package; (6) demonstration of a greater-thanexpected line resolution capability for the scanning acoustic microscope; and (7) development of a nondestructive technique to measure the onset of second breakdown in forward-biased, medium-power transistors. New tasks were undertaken to investigate techniques for characterizing extrinsic silicon for use in infrared detector arrays, to evaluate the application of x-ray photoelectron spectroscopy to diagnostic measurements in semiconductor device processing, and to develop test structures with integral signalprocessing circuitry to allow high-speed measurement of small currents. The work on the high-speed, dragging-stylus spreading resistance instrument and instrumentation for measurement of capacitance and conductance at applied voltages of up to 25 kV was completed. Work on development of techniques for measuring trace impurities in process chemicals (such as diffusion sources and carrier gases) was terminated because available measurement techniques proved to be too insensitive for conclusive experiments. Unless another organization is identified, the work described in the following paragraphs was performed at the National Bureau of Standards. Materials Characterization by Electrical Methods Work continued on the study of correction factors for four-probe resistivity measurements on slices of small diameter and intermediate thickness. The results of measurements on a 5/8-in. (15.9-mm) diameter, 10-2 cm, p-type silicon slice suggest that conditions may exist for which it is not valid to use as an overall geometric correction factor the product of the individual thickness and diameter factors, as is assumed in the standard four-probe method. The first phase was completed of the study of the effects of specimen surface preparation and probe material on the empirical calibration between specimen resistivity and spreading resistance. The calibration relation was found to depend on both surface preparation and probe type for n-type silicon; no probe dependence was found for p-type silicon. Nearly linear relationships between spreading resistance and resistivity could be obtained on p-type silicon with (111) surface orientation and, for certain probe materials, on ntype silicon with (100) surface orientation. For no combination of surface preparation and probe materials studied could a linear relationship be obtained for n-type silicon with (111) surface orientation. Preliminary evaluations of simplified correction-factor algorithms for rapid calculation of resistivity or dopant density profiles from spreading resistance measurements on beveled sections of semiconductor structures have been completed at Solecon Laboratories. Algorithms based on the local slope of the measured spreading resistance profile were found to give satisfactory results under certain conditions. In work at RCA Laboratories, conditions were found which permit reasonable probe life in the high-speed, dragging-stylus apparatus for measuring spreading resistance. Results of measurements along the diameter of a heavily striated silicon slice demonstrated that resolution comparable with that of a conventional stepping-type instrument could be obtained with the high-speed instrument in 1/20th of the time. HIGHLIGHTS A new method was developed to extract background dopant densities in the range from 1016 cm 3 to about 1018 cm -3 from junction capacitance-voltage data. This method is intended for use in cases in which the junction is prepared by diffusion into a uniformly doped substrate. In the method, it is as- sumed that the diffused layer profile in the region of the junction is approximately Gaussian. In addition, work is continuing on the examination of the range of validity of the Schottky equations, which are commonly used to extract dopant density from capacitancevoltage data. The methods for developing theoretical expressions for carrier mobility in silicon as a - function of temperature and dopant density are being extended to cover the case of hole mobility in p-type silicon. In this case, there is an additional complication because of the complexity of the valence band structure. Initial work is being concentrated on boron-doped silicon; silicon doped with other acceptor impurities will have slightly differ- ent mobility-versus-dopant density characteristics because of differences in fractional ionization occurring in the dopant density -3 range from 1016 to 1019 cm even at room temperature. A new task was initiated to study methods for characterizing extrinsic silicon for use in infrared imaging arrays. Initial efforts are being concentrated on evaluation of various techniques to detect extraneous levels in indium-doped silicon; particular emphasis is being placed on study of the so-called Xlevel, which appears to be an acceptor state located between the indium level and the valence band. The presence of this level at concentrations considerably lower than the indium concentration reduces the performance of indium-doped infrared detectors and makes it necessary to operate them at lower temperatures to obtain the best signal-to-noise ratio. Measurements of thermally stimulated current and capacitance were made on a gold-doped ptype silicon MOS capacitor. As expected from theoretical considerations, the phase I thermally stimulated current and capacitance responses of the gold donor center in this structure were the same as responses of the center in an ntp junction diode. In addition, measurement procedures were established to characterize defect centers by analyzing their transient capacitance response under isothermal conditions. Progress was made toward all three of the objectives of the task to evaluate the use of thermally stimulated current and capacitance measurements as a means for characterizing defects in power device-grade silicon wafers. An improved thermally controlled chuck and automatic wafer prober was assembled and tested; the characteristics of this apparatus meet or exceed all design targets. A processing sequence, in which all temperatures are 400°C or less, was selected for fabricating MOS capacitors for making thermally stimulated current and capacitance measurements. Initial steps were taken toward defining procedures for fabricating test structures which provide access to both sides of the p-n junction in a partially or completely fabricated thyristor structure. Development and initial evaluation of the 25kV system for capacitance and conductance measurements as a function of voltage was completed at RCA Laboratories. This system permits application of the extended-range capacitance-voltage technique to metalinsulator-semiconductor structures with insulator thickness in excess of 150 μm. The automated data collection and analysis system for making photovoltaic measurements to determine resistivity variations along the diameter of circular slices was assembled and is being tested. This system is intended to be used to measure slices suitable for fabrication of high-power thyristors and rectifier diodes. The system allows a user to measure the resistivity profile along a slice diameter in about 2 min with some user-system interaction after the measurement run has begun. Although measurement of the average resistivity by the van der Pauw technique has not yet been automated, this measurement can routinely be made using the same probes as in the automated system to make contact to the slice. HIGHLIGHTS A num ber of algorithms have been developed which give varying degrees of recovery of original information in the presence of asymmetric Gaussian instrument functions. Similar algorithms have been developed to treat the problem of recovering signal from noise. The study of the effects of various experimen- nation of data-reduction algorithms. Boron cannot be detected by conventional neu- A new task was undertaken at the Jet Propul- Work on a rapid, nondestructive reflectance Materials and Procedures for Wafer Processing HIGHLIGHTS Photolithography - Additional investigations were completed of the effects of the operating conditions of an optical microscope on line-width measurements in transmitted illumination. In particular, the effects of a finite transmittance of the opaque background and of spherical aberration and defocus were studied. The optical and mechanical performance characteristics of equipment required for accurate measurement of the width of photomask lines as narrow as 0.5 μm have been defined. The NBS optical microscope system can presently measure line widths to within an uncertainty of ±0.2 μm; if the scan is repeated at the same position on the line, the the uncertainty is less than ±0.05 μm. It appears that the edge quality of the line may be the limiting factor in the accuracy and repeatability of the measurement of the width of lines narrower than 10 μm. adequate calibration may be possible by use of appropriate test conditions. Additional studies were completed on selected aspects of the particle-impact noise detection test for screening devices for the presence of loose particles in the package. Principal emphasis was placed on the determination of the effectiveness of various couplants for transmitting the high-frequency vibrations generated by the impact of a particle against the package wall. At this stage of the investigation it appears that, of the three couplants recommended in the test procedure being proposed for inclusion in MILSTD-883, the liquid couplant is most effective in transmitting acoustic emission signals and provides most repeatable results. Device Inspection and Test Investigations at Stanford University and Hughes Research Laboratories of the scanning acoustic microscope for examining semiconductor devices and integrated circuits were directed primarily toward examination of a variety of test specimens in order to establish an improved understanding of the information which can be obtained with this instrument. The microscope was found to be as easy to operate as an optical microscope; its simple controls and lack of a vacuum system render it more convenient to operate and maintain than a scanning electron microscope. Moreover, it is superior to the scanning electron microscope for the nondestructive examination of insulating surfaces because charge does not accumulate and degrade the image. Edges and material differences can be displayed with greater clarity than with either optical or scanning electron microscopes. Various levels in an integrated-circuit structure can be displayed in a series of micrographs readily by making small changes in the lens-to-specimen spacing. The microscope has a resolution capability greater than expected, about 1/3 of a wavelength at 10-percent contrast. The microscope has also been successfully utilized to view gallium arsenide field-effect and transferred-electron devices, aluminum fuse links, and clad and unclad optical fibers. The predictions of a physical model which was proposed to explain the relationship between thermal instability and second breakdown in forward-biased medium-power transistors were confirmed experimentally. A method for nondestructive determination of safe-operatingarea limits based on the exclusion of current constrictions and hot spots was developed. This method is also suitable as a nondestructive test for second breakdown under condi |