Narrower linewidths have been axiomatically desirable in semiconductor device geometries for over two decades, while rapid improvements in optical lithography were largely responsible for the tremendous growth rate in integrated circuits. Projecting the course of future development in lithography is perilous, however, because we are approaching serious limitations in both device design and optical lithography. This special issue of Optical Engineering is aimed at clarifying some technical factors that will affect the course of development, and it properly begins with an introduction to the subject by A. F. Tasch. Tasch carefully considers a growing awareness that registration at each edge on the etched substrate is the most important issue in nanometer lithography, and that major obstacles also exist in scaling down the size of metal oxide semiconductor (MOS) devices. A slow pace in finding solutions to device design limitations may retard the need for reducing linewidth and, in turn, profoundly affect the future course of nanometer lithography.

This paper examines the impact of submicron metal oxide semiconductor (MOS) integrated circuit technology on submicron lithography, and contrasts the lithography picture today with that for submicron features. A considerably larger number of factors must be dealt with rigorously because they either do not scale with decreasing dimensions or they do not lend themselves easily to more rigid control so that it has become disproportionately difficult to reduce their effect. In addition to the lithography issues, other serious device technology limitations arise at submicron dimensions. These have to do with device isolation, gate insulation, parasitic resistance and capacitance, interconnectivity, particle-induced upset, and hot electron effects. These issues must also be successfully resolved if submicron dimensions are to be successfully exploited in submicron integrated circuits.

The acrylate family of resists based on polymethyl methacrylate (PMMA) has paved the way for submicron lithography. The physical and chemical modifications, which have improved the performance of PMMA, are reviewed. These improvements include sensitivity to radiation (electron-beam, optical, ion, and x-ray), thermal and plasma process stability, and superior film quality. Structural modifications of the acrylate polymer chain have been the main contributors to enhanced lithographic performance.

A hybrid lithographic process, utilizing both electron-beam (E-beam) and conventional optical exposure techniques within the same device level, has been developed using a commercially available positive photoresist. Following E-beam exposure of the <3.0 um geometries and optical exposure of the larger-sized patterns, both sets of images are developed in a single step. Using this process, functional complementary metal oxide semiconductor (CMOS) devices have been fabricated with polysilicon gate lengths of 0.75 and 0.50 Am. The effect of E-beam dose upon the submicron gate critical dimensions has been determined to be significant due to the proximity effect. The performance of this lithographic process will be compared to that of the hybrid process developed by Berker and Casey, which also utilizes a positive photoresist, but in an E-beam tone reversal mode.

Numerous types of electron-beam lithography systems have been designed and built over the years. The variable shaped probe with vector deflection type has emerged as the most common of the recent designs. This architecture's popularity stems primarily from the superior writing speed for any given resolution and dose requirement by virtue of the ability to expose numerous pattern elements in parallel while at the same time retaining the flexibility of digital pattern description. High currents for rapid registration mark acquisition and relatively low circuit speed requirements are also advantages of the variable shaped spot vector deflection design.

The use of low energy electron beams (<5 keV) to expose polymeric resists has been limited due to worse aberrations, lower available current density, and the difficulties involved in spinning on a very thin continuous resist layer. The first two problems can be eliminated by applying a retarding electric field directly above the workpiece. Computer simulations showing space charge effects on the probe size and improved aberrations coefficients are presented. An inorganic resist, GeSex, is ideally suited for this lithography due to the very shallow active layer ~0.01 um. Experimental results showing the favorable effects due to lower electron landing energy on the required dose and proximity effects are also presented. Lateral scattering in the resist is simulated by Monte Carlo techniques for 2 and 20 keV isolated, unexposed, submicron lines.

A full-field x-ray exposure system based on a 4 kW stationary palladium source emitting the 4.37 A La line has been developed and evaluated in the fabrication of silicon integrated circuits. It has been used successfully in the patterning of 1 um design rule high performance n-channel metal oxide semi-conductor (NMOS) circuits of various complexities from ring oscillators to large scale custom logic and memory devices. A shadow mask consisting of a 6µm boron nitride membrane and 0.7 pm gold-tantalum features is used to transfer patterns into a chlorine-based x-ray resist sensitive to the 4.37 A radiation. Pattern transfer to the silicon surface is faithfully reproduced through the use of a trilevel structure and anisotropic reactive ion etching. Alignment of various levels is accomplished optically through the use of a bifocal, split-field microscope at high magnification and the use of dark-field edge sensing for improved contrast. Linewidth control and registration data are presented for various NMOS circuit patterns.

Fresnel zone plate patterns, when printed on masks and wafers and suitably illuminated, form images that are useful for alignment in microlithography. These images are observed to be largely unaffected by poor edge definition or by pattern defects arising from variations in processing and linewidth control. This paper describes how these zone plates are used for automatic alignment in x-ray lithography. To obtain high contrast bright images with good signal-to-noise ratio (important in automatic alignment), the zone plates are illuminated with a collimated HeNe laser beam. The images formed at the focus are found to be nearly diffraction limited with diameters determined by the zone plate aperture ratio. With zone plates approximately 100 um in diameter and 300 um focal length, we have observed signal-to-noise ratios better than 20 dB and have been able to obtain position accuracies better than ±0.1 um. Our geometry is chosen so that magnification errors in the mask or wafer up to ~1 um are automatically compensated. A microprocessor-based system that performs all the alignment functions automatically is described.

Masked ion beam lithography (MIBL) is a high resolution pattern replication technology that uses a collimated proton beam directed through a patterned mask to expose a resist-coated wafer in proximity to the mask. A major step in the demonstration of MIBL as a practical lithography for submicrometer device processing has been achieved with the fabrication of functional n-channel metal oxide semiconductor (NMOS) devices. We present details of the mask fabrication, resist and process technology development, and the resulting NMOS test chip characteristics.

A broad range of materials and processing techniques amenable to producing resist systems for ion beam lithography are discussed. The effect of random fluctuations in exposure dose on feature size for a Gaussian beam of constant shape is calculated. The results of Monte Carlo simulations of exposures of PMMA on silicon by 50 keV H+,100 keV H2+, and 150 keV Li+ ions are presented, and it is shown that feature resolution is fundamentally limited by the physical processes through which energy is deposited.

A study was conducted using electromagnetic theory to determine the parameters necessary to optimize laminar groove gratings for use in the vacuum ultraviolet. This study predicted that the absolute efficiency of correctly optimized laminar and sinusoidal groove gratings should be about the same, and that triangular groove gratings should show higher efficiency. For the example calculated, it was shown that when compared to a triangular groove grating, laminar grooves should show far superior higher order attenuation at wavelengths around blaze and be better or equal up to about twice the blaze wavelength. From then on, the triangular profile is better. Actual results measured showed good correlation with predicted values.

The use of a telescope with a "one-dimensional" aperture (10 X800 cm, for example) is suggested for high angular speckle interferometry. In the image plane, the speckles are elongated in a direction perpendicular to the entrance pupil. For a given amount of atmospherical turbulence, a one-dimensional pupil gives a contrast gain in the high frequencies region compared to the circular objective having the same resolution limit. Associated with a spectroscope, a one-dimensional telescope allows the representation of the spatial-spectral plane with no loss of photons. This is obtained because the image of the telescope aperture is the entrance slit of the spectroscope. It permits precise, high resolution astronomical measurements as a function of wavelength, as well as velocity speckle interferometry at a high level of photons with CORAVEL-type experiments. The study of a domeless, altazimuthal mount, low cost prototype of 4 cm X80 cm, intended for solar observations, is currently underway at the Nice Observatory.

This paper describes a simple method of measuring the refractive index of a liquid. The basic element is a thin wedge cell, and the source of light is a low power He-Ne laser. Conditions are derived to obtain the best possible accuracy using this setup. The same setup can also be used for the determination of the refractive index of a solid medium such as glass.

Several methods for the measurement of very long radius of curvature of a convex or concave spherical mirror are reviewed in this paper. These methods have relevance in the measurement of the radii of curvature of laser mirrors. The accuracies achievable with the various methods are compared and summarized.

A method of reducing the cost of large optical telescopes is the use of a segmented primary mirror. The computer-controlled polisher approach is ideally suited to the fabrication of telescope mirror segments. Since the process achieves controlled material removal by regulating the tool motion, there is no need for any surface symmetry. Recent experiments have demonstrated the ability of this process to figure mirror segments.

Gold transmission gratings with 0.3 Am spatial period, 0.6 Am thick have been fabricated and are being used for the spectroscopy of high temperature, laser-produced plasmas. A transmission grating has been coupled with a soft x-ray streak camera, producing the first time resolved (20 psec resolution) continuous x-ray spectrum over the range 0.1 to 1.5 keV from a laser-produced plasma. In addition, a transmission grating has been coupled with a high resolution, Wolter-design x-ray microscope to produce an imaging spectrometer of unprecedented spatial and spectral resolution. In laboratory experiments the imaging spectrometer demonstrated a spectral resolving power, (A/AA) of 200 and a spatial resolution of 1 µm.

Techniques for combining individual semiconductor lasers for use as a laser communication transmitter are discussed. The geometries, aging characteristics, and lifetimes of semiconductor lasers are reviewed, and the resulting effects on the laser communications transmitter are assessed.

A low contrast phantom designed specifically for the evaluation of the image quality obtained from a digital fluorographic imaging device (DFID) is described. The design rationale and the image quality evaluation procedure utilizing the phantom are discussed along with the test results.

The processing requirements for future imaging systems will overwhelm existing ground processing capabilities that are at present loaded to capacity. This bottleneck would be relieved by implementing intelligent systems onboard the spacecraft which are capable of transforming raw sensor data into useful information prior to transmission. As an initial step toward realizing this objective, an adaptable data processing system is being developed which will accommodate high speed multispectral image data in real time. The development includes all test and support equipment required to evaluate and demonstrate the operational performance of the system in terms of the applicability of the design to high speed onboard processing. Demonstrations will be conducted at NASA Langley Research Center throughout the first half of 1983.

We describe an infrared imaging photometer employing a monolithic 32 X 32 pixel bismuth doped silicon charge injection device array. The device is primarily useful in the 8 to 13 karl atmospheric window. The detector is sufficiently sensitive to provide good performance on ground-based telescopes and promises to be very good for low background space flight operation.

Autonomous operation of tracking systems implies the need for abilities which are predicated on discriminating a target from nontarget objects and noise. Intelligent discrimination is most often hampered by a lack of explicit information, hence requiring elaborate processing to extract implicit informa-tion for an electronic "brain." The fact that it is now possible to construct sensors which provide not only brightness information on a two-dimensional map, but also motion and range on equivalent maps, bypasses the aforementioned processing to directly provide sufficient explicit information for the discrimination of targets. The only processing then required is to combine the information in a composite map such that targets stand out clearly from non-targets. This is particularly true of airborne targets where range and motion are generally quite distinct relative to their background. This general philosophy pertains as well to ground targets, in that sensing directly those characteristics which distinguish them from other objects greatly facilitates the tasks of locating and identifying such targets. For ground targets, sensors designed to sense characteristics such as straight lines, smooth surfaces, color, symmetry, and motion would, by use of appropriate composite mapping, detect and acquire predefined classes of targets with high levels of discrimination.

An optical open area ratiometer for measuring the percent open area of perforated sheet materials has been developed. The device uses a technique based on Fourier optics in which a detector is employed to measure a small central portion of the light in the Frau nhofer diffraction pattern of the perforated sheet. This paper discusses the associated problems and results of testing the accuracy of the device. The accuracy is shown to be on the order of 2% of the open area ratio.

A new kind of surface-measuring machine has been developed under government contract at Itek Optical Systems, a Division of Itek Corporation, to assist in the fabrication of large, highly aspheric optical elements. The machine uses four steerable distance-measuring interferometers at the corners of a tetrahedron to measure the positions of a retroreflective target placed at various locations against the surface being measured. Using four interferometers gives redundant information so that, from a set of measurement data, the dimensions of the machine as well as the coordinates of the measurement points can be determined. The machine is, therefore, self-calibrating and does not require a structure made to high accuracy. A wood-structured prototype of this machine was made whose key components are a simple form of air bearing steering mirror, a wide-angle cat's eye retroreflector used as the movable target, and tracking sensors and servos to provide automatic tracking of the cat's eye by the four laser beams. The data are taken and analyzed by computer. The output is given in terms of error relative to an equation of the desired surface. In tests of this machine, measurements of a 0.7 m diameter mirror blank have been made with an accuracy on the order of 0.2µm rms.

It is with some trepidation that I undertake the use of the editorial as a means to express something of my individual approach to doing science. I will share the blame for my decision to "go public" with these thoughts with some outstanding researchers around the world who have urged me to do so. In addition, I will make this space available to credible scientists who wish to express other approaches. Part of the joy and charm of science is that, as a human activity, it is a highly individualized affair. All contributors with long careers develop a style of thinking that is unique and readily recognizable by their colleagues. I feel certain I could recognize the author of a paper by one of the "old timers" in my field without reading his or her name. Thus, rny remarks are of a personal nature and may not be useful to most readers. With all of these caveats expressed, I will delay no longer.