Since the beginning of the, use of Electron Beam Lithography there has been a continuous pressure for very fine lithography. This was initially developed for high frequency FET's and these applications remain a powerful user of nanolithography. [1] In addition since those early days, there has emerged the requirements for integrated optics. Recently the use of electron beams to write distributed feedback laser systems has provided greater flexibility in the selection of their operating wavelength. [2] The use of electron beams to write fine holographic patterns of arbitrary complexity has been a growing application. What all these devices have in common, is a high range of complexity coupled with very fine demanding lithography. In addition, there is a strong requirement for ultra fine lithography for physics experiments. These include the use of Josephson Junctions and associated systems for flux entrapment and fundamental physics experiments, as well as the ability to use complex arrays of regular patterns for electro magnetic experiments. [3], [4] In this case the requirements are characterised by a need for very simple fine structures, together with coarser connection to the outside world.

The production of VHSIC and microwave devices has shown that submicron lithography by E-beam, Optical, or X-ray systems is rapidly becoming feasible. There is, however, a developing demand for a nanolithography tool for producing Josephson, GaAs, optoelectronic devices and for investigating the physics of scaling of semiconductor devices.

Performance tests on the Waferwriter TM system, a high brightness electron beam direct write system, are described. The Waferwriter system is an ultra-high current density, variable size round beam direct write system, capable of exposing existing popular photo-resists at higher rates than conventional shaped beam systems in the micron and submicron feature size range. Performance tests are described covering overlay accuracy, alignment, resolution, and throughput. The system is the first to meet the DoD's VHSIC requirements for E-beam direct write systems.

In the past few years, e-beam technology has gained acceptance as a viable production technique for lithography. This is particularly so in the mask-making areas of semiconductor manufacturing. Only a few manufacturers of semiconductor devices have begun to use e-beam lithography directly in the lab line. This has not been because of a lack of technology but rather lack of good economics. E-beam technology has been suitable for mask-making bectuse the economics make sense, and the mask quality is far superior to that obtained with optical techniques. E-beam has not been suitable for direct writing on the wafer because the throughputs have not been at a level which would make direct write economically viable. Major efforts have been underway by several semiconductor equipment manufacturers to develop a new generation of electron beam lithography systems for direct write on wafer applications. These development efforts are now coming to fruition. One such system is the VLS-1000. The VLS-1000 is a variable shaped electron beam lithography system optimized for direct write on the wafer. This system incorporates several innovative technologies to overcome the hurdles of low throughput and high system cost. A brief description of the system will be given highlighting some of these innovative technologies. One of these, the microseal, has allowed a significant reduction in system size and cost. The microseal allows introduction of wafers directly into the system without the normal incompatibilities and difficulties of vacuum chambers and vacuum locks. Performance results of the microseal will be given. The use of e-beam technology is most viable in the submicron range of feature sizes. Although high accuracy and high. throughput present design conflicts, the use of rigorous error and throughput budgets has allowed the VLS-1000 to provide the highest accuracy performance and competitive throughputs.

Lithographic pattern quality is of paramount importance for the fabrication of microcircuits with submicrometer critical dimensions. The AEBLE 150 direct-write e-beam lithography equipment recently announced by Perkin-Elmer is destined for application in the submicrometer domain, and it will employ on-line quality measurement tools to ensure consistent machine performance and productivity. In this paper, we discuss the strategies to be employed in developing pattern quality measurement tools for AEBLE, and we illustrate early results obtained using the prototype AEBLE and mensuration techniques developed for Perkin-Elmer's mask-making product, MEBES® III. These preliminary results will be used to guide the evolution of AEBLE's quality-assurance tools. We present a scenario for that evolution, including a discussion of measurable parameters and user interfaces.

This paper discusses signal processing procedures which have been developed for three important beam adjustments required by variably shaped electron beam lithography systems: dynamic focussing adjustment, deflection distortion correction and beam size adjustment. Precise and speedy measurement of edge slope, beam position and beam size are all necessary in this context. To execute these measurements, backscattered electron signals are stored in a buffer memory following digital scanning of a shaped beam across a fiducial mark. These signals are then digitally processed in a control computer. These procedures have been applied to an EB lithography system (HL-600). Pattern accuracy of 0.2 μm over a 6.5x6.5 mm field and overlay accuracy of 0.1 μm were obtained.

The era is fast approaching when light optics must utilize all of its options; higher numerical aperture lens, shorter wavelength and partially coherent illumination along with process aids such as multilevel resist processes, to name a few. The relative production process difficulty of light optics and x-ray become equivalent in the 1.25μm regime. With x-ray relieving the imposed limitations of diffraction coupled with recent advances in x-ray sources, masks, alignment techniques and photoresists, it is now time to begin serious x-ray process development.

The characteristics of CPMS (chlorinated polymethylstyrene) to Pd Loc were investigated to obtain the high performance negative x-ray resist. From the results, the high sensitive x-ray negative resist (Dg^0.5=17mJ/cm²), CPMS-X(Pd), were obtained by determining both the optimum molecular weight and the optimum chlorine content to Pd Loc. It was found that this resist also has both high contrast and high dry etching resistivity. This resist can be believed to be a milestone for practical use of x-ray lithography for future VLSI devices.

X-ray lithography has matured from a research and development phase to an implementation phase. Accordingly, the concerns have shifted from imaging issues to those of registration, critical dimension control, step height coverage, and system repeatability. In this paper, results will be discussed relating to x-ray printing and registration for full field alignment systems with 100mm field diameter using optical verniers, SEM (scanning electron microscope) and electrical wafer probe techniques. These results will encompass micrometer and submicrometer imaging using single 'level and tri-level processing techniques.

A dedicated X-ray lithography system has been developed on Beam Line III-4 at the Stanford Synchrotron Radiation Laboratory (SSRL). The research program at SSRL includes studies of the spectral sensitivity of X-ray resists, mask characterization, and the applications of novel X-ray optics. The facilities available for this work and recent results in these areas will be described.

X-ray lithography has been applied in a single-layer resist process to fabricate n-channel enhancement field-effect transistors with effective channel lengths (L_eff) and channel widths (We_ff) as small as small as 0.9 Lim and 0.5 Lim, respectively. The yields on 3" wafers ranged from as high as 50 % for the smallest MOSFETs with L_eff/W_eff of 0.9/0.5 to above 90% for those with Leff/Weff of 2.2/0.5 and 0.9/1.7 without process optimization. This report outlines the mask technology and the device fabrication process. The MOSFET performance is discussed with emphasis on threshold voltage and subthreshold slope uniformity and on wafer-to-wafer variations.

X-Ray lithography: Can it be justified? This is a question which is being asked in many circles today. The answer is, unfortunately, not black or white nor is it easy to determine. However, there are today more serious players in the x-ray field than there were a few years ago and maybe this is an important indicator. In this paper we will try to establish if x-ray lithography can or can not be soundly justified.

There are substantial advantages to bright plasma sources for submicron X-ray lithography. For example, throughputs of 40 wafer levels per hour can be maintained using existing, high resolution, single level resists. The high X-ray brightness of the discharge plasma source is achieved by operating at much higher energy density than conventional sources. The X-rays are emitted from a high temperature plasma, which is magnetically confined and dissipates its energy isotropically. Over the last two years, a significant effort at Maxwell Laboratories has been carried out to characterize the Z-pinch plasma source for applications such as lithography and microscopy. The technology has been brought to the point where high reliability, commercially available sources have been developed for X-ray microscopy. The state-of-the-art is discussed and projections are provided on the capabilities of high power bright discharge sources for lithographic applications.

The development of high-brightness liquid metal ion sources [1] has made it possible to produce focussed ion beams with submicrometer dimensions and moderate current density (about 1 A/cm²). Several classes of applications for focussed ion beam systems have been proposed including direct implantation, lithography, and micromachining. Though focussed ion beams have some unique and attractive properties, they are unlikely to replace mainstream lithography and broad-beam implantation technologies. For instance, for applications such as high-dose source-drain implantations, the writing rate would be limited to only a few thousand pixels per second, resulting in excessive writing times. On the other hand, for certain niche applications a focussed ion beam may be the method of choice. As an example, a custom wafer requiring threshold adjustment implants (10¹³ cm^-2 and 1% coverage) could be processed in less than an hour and would require no mask generation or resist processing and exposure steps.

MicroBeam has under development a focused ion beam instrument intended for a wide range of applications, including microscopy (using secondary electrons), high spatial resolution microanalysis (using secondary ion mass spectrometry), and microfabrication (using ion implantation and sputtering). High collection efficiency SIMS optics enables gray scale elemental maps to be made with resolutions of 200 angstroms. Results from early application studies are presented, including high spatial resolution SIMS in microelectronics, microbiology, and cosmochemistry.

The development of liquid metal ion source (LMIS) technology for use in electro-optical lenses has been largely responsible for the significant growth of interest in focussed ion beam (FIB) systems and their applications during the last decade. Among the earliest work, using capillary nozzle sources, an ion microprobe for surface analysis was developed (1), but following the invention of the needle LMIS (2), most applications work was concentrated upon the use of FIB systems of relatively high energy for ion beam lithography of submic'ron structures in resist (3,4). More recently, the emphasis of microfabrication work has shifted away from resist exposure towards direct FIB implantation doping, requiring the development of dopant-bearing alloy LMIS's (5,6). Most groups have adopted the use of EXB Wien filters (5,7) to separate the dopant ion from the beam, with one notable exception where a symmetrical four sector magnetic filter has been used (8). Growth of interest in all aspects of the technology has been most noticeable in Japan where development of UHV systems for FIB doping of MBE - grown III-V semiconductor material for opto-electronic devices has been the major interest in FIB in a number of laboratories.

Ion beam lithography is developed in three directions: Focused (FIBL), Masked (MIBL) and Ion Projection Lithography (IPL). IPL uses ions as information carrier in a demagnifying step-and-repeat exposure system. With an Ion Projection Lithography Machine (IPLM) a geometrical resolution < 0.25 μm can be obtained combined with a very high depth of focus:more than 100 μm. The high power densities possible with IPL permit not only pattern transfer in conventional organic resists but extend lithography to new processes using resistless ion beam modification techniques of materials. Experimental results obtained with a laboratory type Ion Projection Lithography Machine IPLM-01 are presented.

A focused-ion-beam writing system for experimental use will be described. This system was designed to be simple and to be flexible enough for doing experiments. The system configuration, ion sources, and some considerations regarding alignment signals will be mentioned.

Currently, the integrated MOS dynamic RAM has as many as 256 thousand memory cells per chip based on 2 pm photolithography. Figure 1 shows the history and the prospects for progress in microfabrication technology. Feature size versus year, as reported by Bossung in 1978, is shown, as developed from independent analysis by Moore, Noyce and Gnostic concept. Circles numbered 1 and 2 show that 64K- and 256K-bit RAMs were developed in 1981 and 1984, and that their feature sizes were 3μm and 2μm, respectively. It is significant that the predictions and the real developments are so close. Furthermore, since the basic process for 3 M-bit RAMs based on 1.3μm microlithography has already been reported in conference, it is highly likely that they will become commercially available around 1987, as predicted by the circle numbered 3 based on 1.3μm microlithography.

The limits of submicron patterning by optical lithography were estimated by computer simulation. As a result, the optimum numerical aperture, NA, which gives the highest resolution was determined. Assuming that the permitted defocus value is +1 pm, the lithographic resolution of about 0.7 pm was obtained with NA 0.5 and near-UV exposure. A 0.5-μm resolution was obtained with NA:0.35 and deep-UV exposure. In addition, less than 0.5-μm resolution was suggested using resist system technologies and optical exposure tools that optimize NA as well as the exposure wavelength. Furthermore, it is shown that the goal of half micron resolution will be industrially achieved before the end of 1980's by optical lithography rather than by electron-beam lithography.

The current model of an optical step and repeat system for use in production line has a projection lens with a resolution of l.0m and provides a machine-to-machine overlay accuracy of approximately ±0.2 pm (95% range). The accuracy is to be further improved with a newly developed field-by-field alignment system. The i-line projection lens, developed for higher resolution, adequately resolves 0.8um lines and spaces.

A simple, low cost, multipurpose electron beam lithography system for sub-micron pattern was developed for personal use in laboratory. It has a capability of circle pattern generation and saves a time of pattern drawing for development of new device. There are several examples of fine pattern with few-tenth μm line and space which are drawn by the system.

A three layer resist system using spin-on intermediate layer has been described for optical and e-beam lithography. A thick bottom layer of positive photoresist, 0FPR800 (Tokyo Ohka) was spun on to the substrate, flood-exposed by a UV light, and baked for planarization. A thin intermediate layer of spin-on glass (SOG) or spin-on indium tin oxide (ITO), and a thin top layer of OFPR800 or polymethyl methacrylate (PMMA) were succes-sively spun on to the bottom layer. Transfer of the resist image pattern into the inter-mediate layer was performed anisotropically by reactive ion etching (RIE) in a C₃F₈ plasma for SOG or a CC1₄ /N₂ plasma for ITO. The pattern in the intermediate layer was then replicated in the bottom OFPR800 layer by RIE in an 0₂ plasma. Experimental details are described. The planarizing characteristics of bottom layer, OFPR800 increase with UV exposure and increasing baking temperature. For optical lithography, the complexity of bottom layer processing is discussed with particular emphasis on the planarizing character-istics, the absorption of the exposing wavelength, and the alignment accuracy. For e-beam lithography, spin-on indium tin oxide (ITO) has been developed to prevent the charging up of e-beams in thick bottom layer. High resolution and good CD (critical dimension) control is achieved on the topographic substrate in optical and e-beam lithography.

In order to develop new generation VLSI devices ( 4 - 16 M Dynamic RAMS ), a submicron fabrication technology is strongly desired. With conventional optical lithography methods, there is a inherent limit of 0.5 - 0.7 μm linewidths. Focused Ion Beam ( FIB ) lithography is expected to emerge in the future because of its many advantages for submicron lithography. Both a submicron Si MOSFET with a high switching speed and a GaAs MESFET with a 0.25 μm gate have been fabricated using FIB lithography technology.

Laser Irradiation of solid targets has been demonstrated to produce an intense source of soft x-rays. Conversion efficiencies of greater than 10 percent of the laser radiation into x-ray emission at energy over 1 keV radiation into 2n ster. have been demonstrated with sub-joule laser pulses. This high conversion efficiency is achieved by using sub-nanosecond pulse lengths, tight focusing, and shaped pulses. A laser capable of producing the requisite high quality laser pulses at 100 Hz repetition rates has been constructed using newly developed slab techniques to maintain good beam quality at high average powers. A windowless target chamber has been designed to mate this source to x-ray stepper/aligners. The present laser permits average x-ray intensities of 2 mW/cm² to be realized at the resist behind typical current BN x-ray masks, with negligible penumbral effects. This is almost an order of magnitude higher than possible with conventional electron beam impact sources. The present source, as well as upgraded laser systems based on this concept, should be a significant factor in the commercialization of production x-ray steppers.

The technology for manufacturing boron nitride x-ray masks used with a palladium anode x-ray source is described. There are important relationships between LPCVD boron nitride process parameters and critical membrane properties. The LPCVD process variables affect the alignment wavelength transmittance, membrane tension, and tension uniformity, the latter properties having significant influence on mask flatness. Subtractive patterning of the gold x-ray absorber layer is easily accomplished using tantalum layers as etch stops. Additive patterning by electroplating is a promising technique for producing x-ray masks for submicron imaging.

A VG HB5 STEM has been modified to allow computer controlled contamination writing on thin Carbon substrates. Line widths of 20 nM and height to width aspect ratios of better than 5:1 can be drawn. This writing has been used to fabricate Fresnel zone plates. A patching process is involved with registration to an accuracy of 2.5 nM. Methods and techniques have been developed to compensate for non-linearity of the field scans together with specimen drift. Zone plates of 50 NM diameter and outer zone widths of 40 nM have been con-structed.