E-beam lithography systems are being employed as pattern generators in the semiconductor manufacturing process. The TV-like raster exposure of a scanning electron beam under computer control provides the resolution, flexibility, and accuracy needed for the generation of high density integrated circuit patterns. E-beam mask making has become the technology of choice, while e-beam direct writing on the wafer has remained largely a niche application. Here the throughput handicap of serial exposure presented an economic hurdle, which limited applications to exploratory research and prototyping rather than manufacturing--with few exceptions: IBM, for example, has applied its internally developed high-throughput EL-series systems worldwide in large scale manufacturing of ASIC-type bipolar logic products. The recent progress in state-of-the-art of manufacturing-oriented e-beam systems for mask making and direct writing, together with results achieved with these advanced systems, is the subject of this paper.

The MEBES IV Electron Beam Lithography System was developed to meet requirements for advanced maskmaking. This necessitates support of 16- and 64-Mbit DRAM production and early development of 256-Mbit DRAMs. Using the original MEBES concepts and Etec's many years of experience with MEBES manufacturing, several major subsystems were redesigned, including the electron beam column and electron source. New test methods and test patterns were also developed to characterize system performance. As a result of the combined efforts of Etec Manufacturing and Engineering, a number of MEBES IV systems have already been built and tested. This paper provides a brief description of the MEBES IV systems. The new test patterns and methods are discussed. System performance data collected during factory acceptance of MEBES IV-LaB6 and -TFE (thermal field emission) systems are also presented.

Fabrication of integrated circuits at subhalf micron geometries is currently feasible only using advanced lithography technologies such as direct write e-beam and x-ray systems. These systems are extremely expensive and have low effective throughputs for a production environment. A mix-and-match approach using optical steppers for noncritical levels can dramatically increase productivity and control total lithography costs. A major impact for mix- and-match lithography is the total root mean squared alignment errors between systems. Implementation of a larger overlay budget to accommodate mix-and-match errors adversely impacts design rules for submicron technologies. However, a maskless lithography tool such as direct write e-beam offers the potential to compensate for systematic errors during wafer patterning and dramatically reduce the overlay budget for those layers. At TRW, a mix-and- match scheme has been developed between a Hitachi HL-700D e-beam direct write system and a Ultratech 1500 wide field 1X stepper. A previous analysis using only the linear distortion terms between these systems was found to be inadequate to fully explain the observed overlay. In this study, both linear and higher order distortion components are extracted using a large number of distributed measurement sites in the stepper field. These distortion terms are then analyzed to determine their source. Compensation techniques including both system hardware adjustments and e-beam software are investigated to enhance registration capabilities.

A high precision electron beam lithography systems which can be used to make reticles for 0.3 micrometers devices has been developed. This system is an enhanced model of the Hitachi electron beam lithography system, HL-700M. Key technologies used in this system are (1) the minimum address unit (0.0125 micrometers ) and the stage-positioning measurement unit (0.005 micrometers ) to correspond with higher precision specifications, (2) the refined beam correction functions and (3) the efficient environmental controls. The items of improvement on environmental controls are to design an anti-vibration column and to adopt a reticle temperature control system. The main specifications of this system are (1) the positioning accuracy: 0.06 micrometers , (2) stitching accuracy: 0.05 micrometers , (3) pattern width accuracy: 0.05 micrometers and (4) throughput: 0.5 reticle/hr. The system enables one to write phase shift masks using the direct writing mode in HL-700D.

The possibilities of sub-quarter-micron pattern fabrication by e-beam lithography with single- layer resist was studied on 0.5 micrometers thick W x-ray mask absorber. Calculation was made to evaluate the parameters determining the e-beam dose profile in the resist. It was found that at the incident energy of 40 keV pattern contrast in the resist, whose thickness is 0.2 micrometers , is homogenized through the depth. The experimental result proved that 0.15 micrometers line/space can be resolved by using a high contrast resist with this thickness.

Electron beam lithography for the fabrication of 1X x-ray masks and optical lithography system such as Markle-Dyson system requires precision of edge placement of mask less than 100 nm. In this paper we report experimental results with low voltage exposures (2 keV). 100 nm features were delineated in 66 nm resist in both sparse and dense patterns on bare silicon and 300 nm Au/Si confirming the predicted dramatic reduction in proximity effects. In addition, both sparse and dense 100 nm patterns were resolved over doses from 16 to 28 (mu) C/cm2 on bare silicon and 14 to 27 (mu) C/cm2 on gold substrate illustrating the high dose tolerance enjoyed at low voltages. Electromagnetic interference and the lower electron beam brightness are not limiting factors for the low voltage exposure.

A steered beam lithography will represent an essential part of the technology to meet the future need for ultra-high resolution mask making and direct write. Conventional high voltage e-beam lithography is being developed to meet these challenges. However, there are a number of physical limitations (proximity effects, resist sensitivity) which must be overcome. To do so will prove to be extremely expensive if in fact these problems can be overcome. There are significant advantages in going to extremely low energies in e-beam lithography. Proximity effects are eliminated although the electron-optics become more exacting. The need to focus a low energy e-beam can be eliminated by maintaining a sharp tip close to a surface as in a scanning tunneling microscope (STM). We have demonstrated technologically useful lithography with the STM operated with 4 - 50 V between tip and sample. Patterns have been defined in e-beam resists and by selective oxidation of semiconductor substrates under the action of the STM tip. In both cases the pattern can be transferred into the substrate with a dry etch. Sub 50 nm resolution is routine on a variety of substrates. A viable lithographic technology has been demonstrated in the research laboratory. However, several key issues must be addressed to develop a technologically viable lithography system compatible with existing microfabrication practice. These issues include: registration using the imaging properties of the STM for alignment, pattern accuracy and throughput. Advances in STM speed are described and suggestions made for improving lithographic performance with multiple sharp tips (each with an independent servo loop). The potential pay-off is high as a low voltage lithography tool will involve significantly less capital investment (and support cost) than the next generations of high voltage e-beam lithography tools.

The manufacture of state-of-the-art integrated circuits uses UV optical projection lithography. Conventional wisdom (i.e., the trade journals) holds that this technology will take the industry to quarter-micrometer minimum features sizes and below. So, why bother with X-ray lithography? The reason is that lithography is a 'system problem', and proximity X-ray lithography is better matched to that system problem than any other technology, once the initial investment is surmounted. X-ray lithography offers the most cost-effective path to the future of ultra-large-scale integrated circuits with feature sizes of tenth micrometer and below (i.e., gigascale electronics and quantum-effect electronics).

In our previous paper which we presented here two years ago, we described the ALF (Advanced Lithography Facility), IBM's new facility for X-ray lithography which was built as an addition to the Advanced Semiconductor Technology Center at IBM's semiconductor plant in Hopewell Jct., NY. At that time, we described the structure, its utilities, facilities and special features such as the radiation shielding, control room, clean room and vibration resistant design. The building has been completed and occupied. By the time this paper is presented the storage ring will be commissioned, the clean room occupied, and two beamlines with one stepper operational. In this paper we will review the successful completion of the facility with its associated hardware. The installation of the synchrotron will be described elsewhere. We will also discuss the first measurements of vibration, clean room cleanliness and the effectiveness of the radiation shielding. The ALF was completed on schedule and cost objectives were met. This is attributed to careful planning, close cooperation among all the parties involved from the technical team in IBM Research, the system vendor (Oxford Instruments of Oxford England) to the many contractors and subcontractors and to strong support from IBM senior management. All the planned building specifications were met and the facility has come on-line with a minimum of problems. Most important, the initial measurements show that the radiation shielding plan is sound and that with a few modifications the dose limit of 10% of background will be met. Any concerns about an electron accelerator and synchrotron in an industrial setting have been eliminated.

Preliminary evaluation of a 1:1 proximity x-ray stepper, built by Hampshire Instruments is discussed here. This stepper, model 5000P, is the first commercial system that uses a laser- generated plasma x-ray source. It was extensively tested at the supplier's facility and in July 1991 it was shipped to AT&T Bell Labs, where its installation is nearly completed.

Many requirements of a point-source x-ray membrane mask are unique, in comparison with masks used with synchrotron sources. Membranes must be thinner, flatter and stronger, for example. We characterize our choices of polysilicon and silicon-rich nitride for membranes, and a simplified one-piece (monolithic) mask blank. We use a dual strength testing technique which discerns whether membrane strength is limited by defects in the film growth process, or defects in the attachment of membrane to support. We find our membrane films are stronger than previously thought, and that improvements in membrane/support attachment will effect a more durable membrane mask.

X-ray lithography using a synchrotron light source has received considerable attention in recent years as a method for producing semiconductor device dimensions smaller than 0.35 microns. A number of synchrotrons or Electron Storage Rings (ESR) have been built around the world as possible light sources for lithographic applications. IBM has built its Advanced Lithography Facility (ALF) for the purpose of exploring synchrotron based X-ray lithography for device manufacturing. The ALF has the superconducting HELIOS compact storage ring, built by Oxford Instruments Ltd, at its center. The subject of the present paper is the design of the beamlines which connect this synchrotron light source to the Step- and- Repeat Aligner/exposure (SRA) tool where the device wafers are exposed to the synchrotron light.

As part of the development of a synchrotron-based x-ray lithography beamline, progress has been made in designing an optics system which meets the current industrial requirements. The primary goals of the optics are to achieve 0.25 micrometers line resolution over a 50-mm-wide by 25-mm-tall exposure field. The beamline optics must also provide a 5% power uniformity over the exposure field and enough power to achieve a 60-wafer-per-hour throughput with 20 exposure fields. In this paper, a description of a beamline optics system is presented for use with the Superconducting X-ray Lithography Source (SXLS) being built by Brookhaven National Laboratory with assistance from the Grumman Corporation. The basic design approach, performance characteristics, and some design data will be discussed.

This paper presents the transient modelling of an x-ray lithography mask undergoing continuous wave x-ray exposure. Two- and three-dimensional finite element models are developed to analyze the heat transfer mechanisms in the mask-wafer-holder assembly system. An additional structural model is used to calculate thermally-induced stresses and distortions (in-plane and out-of-plane) as a function of time. Parametric studies are performed to identify the effects of the Gaussian beam width, scanning frequency, exposure time and proximity gap setting on the dynamic response of the membrane.

The definition of sub-half-micron gates for gallium arsenide (GaAs)-based field effect transistors is generally performed by direct write electron beam lithography (EBL). Because of throughput limitations in defining large geometries by EBL, the gate-layer fabrication is conventionally divided into two lithographic processes where EBL is used to generate the gate fingers and optical lithography is used to generate the large area gate pads and interconnects. As a result, two complete sequences of resist application, exposure, development, metallization and lift-off are required for the entire gate structure. We report a new hybrid process, referred to as EBOL (electron beam/optical lithography), in which a single application of a multi-level resist is used for both exposures. The entire gate structure, (gate fingers, interconnects and pads), is then formed with a single metallization and lift-off process. The EBOL process thus retains the advantages of the high resolution E-beam lithography and the high throughput of optical lithography while essentially eliminating an entire metallization/lift-off process sequence. This technique has been successfully applied to metal semiconductor field-effect transistor wafers containing devices with dual 0.25 X 75 micron gates connected to 75 X 75 micron gate pads by 5 X 25 micron interconnects. The yields on these wafers have been very high with transistors averaging cutoff frequency values of 42 GHz and transconductance values of 366 mS/mm. Thus, the gate-layer process has been simplified without loss in yield or device performance. We will discuss the entire EBOL process technology including the multi-layer resist structure, exposure conditions, process sensitivities, metal edge definition, device results, and comparison to the standard gate-layer process.

We fabricated 2.6 GHz surface acoustic wave (SAW) filters using an electron beam exposure system. The filter consists of an interdigital transducer of aluminum on piezoelectric material (LiTaO3). Since the LiTaO3 density was high, a large number of electrons were backscattered from the material when exposed by electron beams. When a conventional negative resist was used, the backscattered electrons degraded the image contrast so that resist residue remained on unexposed areas after development. The process margin, which indicates exposure tolerance, was calculated. The process margin was small for LiTaO3, indicating that it is difficult to suppress resist residue when negative resist is used and that a positive resist should be used. To obtain vertical walls of positive resist, we used a 0.4-micrometers double-layer configuration in which the upper layer had a lower sensitivity than the bottom layer. To prevent the piezoelectric substrate from breaking due to the pyroelectric effect, the resist pre- and post-baking temperature was raised gradually at a rate of 5 degrees/minute. The 0.4-micrometers line and space patterns required for a 2.6 GHz filter could be produced accurately. The fabricated SAW filter had a 2.58 GHz center frequency and an insertion loss of 5.0 dB.

A comprehensive investigation on field stitching errors and their effect on the single-mode characteristics of DFB lasers fabricated using e-beam lithography is presented. The stitching errors are associated with small-area, high-resolution electron beam exposure, which has the potential advantage of high-speed writing of laser gratings. Measurements show that the errors are composed of a systematic and a stochastic part. Their effect on the grain margin was simulated both for (lambda) /4 phase-shifted and optimized multiple-phase-shifted DFB lasers. Simulations show that the lasers are insensitive to the systematic part of the stitching errors if the number of errors is large enough. The stochastic part was found to give rise to a variation in gain margin of the DFB lasers. We have concluded that the field stitching accuracy in the high-resolution mode of a commercial system for electron beam lithography is sufficient to provide a high yield of single-mode lasers. However, it is essential that certain precautions are taken considering exposure conditions, and that a fault tolerant laser design is used. Devices with side-mode suppression ratios of up to 48 dB have been fabricated, confirming that small-area, high-resolution e-beam lithography is appropriate in the production of high-quality single-mode DFB lasers.

Silicon field emitter arrays (FEAs) have been fabricated with a unique orientation dependent etching process and oxidation sharpening process to produce uniform and reproducible single point gated structures. Electron emission currents from these single silicon field emitters have exceeded 20 microamperes with extraction gate voltages less than 200 V. These pyramidal field emitters have a 'cone' angle of about 70 degrees with a tip radius of curvature of about 100 angstroms. The gate metallization has been formed from a variety of materials, notably platinum and polysilicon. Similar electron emission results have been operated continually in an unbaked vacuum system in the multi-microampere regime for over 600 hours without a change in their emission properties. Significant numbers of individually addressed field emitters can be fabricated on small silicon chips, and many chips can be fabricated on a silicon wafer, thereby making the cost per chip reasonably low. Microstructural Einzel lenses have also been fabricated with 5 pole electrodes as well as arrays of deflectors. The combination of the technology of microstructural field emission and microstructural lenses and deflectors offer a unique opportunity for nanolithography, novel devices, and electron/ion microscopy. Microstructurally Integrated Lens and Emitter Systems (MILES) offer the potential for massively parallel electron beam applications and electron/ion source redundancy.

Dose correction is commonly used to compensate for the proximity effect in electron lithography. The computation of the required dose modulation is usually carried out using 'self-consistent' algorithms that work by solving a large number of simultaneous linear equations. However, there are two major drawbacks: the resulting correction is not exact, and the computation time is excessively long. A computational scheme, as shown in Figure 1, has been devised to eliminate this problem by the deconvolution of the point spread function in the pattern domain. The method is iterative, based on a steepest descent algorithm. The scheme has been successfully tested on a simple pattern with a minimum feature size 0.5 micrometers , exposed on a MEBES tool at 10 KeV in 0.2 micrometers of PMMA resist on a silicon substrate.

Experimental investigations and special software for the e-beam exposure system gives the possibility of decreasing the influence of proximity effects and field distortion. The results of creating a focusing element for the soft x-ray range are described: amplitude and phase- contrast Fresnel zone plates as well as reflected Bragg-Fresnel lenses on the base of multilayer x-ray mirrors.

Kinematically-mounted x-ray lithography masks are investigated to determine their response to imperfections in mounting, namely, misalignment of the clamping forces. Given the limited error budget for x-ray mask mounting, it is essential to minimize the mechanical distortions in the exposure area. Three-dimensional finite element models of the support ring, mask wafer and membrane are used to analyze the gravitational effects for both horizontal mounting (e- beam patterning) and vertical mounting (synchrotron exposure) coupled with the effects of misaligned clamping forces. In-plane distortions of the membrane are computed and the nodal displacements in the patterned area are uniquely mapped. Results of this study show the resultant membrane distortions due to imperfect mounting to be significant when considering 0.25 micrometers (or finer) technology.

A keV x-ray source based on plasma production using a high repetition rate picosecond KrF laser has been developed. By using a train of 50 ps input pulses to extract a KrF amplifier module an output train of pulses of energy up to 200 mJ has been obtained leading to generation of 1.15 keV x-rays from solid copper targets with a conversion efficiency from laser light to x-rays of over 1.5%. Iron and nickel targets yield softer x-ray spectra with higher conversion efficiencies of 2.3% and 1.8% respectively. When operated at 20 Hz in one atmosphere of background helium gas a point source of 1.15 keV x-rays with an average power of 45 mW is obtained. Initial characterization of the sensitivities of a positive resist, PBS, and a negative resist, Shipley XP90104C, have been carried out to demonstrate the effectiveness of the source.

The concept of X-ray generation in a Spherical Pinch machine was presented at the 1991 SPIE conference (Proc. SPIE, Vol. 1465, p. 318, 1991). The machine is a pulsed power device which drives strong spherically imploding shock waves to create a highly compressed hot plasma of small size which acts like a blackbody radiator. At the present time the machine is operating at an input electrical energy of 18 kJ and is capable of reaching 34 kJ. An efficiency of about 20% has been measured for the transfer of electrical energy to the plasma. The radiation output of the machine ranges over a broad spectrum which peaks near the deep UV (100 eV - 150 eV). Therefore the machine is useful for applications such as UV and deep UV lithography, in addition to soft X-ray lithography. The dosimetry analysis shows that about 100 mJ/cm2 of UV and deep UV and about 3 - 5 mJ/cm2 of soft X-rays per single discharge is available at a distance of 20 cm from the source. The radiation output is now being optimized in order to reach the design level of 10 mJ/cm2/discharge in the soft X- ray region of the blackbody spectrum. The advantages of the machine are, besides compactness and cost effectiveness, relatively long pulse duration (approximately 10 microsecond(s) ec) and low peak power at the mask and wafer. The machine is capable of operating at a repetition rate of 0.1 Hz. A new machine with higher input energy and repetition rate is being designed and constructed and is expected to generate about 12 mJ/cm2/discharge in the soft X-ray region of the spectrum at a distance of 30 cm from the source.

The resisting pressure of deep cylindrical beryllium (Be) windows was studied theoretically and experimentally. A cylindrical Be window (window size, 50 X 50 mm2; thickness, 20 micrometers ; radius, 30 mm) which was fabricated on the basis of the theoretical calculation sustained up to 0.43 MPa (pressure difference) in hydraulic pressure burst testing. Deflection of this foil was 0.2 mm at this pressure. A planar window (diameter, 38 mm; thickness, 30 micrometers ) sustained pressures in the region of 0.25 MPa to 0.49 MPa. Deflections of these foils were in the region of 0.9 mm to 1.4 mm. A deep cylindrical Be window was found to be superior to a planar one in 1 atm. (0.1 MPa) helium gas, in terms of its larger field size and higher transmittance of SR.

Measurements of dose characteristics of PMMA were performed with various resist temperatures. The quantitative data shows sensitivity changes with the temperature increase. The model of resist heating effect is presented and a possible technique to correct distortions due to resist heating is discussed.

Phase-shift photolithography is emerging as an important new technology for sub-half-micron design rule circuits. Unfortunately part of the price paid for the improvements in spatial resolution and process latitude afforded by phase-shift lithography is increased mask defect printing sensitivity. The minimum printable defect size, 0.3 microns (on the mask) for I-line steppers at 0.35 microns, is roughly half that for conventional photomasks. This paper examines the issues associated with extending high resolution focused ion beam mask repair to phase-shift masks.

We have developed an ion-assisted gas etching (IAGE) technique which enhances existing focused ion beam (FIB) failure analysis (FA) cross-sectioning techniques. Preliminary results show enhanced sputter removal rate of certain materials, decoration of material interfaces or boundaries, and preferential etching of materials particularly difficult to image with standard secondary imaging systems. In particular we show two examples of IAGE using XeF2. The first sample is a multilayer of resist on oxides through which contact holes have been made by reactive ion etching (RIE) to an implanted silicon layer. A typical FIB cross-section made without IAGE shows some delineation of boundaries between oxides and a slight cascade of the resist at the top edge of the section. An FIB section made while etching with XeF2 shows improvement in delineation of the boundaries between the oxides and in particular, decorates the implanted layer. To verify the preliminary evidence that implanted material can be selectively etched with XeF2, we used a silicon wafer implanted with As+ to a depth of 850 A, and the resist was stripped. Typical FIB sectioning could not distinguish between the implanted and nonimplanted regions on the wafer, either with scanning electron (SEM) images or scanning ion (SIM) images. However, there was a slight difference in the SIM image of the top surface of the wafer. A very low ion dose combined with XeF2 selectively etched the implanted and nonimplanted regions enough to distinguish them clearly in an SEM image. Thus we have preliminary evidence that IAGE with XeF2 complements existing FIB FA cross-sectioning techniques and has particular potential for defining boundaries of materials which typically have low contrast such as oxides, and as a way of imaging implanted regions.

This paper reports the results of efforts to find conditions under which FIB technology can be used to cross-section resist features. Evaluation of the accuracy of these cross-sections and the convenience or speed relative to cleaving is examined. FIB cross sections through a staggered window pattern on the edge of cleaved segment from a 0.4 micrometers thick novolac (mp 2400- 17) coated Si wafer showed that the ion beam milling relieves the taper present on the profile, and enlarges the dimensions of the smaller features (0.6 micrometers windows). These effects are due to the near-maximum sputter yield at the angel that the resist side wall presents to the ion beam, the size of the ion beam relative to the feature size, and the dose used to cut the section. Alterations in the resist geometry can be minimized by using the lowest possible doses to image and cut the desired features. Redeposition of sputtered material in the window openings does not appear to be a problem. All of the work in this study involves uncoated resist, for both FIB cutting and SEM imaging. A technique for sectioning line and space patterns and then measuring the line profile at any point on a full wafer has been worked out. This technique fulfills the potential speed advantage of FIB ion milling over cleaving. Features in a 7 X 7 array printed on a 12 cm wafer were ion milled at an angle of 45 degree(s) to the wafer surface by dose variation, giving a wedge-shaped trough in the wafer. When tilted to 45 degree(s) in the SEM, the angled wall of the wedge cut presents a surface that is normal to the electron beam, just as in the edge mounting of a cleaved section. Contrast enhanced digital images allow measurement of the line profile, which was found to be accurate within the limits imposed by the ion beam width.

High performance focused ion beam systems utilizing liquid metal ion sources are performance limited due to the chromatic aberration of the optical elements and the finite energy spread of the ion source. Some concepts are presented to reduce the chromatic and spherical aberrations utilizing axially symmetric electrostatic optical elements. Unlike paraxial optics, in off-axis optics the individual aberrations of each lens element interact. This allows the total system aberrations to be minimized by optimizing lens parameters. Fifth order raytracing is used to determine the off-axis aberrations of a two lens system. The aberrations are then calculated as a function of the inter-lens spacing and are shown to have distinct minima. By optimizing the lens acceptance angle and the lens spacing off-axis chromatic aberration and the geometric aberrations (including spherical aberration) may be reduced to values below that achievable using paraxial optics at the same beam current. Novel coaxial optical lens designs are presented which also offers the possibility of chromatic aberration correction and may provide improved performance over conventional optics for special applications.

A new resist processing is presented using the carbonization of organic polymer-based resists due to ion beam exposure and a following dry development under O2-RIE conditions. As an example for this process, a negative tone pattern transfer yielding a structure size down to 50 nm is demonstrated. The influence of the ion exposure dose and the O2-RIE conditions on the pattern transfer are discussed in this paper.

We demonstrate the flexibility of radical-beam ion-beam etching for anisotropic low-damage dry etching and in situ processing of GaAs/AlGaAs structures. High anisotropy and low damage are verified by the light-current characteristics of etched laser facets and from Schottky diode characteristics of etched surfaces. These data show that damage is reduced by increasing the ratio of chemical to physical etching components. We also demonstrate direct- write patterning by selective chlorine etching of focused ion beam-damaged GaAs. The depths and linewidths are measured by atomic force microscopy. The enhanced etching is correlated to the implant-damage distribution. A selectivity of 10 to 1 is attained for doses equal or above 5 X 1014 cm-2, which corresponds to the amorphization threshold in GaAs.

A positive tone chemically amplified resist is optimized for SOR x-ray lithography using an orthogonal response surface experimental design. Seven factors are manipulated in the matrix, including two resist chemistry factors and five process factors. Electrically probed linewidths are analyzed for statistical significance of main effects and interactions. The resist chemistry and process are coupled and advanced lithographic work must consider both together if optimized performance is desired. The shape and size of the Exposure-Gap tree for +/- 10% linewidth control is influenced strongly by resist chemical and process changes and is not just a function of the aerial image. Dose latitude is not correlated with resist sensitivity indicating stochastic effects don't affect the image quality significantly in acid catalyzed resist processing. There is >+/- 10% dose latitude available for quarter micron imaging over a 5 micron gap range. The work shows that a formulation and process exist which is compatible with quarter micron lithography with relatively wide process latitude and straight resist profiles were obtained with good sensitivity.

The intrinsic registration capability of current e-beam lithographic tools has to approach or even go below the 0.10 micrometers value ( + 3(sigma) ). However, the actual level to level overlay measurements are generally found above this limit as resist materials trap the electric charges: the electron beam is deflected by the electrostatic forces, resulting in misalignment and distortion of the delineated patterns. Large pattern displacements can be observed depending on the experimental conditions. This paper aims to compare various charge reducing processes when used under exactly the same conditions (20 keV industrial machine and trilevel resist structure). No system appears as a perfect solution. There are partial solutions, but none is simple and satisfying; this contributes to the picture of electron- beam lithography as a marginal technique for IC manufacturing.

Ground truth comparison between our advanced simulation modules and specific sets of experiments is presented, and simulation fidelity is evaluated. The resists described below are SNR248 negative chemically amplified deep-UV resist, KTI895i positive I-line resist, Shipley ANR negative chemically amplified I-line resist, and SMP1400 positive G-line resist. The simulation is composed of several modules: aerial image calculation, exposure, post-exposure baking (PEB) as reaction-diffusion equations, and dissolution employing reaction-diffusion equations as well as Hamilton-Jacobi systems. The experiments described display variations in exposure energies to determine process latitude and cd control, defocus, as well as line shape. Detailed comparisons between experiments and simulations are exhibited for a wide range of photolithographic applications.

Lithographers have steadily reduced exposure wavelength and increased numerical aperture (NA) to maintain process window and simplicity. The G-line systems of the 1970s gave way to the I-line systems of the late 80s, and then to the deep ultraviolet (DUV) systems of today. This paper describes our characterization of a DUV lithography system for the manufacture of 16-Mb DRAM chips at 500-nm ground rules. The process consists of a positive-tone, aqueous-base developable photoresist with an overcoat for sensitivity control, and an anti- reflective coating (ARC) on selected levels. The exposure tools used are step-and-scan systems with a 0.36 NA and expose bandpass of 240 - 255 nm. Apply and develop processes are clustered with the expose tool to minimize defects, reduce cycle time, and eliminate process variables.

We have fabricated 16M-DRAMs with i-line phase-shift lithography using 8 inch wafers. Edge-contrast enhancement type was used as a phase-shift reticle. The effects on i-line phase- shift lithography were evaluated for the patterns which have a small focus and alignment margin. It was found that the phase-shift lithography has a large effect on enlarging focus margin of small size patterns like 0.5 micrometers . We evaluated the effect of phase-shift lithography by a trial manufacturing of 16M-DRAMs. As a result, i-line phase-shift lithography was found to have a possibility of high yield production of 16M-DRAMs. However, it was also found that there were several rejected pattern widths which were caused by variation of shifter width in the phase-shift reticle. So it is necessary to improve these rejected patterns for the application of this method to 16M-DRAMs.

Demands on advanced exposure tools for ULSI applications have increased rapidly during the last few years. Overlay accuracy, one of the key subjects, has been under continuous development. In X-ray lithography the overlay is mainly determined by pattern displacement on X-ray masks, by process-induced wafer distortions and by stepper-related contributions, generated by a limited alignment accuracy and displacements during the pattern replication. It is the aim of this paper to characterize the stepper-related contributions to the overlay budget. For this purpose we investigated the overlay performance of vertical XRS-200 type X-ray steppers equipped with an advanced optical alignment system. Using a multiple exposure technique, a series of alignment experiments were performed on typical CMOS layers. The results show that the wafer chuck flatness has, beside the alignment accuracy, a major influence on the overlay accuracy.

We analyze the behavior of diffraction integral with the Fresnel number of the order of 0.1 to 5, and discuss the effect of diffraction on the pattern transfer in the x-ray proximity printing process. We investigate the cases of an isolated single space and an isolated single line. The diffraction patterns are constructed for various values of F. The width of the pattern is examined as a function of the minimum dose to develop the resist, which is, in turn, a function of the threshold intensity Ith. We search for the range of F in which the linewidth variation is within +/- 10% of its maximum (or minimum) value. Since 1/F is proportional to mask-to-wafer gap z, the widest range of allowed F value can be adopted as an optimum range. The optimum condition satisfying both the cases is found. The results show that there are the optimum ranges of the mask-to-wafer gap depending on either the linewidths of the mask patterns or the system throughput requirement. Our discussion on the diffraction effect is carried out mainly in terms of the Fresnel number.

The rationale, design, component properties, and potential capabilities of extreme-ultraviolet (XUV) projection lithography systems using 60 - 80 nm illumination and single-surface reflectors are described. These systems are evaluated for potential application to high-volume production of future generations of gigabit chips.

Practical considerations that will strongly affect the imaging capabilities of reflecting systems for extreme-ultraviolet (XUV) projection lithography include manufacturing tolerances and thermal distortion of the mirror surfaces due to absorption of a fraction of the incident radiation beam. We have analyzed the potential magnitudes of these effects for two types of reflective projection optical designs. We find that concentric, symmetric two-mirror systems are less sensitive to manufacturing errors and thermal distortion than off-axis, four-mirror systems.

Based on the simplified yet accurate model we developed, a route to an optimized x-ray lithography system is defined. It predicts a large exposure window with a less coherent illumination than those currently used in most synchrotron radiation based exposure systems. Two different schemes of coherence reduction are proposed and their effects are simulated.

We report on a new modeling tool for the prediction of the cumulative dose delivered to any given layer of a semiconductor process involving x-ray lithography (XRL) steps. In such a process, the layers receive various doses at different steps in the manufacturing. In order to determine the total dose delivered at any given level it is necessary not only to compute the dose for the lithographic step of that particular level, but also to keep track of the past history of the process in order to accumulate the total radiation. For this, one must know the full process, including sacrificial layers, resist layers and layout geometries, since different areas of the surface being patterned will receive different doses because of shielding by other layers. We apply the model to the case of a standard 0.5 micrometers NMOS device process and to an advanced 0.1 micrometers NMOS process. Results will be presented and discussed in detail, for the various assumptions of the lithographic process. The program, written in C under UNIX, is compatible with standard process modeling tools and device layouts.

We report the first results on the performance of the new beamlines installed at CXrL. Three new lines have become operational in 1991, bringing the total of ports in use at the Center to five. The High-Flux beamline (Exposure Station #2) was commissioned in November 1991, and delivers a high-intensity spot of radiation for accelerated mask lifetime studies. Of the new beamlines, the Double Toroid beamline ('ES-3') is the one designed and built for the advanced 0.25 micron stepper, currently under development. The beamline employs an optical scanning method, and delivers a beam with a 2 mm FWHM in vertical, and a designed uniformity better than 2% across the 25 mm horizontal field. A 13 micron curved beryllium window is used to extract the beam in helium. First exposures were achieved in October 1991, and the beamline is now used to illuminate the Two State Alignment (TSA) system. As the name implies, the Karl Suss Stepper beamline ('ES-4') is the beamline designed and built to deliver radiation to a Karl Suss XRS-200 stepper, on loan from the IESS of Rome, Italy. This line was completed in November 1991, and the stepper was installed one week later. All of these three beamlines have exposure stations that are enclosed in a Class-10 cleanroom, in order to provide a contamination-free environment. We will present data on the characteristics of the radiation delivered to the exposure systems and on the initial performances of the various systems.

We describe the implementation of the two-state alignment system into the CXrL aligner, which is developed at our Center for X-ray Lithography. The CXrL aligner is designed to expose sub 0.25 μm feature size integrated circuits. The aligner consists of a three-axes two-state alignment system for alignment error detection and a piezo based precision mechanical stage for alignment error correction. The wafer is held by a precision vacuum chuck, while the mask is held by three vacuum suction cups located around the glass ring. In the prototype, the mask to wafer relative positioning is achieved by 3 motorized stages (for gap setting) and 3 piezo-actuators (for lateral alignment). Since the optical system is designed to be located outside of the synchrotron radiation path, alignment can be performed during exposure. We present the results of the alignment system performance, such as noise equivalent displacement and alignment signal response time. An alignment signal repeatibility of much better than 3σ = 0.07μm is achieved. We also briefly describe the future evaluation of the system, such as overlay measurement of the system using verniers and SEM inspection of some specially designed patterns.

As tolerance as a percent of feature size increases for sub-micron technologies with increased scaling, yield loses due to circuit performance fluctuations will increase. Therefore for sub- micron technologies a tradeoff has to be made between circuit performance yield and the purchase of more expensive processing equipment that can more tightly control critical dimensions. At the same time, the development time of a circuit that is to be manufactured on a process with higher parameter tolerances will increase, and this has to be traded off with the process development time needed to reduce tolerances. In this paper, the performance yield problem for sub-micron technologies is addressed, as it relates to tolerance in geometric feature sizes and alignment. Using a statistical model of process fluctuations, examples are presented showing that different tolerance requirements are needed for different circuits.