A stepper of new concept has been developed. It uses double telecentric lens and TTL laser scanning method. It is TTL on axis stepper in the ture sense, because alignment is checked directly before exposure. The projection optics can resolve 1.0 μm in production and has small distortion. The alignment accuracy within 0.25 μm ( 3 o) is achieved.

The Laser Step Alignment (LSA) is a through-the-lens, field-by-field alignment system developed for an optical step-and-repeat system with a reduction projection lens (wafer stepper) intended for one micron or submicron lithography. The laser beam used for LSA is focused on a wafer through the projection lens. The alignment marks on the wafer are in the shape of diffraction gratings. The mark detection is performed by moving the wafer stage so that the alignment mark travels across the laser beam. The passage of the align-ment mark produces diffracted and scattered light, which is photo-electrically detected as the alignment signal. The position of the alignment mark is accurately measured by the laser interferomter attached to the wafer stage. The LSA system achieves overlay accuracy (20) less than 0.15 Pm for processed wafers. LSA uses a 633 nm He-Ne laser, so the photoresist is not sensitive to it, and LSA is well applicable to multilayer resist processes. The alignment time is about 0-6 seconds per shot in field-by-field mode. This LSA is attached to the Nikon wafer stepper (NSR), so field-by-field alignment, several types of alignment, and other measurement for compensation are performed. An appropriate alignment may be selected to meet the accuracy and throughput required in each case.

Optical lithography using an i-line wavelength (365 nm) has superior resolution compared to longer g- or h-line wave-lengths. This is due to the inverse relationship between the wavelength of the light and the diffraction-limited resolution of an imaging lens. The higher absorptivity of typical positive photo-resists at the i-line wavelength also enhances resolution. The half micron resolution capability of the Zeiss i-line 10:1 projection lens has been reported in a previous papers. This paper describes the performance of the THE 800SLR® Wafer Stepper with an i-line imaging lens. The i-line optics for the wafer stepper have presented difficulties in the areas of illuminator optimization, i-line spec-tral filter development, and wafer-to-reticle alignment. The reduced linewidth (0.5 micron) requires tighter system tolerances in the performance of the wafer stepper for linewidth control. The error budget to achieve these tight tolerances in the optical, mechanical, and electrical areas explained later in this paper. High throughput combined with submicron system capability enables i-line optical lithography to be superior to other non-optical lithography methods in production applications.

The performance of a wafer stepper alignment system based upon phase gratings etched in the silicon has been investigated both experimentally and theoretically. Measurements of the alignment accuracy as a function of the light intensity reflected from the wafer show that the accuracy is better than 0.1 micron down to reflectivities as low as 0.15%. The various factors influencing the alignment signal by processing layers have been analyzed. A standard marker geometry is given, which produces a 0.1 micron alignment accuracy for the vast majority of processing conditions. For some special cases such as planarizing resist layers a different marker geometry is preferrable.

A feasibility study of 1X optical stepper for one micron lithography in a manufacturing environment is conducted. The present discussion concentrates on the availability and per-formance of production-worthy 1X optical stepper, and the manufacturability of one micron reticle including the 0.5μm defect inspection. The Ultratech Stepper Model 1000 with one micron lens and the KLA-208 automatic reticle inspection system are evaluated and their performances are discussed.

VLSI products require extremes in lithographic capabilities. Several masking levels typically demand leading-edge lithographic capabilities, while a number of other levels typically have less stringent requirements (approximately 5pm). Similar statements may be made regarding typical overlay requirements. This leads to the consideration of mixing 1X aligners with step-and-repeat 5X reduction systems. The objective of this procedure is to minimize costs and meet lithographic requirements with increased real-time throughput. There are three key issues to be dealt with: 1) Mixed tool set overlay, 2) Relative process/tool cost considerations, 3) Logistics of line operation. This paper describes mixed-tool set overlay results and a description of the techniques used to minimize them.

The principal requirements for a Submicron CMOS process are the ability to generate submicron geometries, and the ability to register the several masking layers closely enough to take full advantage of the benefits of those geometries. This paper discusses the suitability of a GCA 4800 Wafer stepper to fulfill those requirements.

The effect of reticle sizing has been characterized for a single layer resist process employing a hotplate postexposure bake. Results are presented which show optimum CD control is obtained for feature sizing .25 to .3 microns less than the reticle size for clear field reticles. Process margins relative to exposure dose, topography and focus are presented.

A direct-reticle-referenced die-by-die alignment system operating at the actinic wavelength has certain unique advantages and some advantages in common with other systems. The unique advantages are the elimination of baseline and baseline associated problems and the minimization of optical aberrations of the alignment signal. The other advantages are the capability of video collection, digitization and manipulation of large data bases, the opportunity to chose and vary algorithms to accommodate processed wafers and the minimal demands imposed upon processing. The cost of using the actinic wavelength is the limitation in the number of photons available for alignment and the resultant reduction of the signal-to-noise ratio. There are certain problems generic to most other die-by-die alignment systems. Both optical and mechanical complexity is added to the lithographic system. Thin-film structures must be tailored to accommodate the demands of contrast, and photoresist modulation effects can introduce uncertainties in alignment. We shall briefly describe these effects in the context of the Field Alignment System, describe some of our measurements on this system, particularly those pertaining to the efficacy of the algorithm and comment upon the alignment capability of this system.

A retrofit illuminator for the High Performance Condenser for the 100, 200, and 300 series Perkin-Eimer Micralign projection aligners is discussed. Comparisons of the major performance features between the two systems is made shoeing dramatic improvements in wafer through-put, a reduction of nearly 2,000 watts of electrical power consumption, reduction of the arc slit width, and significantly reduced maintenance time. Some SEM's of typical lithography are also presented.

A Contrast Enhancement Material (CEM), G.E. Altilith CEM-388, has been gaining acceptance on semiconductor fabrication lines using h line step-and-repeat systems. Although a bilevel system, the use of CEM-388 is relatively simple, and has shown significant increases in practical resolution when used at 405 nm. A considerable advantage may also be gained by using this system with Micralign projection printers, especially in improving resolution over steps. This paper will present the practical aspects of using CEM-388 in such systems and illustrate the effects on resolution, step coverage, C.D. control, and wafer throughput. The presence and control of inhibition effects will also be discussed.

The effect of atmospheric pressure variations on the reduction ratio of a stepper lens is discussed. Data for a 5X lens is presented which demonstrates a dependence of the reduction ratio on the atmospheric pressure. The dependence can lead to a .3 um error in the reduction of an 11 mm field. A model based on thin lens optics is also presented which predicts the observed trends. Methods of minimizing the impact of this effect are discussed.

We describe a method which minimizes the time required to set up and sustain optimum overlay performance of the DSW step-and-repeat system throughout a product wafer run. Improvement in both system performance and productivity is realized. Fundamental to the technique are the latent image formed by the exposure of photoresist, the AWA' global alignment system and the laser-metered stages of the DSW® system. We eliminate the need to develop wafers and to read verniers during the DSW® system set-up. Furthermore, we demonstrate the potential for on-the-fly calibration using product reticles and product wafers. The approach is generally applicable to the overlay calibration of optical exposure tools.

The use of narrow band illumination for automatic alignment can result in thin film interference effects that can make optical alignment difficult. Depending upon the thickness of the photoresist and the types and thicknesses of the underlying film structure, the reflectivity of the alignment marks are enhanced or decreased. This can also produce fringes of alternating light and dark areas that interfere with the system's ability to recognize the alignment marks accurately. Small changes in the thicknesses of the photoresist or the underlying films can result in large changes in the quality of the alignment marks. In the past, most efforts to overcome these problems have relied upon efforts to accurately control the thickness of the optical films, bleaching of the photoresist, or changing the wavelength of the alignment illumination. This paper describes a technique that is used to assure that good alignment mark images are obtained under a wide range of film thickness variations. This technique has been developed specifically for use with Eaton-Optimetrix wafer steppers, but the principles are applicable to other types of steppers. The technique that has been developed takes advantage of the fact that small changes in film thickness will often cause a poor optical image to become good. The alignment marks for this wafer stepper have been divided into two separate areas. Through the use of a simple technique, that does not necessarily require any additional process steps, the optical thickness and/or structures of these two areas are made to differ from each other. In most cases, at least one of the two areas will produce an alignment image of good quality. During alignment, the better of the two alignment mark areas is used by the stepper for aligning the wafer to the projected image of the reticle. This paper discusses the theory and application of the new alignment mark (Double Structured Alignment Mark). The characteristics of the alignment marks are evaluated by comparing the theoretical results with plots of actual data. Test results and photographs of the actual alignment mark images are shown to verify the effectiveness of the technique.

The need for automatic defect inspection of photomasks has been well recognized by the semiconductor manufacturing industry. Indeed, commercial equipment for this purpose has been available for many years. The method used in the commercial equipment is based on the point-by-point comparison of one die with an adjacent die or with the data base, in a serial fashion. As the minimum features of the production integrated circuit (IC) pattern approach micrometer and submicrometer dimensions and the pattern area increases to about 150 mm in diameter, the inspection time per photomask by this method becomes formidably long, sometimes hours. To reduce the inspection time, radically different inspection methods are clearly needed. One such method is based on an optical information processing technique.1'2 This method has an inherent advantage: the IC pattern and the defect information over the entire photomask are processed simultaneously, or in parallel. However, several difficulties in the past have prevented this method from being practical. Under the sponsorship of Insystems, scientists at the University of Daytsn Research Institute were able to circumvent these difficulties by means of holography.'. Based on the successful results achieved there, Insystems undertook the task of developing an automated photo-mask inspection system, the Model 8405. This system is designed for turn-key operation in IC manufacturing facilities. Major performance specifications of the system are listed below. . Acceptable photomask sizes: standard sizes or others up to 178 mm x 178 mm (7" x 7") . Acceptable photomask thicknesses: standard thicknesses or others up to 6.4 mm (0.25") with or without pellicle . Detectable defect sizes: 0.5 um or larger with or without pellicle . Automatic inspection time: 7 minutes or less for 100 mm x 100 mm photomasks . Hologram or filter production time: approximately 6 minutes . Defect report formats: include computer printout of defect map (indicating defect types, sizes and locations), video terminal displays of holographic defect image and corresponding microscope image The purpose of this paper is to describe this system and present some preliminary results.

The concept of VLSI chip fabrication in a step and repeat lithography based facility varies significantly from that of the conventional 1:1 aligner based process. Whereas a 1:1 mask contains over 100 possible chip sites covering a five inch wafer, a 5:1 stepper reticle may contain as few as one chip site per exposure with two to four being more typical. This small reticle array is the stepped across the wafer to fill all possible chip postitions on the wafer. If a printable defect exists on the reticle then the final dis yield will directly experience a loss of anywhere from 25% (four chips per reticle) to 100% (one chip per reticle) of the potential chips in all lots exposed with that reticle.

Inspection of semiconductor pattern images against their database definition is increasingly used for validation of stepper reticles. Systems like the Cambridge Instruments Chipcheck use transmitted visible light optics to generate a high contrast video signal from the etched chrome plate image. This type of system can quickly and effectively "catch" repeating type defects which would be missed by die to die type comparators. More recently, the incident illumination capabilities of Chipcheck have been utilised to inspect developed resist images for wafer stepper reticle validation prior to production flow. This technique has many operational benefits compared with manual inspection or "glass wafer" techniques both in speed and accuracy.

Current state-of-the-art instruments for automatic visual inspection of masks, reticles, or wafers, are targeted to inspect patterns with line widths as small as 1.5 micron (gm) and defects as small as 0.5 micron. New near term targets for production inspection machines will require inspection of 1.0 micron line width patterns and 0.3 micron defects. The types of inspection machines which will adequately address these new targets will require very robust image analysis algorithms. Some current instrumentation will predictably fail in the sub-0.3 micron to sub-0.5 micron defect size range due to fundamental limitations of traditional image acquisition and processing methods (e.g., pixel to pixel comparison modes). Two studies were carried out to see how defect detection might be improved using a new approach exploiting adaptive signal processing, artificial intelligence methodologies, and CAD directed model matching. The first study investigated the current practice/ limits of defect detection; measuring the reliability with which defects could be detected, measured, and correlated with a CAD data base. We conclude that defects greater or equal to 0.5 micron can be consistently detected at a high (>90Z confidence level); with defects as small as 0.3 - 0.4 micron being detected routinely. In the second study, we explored the realistic potential of defect detectability of the new approach. In this experiment we report very encouraging experimental data extending to line widths of 0.5 micron, and defects below 0.2 micron.

Several specific aspects of a new multi-functional automatic wafer inspection system are discussed and characterization data presented. Li-near dimensional measurement data are presented illustrating system precision and accuracy. The probability of detecting different types of defects is presented for two different pixel sizes. A new method of providing a design or CAD reference, one of three alternative references, is shown.

A computer program has been developed for calculating statistical layout design rules for integrated circuits. The program also calculates the sampling plan required to verify the confidence levels of alignment and circuit element dimensions at develop inspection. The inherent speed and computational accuracy of the computer offers the user a choice of confidence levels depending upon circuit configuration and critical alignments in the process. Finally, the program calculates overall circuit yields as a function of lithography limited margin violations.

Photomasks are in widespread use in production of integrated circuits either in the form of reticles or step-and-repeated working masks. The quality of these masks depends on control in both line dimensions and defect density. Unders'tanding of defect generation mechanism has been a topic of extensive study in the past. In this paper, defect caused by over-lapping flashes, exposed by an optical pattern generator on emulsion photoplates, has been reported. It has been observed that pattern generator data tapes produced by computer-aided-design (CAD) systems often contains flashes that overlap in edges or corners. This causes a distortion in line demensions and can be a serious limitation when the design rules in mask layout are quite tight. In this paper, we have discussed the effect of this overlap, its causes and possible remedies. It has been further shown that the removal of overlapping flashes from critical areas can improve the quality of the reticle.

This paper discusses an optical technique capable of reliably measuring registration to few hundredths of a micron on virtually any layer. It has overcome the accuracy limitations and the proximity effect error present in other optical techniques. In spite of the recent popularity of highly accurate computerized electrical registration measurements optical registration measurements are still popular and necessary. This is because nonconductive layers, including resist, cannot be measured electrically. A quick optical technique with a high degree of accuracy has been developed and put to use. The measurement is made through a microscope, and a computer pattern recognition follows. It is free of the resolution limits inherent in such structures as optical verniers, which are bound to typically 0.1 microns by the grid size used to make the mask. This method employs a direct optical misalignment measurement between two matching structures and is capable of resolving 0.01 microns. It is also free of the proximity effects which make many verniers and pattern recognition schemes inaccurate. Proper microscope calibration, adjustments, and pattern recognition algorithms are key in making this technique work. The apertures must be accurately aligned and the focus properly adjusted to provide the right image. An HP 9000-226 computer has been custom interfaced to a Leitz microscope and a set of algorithms written. The result is a highly accurate, fast, and user friendly optical measurement system capable of measuring registration on all layers.

This paper deals with characterization of a pilot line photo process using an electrical linewidth measurement system.* A functional process for set up of the system is described, as well as the test structures used. Evaluations conducted include overlay stability of a Perkin-Elmer 542 and adjustment, overlay of a Perkin-Elmer 542 to three Perkin-Elmer 240s, overlay to a Perkin-Elmer 542 in another wafer fab, a daily linewidth evaluation on a Perkin-Elmer 542, and a comparison study of the electrical prober to other types of linewidth measuring systems. It will be shown that this tool, with precision in the .001 pm range (3 Q) is capable of providing correlation data for other types of quick feedback, smaller data base systems for linewidth measurement, as well as providing a data base for the proper evaluation of photolithographic processing.

Lack of precision and accuracy of in-process critical dimension (CD) measurements of linewidth continues to be a serious problem at micrometer and submicrometer dimensions. Even with highly repeatable optical linewidth measurement systems, variable "offsets" or errors have been shown to occur with changes in process variables such as thickness of the patterned layer and sublayers and changes in the indices of refraction of the materials. All of these variations result in changes in the optical phase difference which occurs on reflection at the line edge and, therefore, results in changes in the structure of the optical image. Although an accurate coherent optical edge detection method has been developed, it requires accurate knowledge of the phase difference which is not always possible in CD measurements. This paper proposes a new dual-threshold method for edge detection and focusing, based on image theory, which can be adapted to most optical microscope-based measurement systems. It does not require knowledge of the phase difference at the line edge. The accuracy of this criterion is compared to two more widely used criteria, (1) the minimum and (2) 50% threshold, and it is concluded that, when the phase difference is unknown and varies with normal processing, the new dual-threshold method is the superior method.

A monochromatic, waveguide model is presented which can predict the optical microscope images of thick-layer objects including multilayer structures with sloping, curved, and undercut edges, granular structures such as polysilicon, and asymmetric objects. The model is used to illustrate the effects of line structure on the optical image. Qualitative agreement with experimentally obtained optical image profiles is demonstrated. Application of the model to study the effects of variations in layer thickness and edge geometry on linewidth measurements made at different stages of manufacturing an MOS device is discussed.

A laser based linewidth measurement system has been developed for measurement of critical dimensions on semiconductor wafers. The system meets both the accuracy and throughput requirements for micron and submicron semiconductor production requirements.

Demands for higher resolution and larger field size are pushing the limits of optical design and manufacturing technology. A simple method of characterizing lens aberrations is proposed in this paper. The effects of astigmatism, field curvature and other aberrations are analyzed. A method of estimating the variation in focal depth for a given resist process and device technology is also proposed. The combined results can be used to determine the minimum feature size and maximum field size to be used with adequate linewidth control. In addition, the lens characterization method can be used for detecting problems with the lens quality and image orthogonality.

The Positive Resist Optical Lithography model (PROLITH) is introduced. The model predicts resist profiles for submicron projection, proximity and contact printing. Included are models for optical projection systems, the standing wave effect for projection printing, a diffraction model for contact printing, a kinetic model for the exposure reaction, and a kinetic model of the development process. Also included in PROLITH are the effects of prebake conditions and polychromatic exposure.

We have evaluated both present and future G-- and I--line lenses by considering their process sensitivity in 0.9-micron lithography. This was accomplished using a modified and automated version of the U. C. Berkeley simulation program SAMPLE. Our results indicate that resist development latitudes are much more sensitive indicators of lens performance than sidewall angle and thickness contrast. When properly defined, they are also relatively insensitive to standing wave effects. Using a single layer of the AZ1350J 6-line resist for all our comparisons, we find the 0.38 numerical aperture 6-line and the 0.42 I-line lenses give superior latitudes. The 0.28 6-line lense provides inadequate development latitudes for reliable submicron patterning. Details regarding opening latitude, base latitude, top latitude, metrology ratio, sidewall angle, thickness contrast, and exposure dose in response to variations in numerical aperture, wavelength, base nodality, and top nodality are presented.

Optical projection microlithography for fabricating fine line and optical data storage technique ave extended into submicron region. But many new technical problems will be appeared in transferring such exacting structure by optical method. One of the important subjects is the investigation of the light energy distribution on focal plane. this paper described the experimental results using an automatic focusing system which was produced by Tsinghua Univ. within +0.1u accurasy. Those have been colipared with the calculating results of perfect lens and residual aberration system,and the relation of them was also discussed.

A new system for detecting contaminants on patterned Si wafers is described. The system employs a linearly polarized He-Ne laser light to scan obliquely across the wafer surfaces, and optical detection technique to distinguish the contaminants from the patterns by measuring the polarization of the light reflected from both the contaminants and the patterns.

A new aluminosilicate glass (AL) has been developed, which possesses excellent melt-ability and low thermal expansivity. The light transmittance and the thermal expansion coefficient of AL well met the requirements for low thermal expansion photomask substrates (LTE) designated by SEMI. The other major properties of AL for photomask use such as chemical durability, mouse-nips, scratch resistance, and electrical resistivity were mainly studied in comparison with a conventional LTE and a synthetic quartz glass (QZ). Both of the weight losses of AL due to acid and alkaline solution attacks were approximately a half those of the conventional LTE, indicating that pattern damages produced during mask cleaning processes with such chemicals are improved. It was comfirmed that mousenips resulted from a reaction between chromium film and sodium ions in AL were not produced under a heat-treatment condition of 200°C, 60 min. From a comparison of the deformation/fracture parameters between AL and QZ, the mechanical damages for AL were shown to be considerably slighter than QZ in spite of the higher microhardness of QZ. The lower electrical resistivity of AL revealed less dust contamination potential than QZ. The defect level and the improved properties of AL were discussed in terms of optical microlithography and photomask application.

A light intensity monitor has been developed for automated intensity mapping of the stepper exposure field at the wafer plane. The hardware consists of a microprocessor controlled light monitor which continuously collects and analyzes data obtained from multiple exposures in the illumination field.