Semiconductor lithography manufacturing presents a major challenge for the application of classical Statistical Process Control (SPC) methodologies due to the complex nature of this process. For example, difficulties can occur due to inadequate data sampling, nonnormal error distributions, equipment or process instability and nonstationary random errors. Incorrect use of classical SPC techniques can result in the incorrect interpretation of process stability which can have a drastic impact on productivity. Photolithography provides additional SPC challenges due to the inherent multivariable nature of the output variables that are being controlled. This paper examines appropriate SPC and monitoring techniques for stepper control of overlay performance using in-process measurement and analysis equipment to address these issues. The average run length of three charting techniques is compared to quantify the ability of each technique to detect process mean shifts. Shewart, Exponentially Weighted Moving-Average (EWMA) and Cumulative-Sum (CUSUM) charts are analyzed for a baseline process and mean shifts of 0.42, 0.85 and 1.25 standard deviations. These results illustrate the superior performance of a CUSUM chart over Shewart and EWMA charts. In addition, the Shewart chart with Western Electric rules produced false mean shift alarms for the baseline case. The EWMA is also observed to be sensitive to the selection of weighting factors. The effectiveness of plotting individual wafers is compared with plotting lot means. The plotting of individual wafers outperforms lot means in the determination of baseline shifts because of the larger population size of the individual charts.

The phase shift focus monitor is a relatively new technique which can be used to determine the optimal imaging plane for a photo-stepper. This determination ofbest focus is accomplished through phase shifting technology developed and patented by IBM. Adapting this concept to an overlay measurement target has advantages of simultaneous x and y measurements, high precision, and automated data collection. Printed on the reticle are phase shifting structures using the box-in-box pattern. These phase shifting structures produce lateral shifting ofimages at the wafer plane when they are imaged out offocus. This is a direct result from utilizing non- 1 80 degree phase shifting material. When the patterns are imaged in focus, no lateral shifting, or overlay error is exhibited by the box-in-box patterns. It has been found that the overlay error of the printed structures increases with the magnitude of defocus. If overlay data is collected from several different levels of defocus, a direct linear function can be found to relate best focus to the overlay error of the printed box-in-box patterns. Additionally, it can be shown that this relation is dependent upon the NA (numerical aperture) and PC (partial coherence) of the imaging system. This technique can be used as a very powerful analysis tool for characterizing photo-steppers, as it provides for the collection ofvast amounts ofdata about a stepper's condition rather quickly. It is laid out in an arrayed pattern with 121 modules in an 1 1 x 11 matrix. The complete matrix is contained in a 22 x 22 mm field (at wafer level on a 5x reduction stepper).Each module contains the phase-shifted box-in box patterns. Several potential applications have been suggested already. One wafer can be used to collect data on current best focus and field tilts in a few minutes. More static characteristics specific to the lens can be determined as well, i.e. astigmatism and field curvature. All of this information can be collected by measuring the overlay error from several of the 121 module positions in the field. This paper will discuss techniques to use the focus monitor to characterize steppers equipped with variable NA and PC. In addition, it can be shown that the monitor can provide quick and accurate measurements to verify and track the stepper's own internal diagnostics for statistical process control. Keywords: phase shift, focus, photostepper, reticle, numerical aperture, coherence, overlay

This paper describes a technique for quickly achieving photolithography and etch critical dimension (CD) targets using measurement data from an automated CD SEM and a statistical analysis package. The experimental dataset is created from the measured CD response as a function of process input parameters that have been varied in a controlled fashion across a wafer or on multiple wafers. The resulting model is displayed as an array of "prediction profiles" that allow interactive variation of the model components to simulate the response of CD to input changes. The outputs of the analysis are used to determine optimal processing conditions and to provide an estimate of process latitude. Keywords: Automated CD SEM Metrology, Lithography Process Characterization

Iii previous work by Sturtevant et al [1994] it was shown how the use of scatterometry data obtained in real time during the post-expose bake (PEB) step of the photolithography process can improve critical dimension (CD) control. However, it is not obvious that the ad hoc methodology utilized during these experiments conducted at IBM, is the best way to use this valuable information. This paper focuses on the development of more rigorous methods that can incorporate scatterometry data in on-line prediction of the bake time necessary to achieve the target CD. First, a brief description of the system and required tools is given. Then, a model that can relate the measured intensity and the time the wafer has been baked (referred to hereafter as the bake time) is introduced. Then, two theoretical methods that can utilize this model to calculate the bake end point are described. Finally, these two methods are applied to the analysis ofadditional data collected at IBM and operation oftbe complete end point control system is emulated to the extent that is possible. Keywords: end-point detection, process control, deep-UV lithography, scatterometry, post-exposure bake process

Early technology learning for O.25m lithography should confirm predictions of tools and processes needed for printhg these small image sizes. Improvements are felt to be needed in reticle critical dimension (CD) control, stepper overlay, lens design, and resists for imaging O.25.im. Etch technology is also tested for substrate as well as organic ARLs at these line sizes. This paper will present results from efforts to produce O.25jm poly, O.301um contacts/vias, and O.3OMm metal images on a device of realistic topography applications. Poly patterning has been improved by using off axis illumination but the process window is still small. There is a trade off between focus latitude and isolated to dense bias when using annular illumination. While organic ARLs have improved imaging CD control, there is a corresponding loss in CD control during ARL etch. Effects of ARL thickness variation over topography will also cause CD variations when etching to clear the thicker areas. Poly etch itself also contributes to the CD uniformity variations. At the contact masking level the wafer has a fairly planar oxide surface but even with a chemical mechanical polish the oxide thickness will vary due to the underlying topography. Contact patterning studies will include sensitivities of contact imaged CDs to variations in film stack thicknesses and reflectivities. Process windows will be shown for several film stacks with and without ARC for different resist thicknesses. The relationship between oxide thickness and contact CDs will be shown for contacts at O.3Om. For metal etch, results will be shown for msitu and dedicated ARL etch tools using various over-etch conditions. The relationship between the resist thickness required for the current ARL and metal etch selectivity and litho patterning capabilities will be discussed. Results will be shown for metal patterns post imaging, ARL etch, and metal etch at O.301um. Keywords: annular illumination, etch bias, process window, deep ultraviolet, process bias

Over the past ten years lpw kV Electron Microscopy has been the technique of choice for inprocess, critical layer metrology, for leading-edge design-rule devices. However, conventional low kY Secondary Electron microscopy is reaching its limits in its ability to measure near and sub-half micron features at all levels due to charging issues and interpretation of resist profile. A re-evaluation of the strategy for determining CD measurement site becomes increasingly important as site to site differences are more significant at these smaller dimensions. Otherwise an apparently well controlled process measured in a typical site (e.g. array of a memoiy cell) could be failing due to shorts in critical sites. These critical sites tend to challenge the limits of conventional SEM metrology more. A new generation of SEMs offering a variety of techniques to overcome these limitations has recently arrived on the market, these improve visibility and reduce the effects of charging, allowing a more accurate and representative control ofa lithographic process to be made.

The design of the world's first production worthy broadband ultra-violet and visible small spot spectroscopic ellipsometer is described. The instrument, called the Prometrix® UV125OSE, was developed by the Prometrix division of Tencor Instruments in cooperation with SOPRA S.A., a pioneer in the field of spectroscopic ellipsometry. It has the ability to measure both the thickness and refractive index of different layers on a wide variety of materials in multiple layer film stacks. In this paper the optical system will be reviewed and spot size data presented. We will further discuss some of the design considerations such as the angle of incidence and allowed spread of the collection beam. Data characterizing the precision and stability of the instrument is presented for a variety of films including SiO on silicon, and Si3N4 on silicon, and a multiple layer stack of SiO Ipoly-SiISiOon silicon. Keywords: spectroscopic ellipsometry, ultra-violet, thin films, measurements, index of refraction, silicon dioxide, silicon nitride, polysilicon

A new technique that furnishes simultaneous and unambiguous determination of thickness and n(A) and k(A) spectra as well as energy bandgap, Eg , of a wide variety of semiconductor and dielectric thin films deposited on either opaque or transparent substrates will be described. The n(A) and k(A) spectra are determined over a wide range of wavelengths from deep-ultraviolet to near-infrared (190 nm ) 900 nm). Films as thin as 20 A can be measured. The technique is based on reflection/absorption spectroscopy and optical dispersion equations derived by A.R. Forouhi and I. Bloomer. Applications include, but are not limited to: a-C:H on magnetic disks and on A1TiC (with or without a thin SiNX adhesion layer), SiN:H and SiO:H deposited on Si, a-Si:H deposited on oxidized Si, ITO deposited on glass or quartz, SiO on glass, and TiO on quartz. Furthermore, the technique can be used to characterize the damaged silicon surface layer produced by plasma cleaning of bare crystalline silicon wafers.

Titanium nitride (TiN) is increasingly used in multilevel metallization processing for a variety of applications: as an antireflective coating, an aluminum diffusion barrier layer and a tungsten interconnect and plug adhesion layer. Typical process control techniques are optical measurement of reflectivity of film thickness, or four-point probe measurements of sheet resistance monitor wafers. Reflectivity measurements of titanium nitride antireflective coatings at the stepper exposure wavelength are straightforward, but direct optical thickness measurements of TiN layers have been difficult to make. Spectroreflectometers rely on accurate values of refractive index dispersion (the change in RI as a function of wavelength) to calculate thickness. Single wavelength ellipsometers do not collect sufficient data to calculate thickness as well as the real and imaginary portions of the refractive index. Spectroscopic ellipsometry offers the ability to determine thickness and refractive index dispersion. This paper investigates possible correlation between the refractive index of TiN and its composition. Refractive index and thickness of TiN are determined by spectroscopic ellipsometry, and composition is determined by Rutherford backscatter spectrometry and Auger electron spectroscopy. If a correlation can be identified, the possibility exists for nondestructive process monitoring of TiN composition. A standard coherent sputtered TiN and two different CVD TiN processes will be examined.

The progress of integration in semiconductor (LSI) is remarkable. We expect that the mass-production of 256M DRAM shall start in the next few years. The design rules for this level of integration will require even more accurately in a system to measure a reticle pattern's position. To meet the measurement requirements for future IC designs, Nikon Corporation has developed to XY-5i. During development, we introduced some new technologies to improve the systems' measurement performance. In this paper, we will explain the new flexure compensation concept and the reduction of measurement error caused by temperature drift. These are the key technologies used to attain the improved performance. This performance will be demonstrated with actual measurement data from the XY-5i.

A method for effectively utilizing a Scanning Electron Microscope (SEM) for defect review and identification of unpatterned silicon particle wafers, following inspection on a laser-scanning defect inspection tool, is presented. The method involves prepatterning of bare silicon wafers at the extreme edges, typically less than 20 mm from the edge of a 200 mm wafer, using standard (I-Line) photolithographic processing. The registration marks created in this process are used for stage correlation between the SEM and laser-scanning wafer inspection tools. Use of these marks is demonstrated to result in a 50% improvement in particle location accuracy (mean + 2 sigma) over previously reported results. Further optimizations, including modeling and removal of systematic error sources through data transformations, demonstrate the additional improvements in particle location accuracy that are possible.

This paper describes a broad range of design issues that influence the performance of optical equipment for in-line inspection of random (logic) and repetitive (memory) patterns. In particular, we describe the angular distribution of signals from defects on a patterned wafer illuminated by a focused optical beam. We analyze the configuration of both illumination and collection optics to maximize the signal to background ratio for the detection of submicron defects on pattern. In addition, we analyze the distribution of the scattered light as a function of pattern periodicity and orientation with respect to the illuminating beam. The advantages of polarization selection and spatial filtering techniques are explored to enhance the detection sensitivity on repetitive and random pattern wafers. From these results we have developed a new patterned wafer inspection system that offers increased sensitivity and improved defect capture.

An inspection system has been developed that uses high-speed electron- beam imaging combined with digital image processing to automatically locate defects on semiconductor wafers. The system inspects wafers using a low beam energy of 0.8 KeV and a scan speed that is as much as 1000 times faster than traditional SEMs. This paper describes the system and presents characterization testing results of the system's ability to find defects on advanced integrated circuits. Three unique applications for the new automated scanning electron microscope inspection system (SEMSpec) have been found. First, the high resolution of the electron imaging microscope allows the tool to find much smaller defects than optical inspection systems can find. The SEMSpec system found defects smaller than 0.1 micrometers in size, even on densely packed, sub-quarter micron, high aspect ratio, multilayer geometries. Second, wafers that are late in the fabrication process are much easier to inspect at high sensitivity by SEMSpec than by optical means. This is because SEMSpec images are much cleaner than optical images when viewing grainy poly, metal, or other interconnect layers. Third, the SEMSpec system can sometimes find electrical problems by observing charge-induced voltage contrast differences. Voltage contrast occurs when some of the IC features are electrically defective and they charge up differently under the electron beam than they normally would. The SEMSpec system can then identify them as defective even though they might otherwise appear to be perfectly formed.

A set of test masks on which programmed defects are placed has been standardized by SEMI (SEMI Standards P23-93). The masks have been designed to be used for benchmarking the sensitivity of defect inspection systems for reticles and masks. The standard test vehicles consist of two types of masks which have quasi device background patterns (contact holes or through-holes and wiring patterns) with a wide variety of programmed defects. The geometry of the background patterns on the masks have been designed for 0.5 micrometers or less when printed on a wafer with a 5x reduction stepper. Under the coordination of SEMI Japan, the test masks have experimentally been fabricated by a couple of Japanese mask shops, and the masks fabricated have been evaluated focusing on defect sizing accuracy. The mask which has holes as background pattern has been used for defect printability study using a 5x reduction i-line stepper. As a result, it has been confirmed that the test mask is a useful vehicle for the study. The standard masks, therefore, can also be used to make a standard defect spec on which mask suppliers and users agree.

Lithography is the critical process for developing smaller and more dimensionally accurate semiconductor structures more reliably. Many random defects are caused by contamination particles during the lithographic process. Several models have predicted yield loss associated with defect density and particle size. Although some have taken into account the defect growth mechanisms, a more accurate simulator is needed to provide the transformation from the contamination to the defect level. In this paper, the contamination to defect transformation is investigated using the 2D lithographic simulator, METROPOLE. A study is presented that characterizes the effects of particle size, location and composition on the formation of lithographic defects for a 2D metal line structure. Particle characterization and parameterization is carried out for a number of common conditions including different light polarizations, defocus, and nonplanarities. The simulations are checked against defect data from AMD Sunnyvale.

As device geometries shrink to 0.25 micron and smaller, all facets of integrated circuit (IC) processing are being challenged. With device sizes shrinking, so too shrinks the size of a defect that can cause chip failure, and hence yield loss. Contamination free manufacturing practices are becoming critical for successful device fabrication. To accomplish this, elimination of defect sources has a high priority. A defect can be a particle, microcontamination, pattern anomaly, crystalline defect such as a stacking fault, and so on. Defects have become a main source of yield loss to the semiconductor industry. This comes at a time when 90% yield values on mature product cannot increase at the rate that has occurred in the past. The industry is now faced with finding methods of incremental yield increase, in-line, on production wafers. Automatic Defect Classification (ADC) is an important part of SEMATECH's strategy to meet these industry needs.

Rim and attenuated phase shift masks can extend the range of i-line optical stepper lithography to 0.4 micrometers with enhanced depth of field. The repair of defects through subtractive and additive methods using focused ion beams is critical to the future success of the technology. Printability tests and simulation studies demonstrate the hierarchy of defects in which both rPSM and aPSM types are less tolerant of absorbing/attenuating defects, but more tolerant of clear defects than the corresponding conventional BIM. Repair trials reveal the existence of a new type of post repair defect - the phase trench - but also clearly demonstrate the feasibility of repair using FIB methods.

Every artifact measurement standard has some uncertainty associated with its calibration, and the NIST Photomask Linewidth Standards are no exception. The methods of estimating measurement uncertainty, however, and the interpretation of its meaning, have been subjects of discussion for many years. The International organization for Standardization (ISO) has recently published a Guide to the Expression of Uncertainty in Measurement, which outlines an operational procedure for estimating and combining the effects of the various contributing factors. A new NIST policy on measurement uncertainty has led to a change in the method of determining the uncertainty of NIST Photomask Linewidth Standards to comply with ISO recommendations. Consequently, their calibration uncertainty has changed. The new method of evaluating systematic components leads to increased systematic uncertainty, and the new method of combining uncertainty components leads to reduced calibration uncertainty. This paper describes the old and new methods and compares their results.

The 'edge detection' is a known problem in optical linewidth metrology. New illumination capabilities like Real Time Broad Band Confocal Scanning, the combination of conventional transmitted illumination with real time confocal scanning, or the use of UV transmitted illumination, together with the use of finite transmissive/reflective masks require a reevaluation of the traditional edge detection approach. Edge response function and Edge response width were reviewed for confocal scanning illumination and for typical finite transmissive (binary) masks. The proposed edge detection strategy deals with edge position shift with detection threshold in different illumination conditions and the relationship between the detection threshold and 'best' focal plane selected for measurement. 'Edge roughness', another known problem was reviewed from different prospectives and measurements of the edge roughness were performed at various edge detection thresholds. Mapping of mask feature 'aerial intensity' was used to view the edge roughness, edge slope and to locate edge position. Aerial intensity simulations with SOLID-C simulation tool were in good agreement with the experimental data. The edge analysis strategy helped to obtain consistent 3S repeatability of less than 8 nm and good accuracy on various masks substrates, like resist develop-inspect, half tone attenuated and chromeless.

A new approach in design and metrology of the monolithic delay line ICs with current-induced adjustment of the electrical parameters is proposed. The circuitry of the delay lines is based on using of two separate delay circuits for delay of the leading and trailing edges. In order to ensure the accuracy of the delay time the phenomenon of current-induced resistance change in polysilicon resistors is used. The measurement and adjustment of the dynamic parameters of the ICs are made on the silicon wafer directly after fabrication before packaging. The program algorithm and specially software and hardware method of the delay time measurement are presented. The main electrical parameters and fabrication costs of the designed monolithic delay line ICs are compared with the known monolithic delay line series.

The relative misalignment of features produced by different mask levels (i.e., overlay error) is projected to become an increasingly important problem to the semiconductor industry as the size of critical features continues to decrease. In response to this need, NIST has initiated a two-step program to improve the measurement of overlay error, and to develop standards in support of overlay metrology. The approach of the first step is to develop a 'toolkit' of standards for characterizing the tool-induced shifts (TIS) in the optical tools presently used for the measurement of overlay error, with the aim of producing and maintaining overlay tools with negligible TIS. The second step, which necessarily follows the successful completion of the first step, is aimed at developing artifact standards with negligible wafer-induced shifts for the calibration of the TIS-free overlay tools.

Modified illumination schemes for wafer steppers, such as oblique and low partial coherence illumination, can improve the imaging performance for some types of microlithography patterns. Implementation of these methods by simple insertion of apertures into a stepper illuminator, without reoptimization of the illuminator, can lead to deterioration in illumination power and uniformity. The different apertures may also affect lens aberrations, lens distortion and illumination telecentricity and require adjustments for the effective control of dose, focus and reduction ratio. This study has examined the set-up of a modified illumination system, the effects of four different illuminator apertures on several lens parameters and the consequent effect on overlay between layers printed with different apertures. It is found that the lens performance is either not significantly changed or can be compensated adequately in the case of off-axis illumination apertures but deteriorates when using low partial coherence illumination. However, preliminary overlay tests between fields printed using different apertures show acceptable performance in all cases.

In this paper, we address the problem of ensuring accurate overlay in lithography of patterned substrates with low-order distortions. This problem becomes more important in situations where warpage and thermally induced distortion of high-density quasi-periodic patterns in large flat panel displays (FPDs), or high-density memory cells demand appropriate compensation schemes. A recently developed signal processing technique is used to estimate and compensate for substrate distortion. While avoiding time-consuming (e.g. site-by-site) alignment procedures, the proposed technique permits the use of time-efficient alignment procedures (e.g. global alignment) without compromising overlay accuracy. The proposed technique first estimates the existing low-order distortion of the patterned substrate. After estimating the distortion, global alignment schemes can be used to align the image of next layer to the previous one in a few check points. In the final step, the estimated distortion function can be used for compensating the local misalignments.

Comparative analysis of the total overlay accuracy on aluminum (Al) and chemical vapor deposition tungsten (CVD-W) was carried out as a function of the available alignment techniques over the various process conditions. In the case of Al films, they were prepared at the various sputtering temperatures, thicknesses, and step heights. The total accuracy was also examined associated with standard and off-axis illumination (OAl). The effect of the sputtering temperature of single and dual Al deposition has been especially investigated in a view of surface roughness, which significantly affects total overlay accuracy. It was observed that CVD-W film can be an alternative of Al film. As the Al film deposition temperature decreases, it was found out that random requisition error with a large distribution was improved by implementing an off-axis alignment (OAA) technique with a visual broadband light source. Meanwhile, Al thickness was not so much critical to overlay accuracy comparing with Al sputtering temperature. And also scaling error determined by asymmetric alignment mark shape was minimized according to the optimization of exposure data file. Minimum detectable step height of alignment mark was down to 300 angstrom using a sensitive alignment techniques over the Al film.

Chemical-Mechanical Polishing processes are helping chip makers solve problems with multilevel metals, smaller device features and narrower stepper lens depth of focus, but are creating new challenges for other segments of the semiconductor industry. In particular, planarized metrology targets can cause overlay measurements to be inaccurate as well as imprecise. Coherence Probe Microscopy (CPM), when applied to overlay metrology, provides a solution for registration measurements on chemical-mechanical polished wafers. This paper describes the use of CPM measurements technology to characterize chemical-mechanical polishing wafer processes. The presented results show a comparison of misregistration measurements from planarized wafers before and after etch and discusses the impact on registration by chemical-mechanical polishing.

Since the most advanced circuits on the market are drawn using 0.35 micron design rules the Scanning Electron Microscopes (SEM) is currently being used for the microelectronics field. SEM take the place of optical microscopes both for observation and for Critical Dimension Measurements (CD). The images, or signals, given by SEM are not so reliable with time as those given by optical means, and frequent settings of SEM are necessary to avoid bad image quality. A semistandardized procedure recommended by SEM manufacturers is recommended to measure the resolution: A carbon sample is covered by gold particles or by Latex Spheres to give the maximum signal contrast. The transition quality between gold and carbon evaluated by measuring the gap between two isolated particles is called the resolution. The previous method cannot give a good result in terms of both precision and repeatability. In order to overcome the drawbacks, the operator's eye is replaced by image densitometry analysis and the measurement accuracy is improved by averaging the signal: The image Fourier transform provides two dimensional spatial frequency information. The results are robust and objective and lead to a quantitative resolution and astigmatism measurement. This method is fast (30 seconds), non operator dependent, automatic and the SEM images can be monitored in line.

The Hitachi Model S-6280 high resolution field emission Scanning Electron Microscope (SEM) is used at Motorola's MOS-12 facility for Critical Dimension (CD) measurements of less than 1 micrometer in size. This tool is used to measure 200-mm diameter wafers with a low accelerating voltage in the range 0.7-1.0 kV. During new process development the tool is run in both manual mode and in a fully automated mode utilizing patten recognition for global alignment as well as feature CD measurement. Results on the reproducibility of fully- automated measurements show that this technique provides a significantly improved measurement capability compared to the standard manual technique.

In this work, we address the 0.25 micrometers yield management applications of multimode electron beam imaging using both backscattered and secondary electrons. The prospects for achieving aggressive performance specifications for quarter-micrometer process control in imaging, metrology and productivity are analyzed, and supporting examples are provided.

For effective critical dimension metrology at geometries of 0.5 micrometers and smaller, it is essential to understand the true edge location, as represented in the SEM image profile of the feature. Total uncertainty associated with the linewidth measurement is often greater than the entire metrology error budget allocation. Capability to accurately predict (or model) the observed signal output for any type of feature is required to reduce measurement uncertainties to acceptable levels. The model, once validated, can then be used as the basis for measurement algorithms, error analysis, system calibration. In a manner analogous to the development of the optical photomask standards we investigate the capability to model, using a Monte Carlo based technique which incorporates measurement system geometries and fields, the output of an in-line, backscatter detection based, CD SEM. A comparison of line waveform profile results obtained from measurements taken of several simple resist-on-silicon test structures, with varying sidewall profiles, against simulations of the same structures is made. The line geometry for simulation input is derived using extensive analysis of in- line SEM and laboratory SEM systems for top down and crossectional measurement and an Atomic Force Microscope (AFM) for layer thickness and sidewall angle measurements. The results of these measurements is a trapezoidal line geometry for simulation input. Measurement profiles from simulation and experiment are compared and highlight the importance of; SEM operating condition (resolution/spot size) and linewidth subtleties (edge rounding at the top of the line, presence of a 'foot' or antifoot at the bottom of the line). Capability to use simulations to determine edge positions and the future work required to accomplish this are also discussed.

This paper deals with a technique recently developed at France Telecom/CNET for increasing the electrical conductance of novolak and polyvinylphenol based photoresist patterns and consequently for avoiding any charge effect occurring during SEM inspection. The process consists in subjecting the photoresist patterns to an argon plasma, which, under specific conditions, creates the graphitization of the surface of the photoresist, leading to a significant decrease in electrical resistivity and total suppression of the charge effect. When the size of the hole becomes sufficiently small, and despite the high resolution capability of the equipment, only the top of the hole is well defined. Due to charge effects, the image at the bottom of the hole is badly resolved and therefore the dimensions cannot be registered. However if the resist layer is previously exposed to an argon plasma, the charge effects totally disappear and the contour of the image can then be very clearly defined, by the sharp transitions of the intensity profile. The process is cost effective, principally nondestructive and has been demonstrated to be a real industrial method. When wafers are treated using the plasma process described above, an automatic, highly reliable, nondestructive inspection of a 7 X 7 matrix can be performed in less than 45 min. The results include top and bottom measurement using standard low energy SEM CD equipment.

To meet the requirements of 0.25 micrometers integrated circuit CD-Metrology, new detection and SEM optics have been developed which provide: (a) 5 nm resolution at 0.6 keV and (b) Spatial metrology using a split in-lens secondary electron detector. The superb resolution at low beam energies is achieved using a ChromaCorrected objective lens having overlapping magnetic and electrostatic fields. The lens was numerically optimized to achieve exceedingly small chromatic aberration coefficients at the low beam energies. The detector is inside the electron column and above the objective lens. This permits the use of a very short working distance, which enhances both resolution and detection efficiency. The concentric design permits collection of the secondary electrons with zero aberration of the primary beam. The detector plane is split into four quadrants, so as to enable the use of spatial metrology algorithms for topographic measurement.

This paper presents the methodology used to perform an evaluation of automated CD metrology SEMs for sub-half micron process control. The paper describes the evaluation strategy, the procedure used to collect and analyze the evaluation data and concludes with recommendations on how the procedure can be improved. The evaluation process was designed to estimate metrology capability and review specific application requirements envisioned for a leading edge semiconductor development and manufacturing facility at Motorola. The evaluation process consisted of a quantitative evaluation of measurement performance specifically examining the reproducibility, linearity, automation success rate, and throughput of the instrument. In addition, capabilities such as user interface, computer integration, job transportability, and technology roadmap were assessed qualitatively. Although a particular evaluation of automated CD SEMs is considered here, the principles used to develop the evaluation procedure can be applied to metrology tools in general. A discussion of the application and desired functionality of CD metrology instrumentation including performance criteria is presented for completeness.

Linewidth measurement in quarter micron devices involves some difficult problems, such as charging of photoresist films. Sample charging prevents not only accurate linewidth measurement but also inspection of high aspect ratio holes. Then we proposed the back scattered electron assisting LSI inspection (BEASTLI) method. In BEASTLI, a 200 keV scanning electron beam is used without being influenced by the charging of insulator. As BEASTLI is extremely useful for examining the shape and dimensions of devices, time and labor is saved when optimizing process conditions. However, owing to irradiation damage to semiconductor devices, BEASTLI could not be used as an in-line CD measurement system. In order to realize 'damage free inspection', we evaluated irradiation damage, such as V_TH shift, increase of junction leak current, deterioration of oxide reliability, and hot carrier degradation. We evaluated these characteristics using MOS transistors. As a result of the investigation, we could apply the BEASTLI method for in-line CD measurement.

Photoresist linewidths are presently controlled by inspecting wafers in plan view with a 'metrology SEM'. No standard reference materials exist for checking this measurement. We show the feasibility of creating an in-house quasi-standard for this purpose. Using a modern i-line stepper (0.54 N.A.), we produced three specimens of 0.5 micrometers wide photoresist line test patterns which presented both aberrated and good line shapes. We examined the specimens in cross-section using the AFM and in plan view using the SEM. The AFM images had crisp edges defining the line profile. We measured pitch, linewidth, and slope angles from the AFM images and compared these with the pitches and linewidths reported by the SEM. We found differences for features that had curved sidewalls. We discuss the basis for edge contrast in the AFM and propose a method for mathematical analysis. This work shows that the AFM can directly examine resist line profiles and provide images of useful precision, without adding a conductive coating as is commonly done in high resolution SEM. Any added coating should be avoided on a standard, since it would modify both the physical width of the structure and its electron scattering characteristics.

NIST personnel (J.S. and T.M.) visited 23 IC manufacturing companies and equipment suppliers during 1994 to determine semiconductor industry needs for scanned probe metrology. NIST has initiated projects addressing some of the need identified. When complete these projects will enable improved metrology using the scanned probe microscope. Industry needs include pitch, height, angle, and width calibration artifacts, and understanding of the effect of humidity on AFM measurements, and tip metrology techniques. To meet these needs we have designed and built a Calibrated Atomic Force Microscope (C-AFM) with interferometric position control. This AFM is capable of making accurate measurements. We present the operational characteristics of the instrument, accurate X, Y, and Z pitch measurements on a commercially available artifact, measurements on a prototype surface roughness artifact, and a promising technique by which to make accurate linewidth measurements.

As a design rule of LSI device gets smaller, critical dimension (CD) control becomes increasingly difficult. This creates a new technology to CD control accurately. The technology utilizes intensity of light diffracted from latent image consisting of periodic patterns in undeveloped photoresist, and its possibility has been reported for several years. We have developed a new method of the CD control by monitoring profile of the latent image using atomic force microscope (AFM). The reduction of CD variation using this method is achieved by controlling development time from the relationship between the profile of the latent image and CD after development. We have tried to apply this method to a resist process in KrF excimer laser lithography and found usefulness of the method for 0.25 micrometers lithography.

Scatterometry, the characterization of periodic structures via diffracted light analysis, is shown to be a viable and versatile metrology for critical dimensions as small as 0.24 micrometers . Scatterometry is rapid, nondestructive, inexpensive, and potentially useful for on- line control during several microlithographic processing steps. This paper discusses two recent studies in which scatterometry was applied to the measurement of developed photoresist patterns. First, scatterometric measurements of developed resist lines in the 0.38 micrometers to 0.70 micrometers range will be presented. Results from four sample wafers are shown to be consistent with SEM measurements. For one wafer, the average deviation of scatterometry linewidth measurements form top-down SEM measurements, over a broad exposure range, is 14.5 nm. Moreover, our scatterometer is shown to be highly linear with the SEM; linearity coefficients have typically been above 0.99. The goal of our second project has been to determine whether scatterometry measurements are affected by variations in the integrated circuit production process. A set of twenty-five wafers was fabricated with deliberate variations in the exposure dose and the underlying film thicknesses. We are presently investigating the effects of the film thicknesses on the measurements of critical dimensions (CDs) as small as 0.24 micrometers . Preliminary results indicate that CDs and multiple thin films can be simultaneously measured by applying multi-parameter prediction algorithms to the scattered light data. Results from four different prediction algorithms are given. Finally, the repeatability of the scatterometer is shown to be excellent: 0.5 nm for consecutive measurements and 0.8 nm for day-to-day measurements. The results of an extensive repeatability/precision experiment are presented.

A novel design for an ultraviolet critical dimension measurement transmission microscope utilizing the Stewart platform as the rigid main structure has been implemented. This new design shows improved vibration characteristics and is able to accommodate large specimens. We present alignment procedures and their relevance to obtaining accurate linewidth measurements. Initial measurements with the new system comparing visible and ultraviolet wavelength illumination show expected characteristic dependence of the intensity image as a function of wavelength.

The National Institute of Standards and Technology has designed and built two electrically calibrated laser calorimeters as primary standards for absolute energy measurements at the wavelength of 248 nm. Under the sponsorship of SEMATECH, NIST developed the calorimeters to improve measurement of dose energy in excimer laser based microlithography. The calorimeter system can be used to calibrate transfer standards which in turn can be used to calibrate detectors employed for energy measurements of semiconductor wafer exposure. The excimer calorimeter uses a glass filter which functions as a volume absorber that allows collection of nanosecond pulses of laser radiation without suffering damage. The measurement range of the calorimeters is 0.3-25 J, but can be extended to 1 mJ with beamsplitters. Electrical calibration of the calorimeters shows a standard deviation in the calibration factor of less than 0.5% for entire energy range. The total uncertainty of typical power and energy meter calibrations is approximately 2%.

The effect of defocus on the printability of submicron 5X reticle defects at g-line, i-line and duv wavelengths is assessed. Imaging of a specifically designed test reticle incorporating defects whose size, and proximity to adjacent features, varies within submicron line/space arrays was studied using the simulation package SOLID. The results were assessed and initially presented by plotting minimum printed defect (MPD) vs. array linewidth (L/W) to compare with a previously reported practical study. Following this the results were replotted as curves of MPD vs. defocus at differing linewidth values. Finally, the resist modelling program has been used to create 3D images of the printed defects and thus demonstrate the effect caused by defocus on the resist profiles. Results on defect printability enable future reticle procurement specifications to be established.

'Swing curves' derived from the modelling program SOLID are presented for a typical conventional i-line resist. At a constant sub-half-micron feature size, results illustrate the effect of varying stepper lens conditions on 'swing curve' amplitude and indicate that, whilst initially defined by reflectivities, resist thickness and resist absorption coefficient, it increases as aerial image peak intensity decreases. 'Swing curves' are drawn in terms of exposure-to-size vs resist thickness which has the advantage of always producing complete plots.

One of the main problem for 0.25 micrometers features will be the depth of focus; Deep UV source will be necessary, but anyway DOF will be in the one micron range. So will it be possible to have CD in specifications all over the field? Many parameters have to be checked and monitored but the photoresist thickness on wafer is an important part due to the interferences franges created inside the photoresist during exposure. In order to reduce this phenomenon, we are used to choose the thickness on the flat part of the swing curve, for maximum or minimum coupling. As the wavelength is being lower, from 365 nm on i-line steppers to 248 nm on Deep UV steppers, this flat part length is reduced, so to obtain the same CD's repeatability the photoresist thickness uniformity must be improved in Deep-UV. This paper presents an effective method to obtain a Deep UV photoresist uniformity allowing 0.25 micrometers CD's in specification on the whole 8' wafer. A great number of process parameters may affect the resist coating uniformity, so we have decided to use the Designs Of Experiments to solve such a complicated problem with a minimum number of experiments. Due to the centrifuging process used to coat the resist on the wafer, the main uniformity default that can appear is a radial default. So we have chosen a combination of Taguchi and Response Surface Methods, in order to minimize the thickness variations versus the radius. First of all, to evaluate the effects of the 8 process parameters on this response, we have performed a screening design of 16 experiments. A statistical analysis has shown that the thickness variations depend on only 5 parameters. Then a Response Surface design of experiments has been achieved, with 27 experiments. A very accurate modelization of the uniformity default has been carried out from this design, and the optimization has led to a thickness standard deviation of only 3.3 angstrom on the whole wafer. Moreover the quality of the modelization, predictive in the complete domain, will allow to work out further problems.

A simple null modulation-polarization method of measuring optical constants of metals has been adapted for operation with a KrF 248 nm excimer laser. The approach requires only 3 optical components to extract the real and imaginary parts of the index of refraction (n,k). Experimental results will show good agreement to reference values for several metals (Cr, Au, Al) and Si.

Process variability in a photolithographic process can arise from a number of sources including photoresist developer assay variation. A change in activity of only +/- 0.001 eq/l in a developer solution with a normality of 0.2624 eq/l can consume or exceed the allowable +/- 5.0% CD specification set by most fabs. This change, which represents a relative error (error/setpoint) at 3(sigma) or only +/- 0.4%, is typical of the allowable variability from developer suppliers. Developer production systems with tighter assay tolerances should reduce process variability and increase process robustness. This study compared the effectiveness of several control techniques for blending tetramethyl ammonium hydroxide (TMAH) developer. Techniques with feedforward and feedback control were investigated. An analysis of sources of error in developer blending systems indicated that feedforward techniques could not achieve the desired relative error of +/- 0.4%. An experimental study was undertaken to determine the relative error of three feedback control methods. The feedback techniques yielded considerably tighter assay control than that expected from the feedforward techniques. An analog setpoint control algorithm used with conductivity measurements provided more precise control than a discrete setpoint control algorithm. However, only feedback control with titration met the goal, achieving a relative error of only +/- 0.13% at 3(sigma) (+/- 0.00034 eq/l).

A new NIST SEM magnification calibration standard has been fabricated and characterized in production prototype form. The SRM 2090A samples contain structures ranging in pitch from 3000 micrometers to 0.2 micrometers and are useful at both high and low accelerating voltages. The design and fabrication of the samples has incorporated many of the improvements suggested through two previous prototype series. Since the low- accelerating voltage, critical-dimension measurement scanning electron microscope has assumed an important role in modern semiconductor process control. We investigate the use and performance of the samples in a representative instrument. Scan area contamination from a prolonged high magnification scan is shown to be a significant issue. A cleaning procedure developed for a previous prototype series has been modified to obtain acceptable results on the present samples.

The definition of"Moore's Law" has come to refer to almost anything related to the semiconductor industry that when plotted on semi-log paper approximates a straight line. I hesitate to review it's origins and by doing so restrict it's definition. However, today I will review the history and past performance relative to predictions and show where the advances have come from. I will leave the future performance up to you. Certainly continuing on the same slope doesn't get any easier. It presents a difficult challenge to the industry. The original paper that postulated the first version ofthe "law" was an article I wrote for the 3 5th anniversary issue ofElectronics Magazine in 1 965 . My assignment was to predict what was going to happen in the semiconductor components industry over the next ten years --to 1975 . In 1965 the integrated circuit was only a few years old and in many cases was not very well accepted. There was still a large contingent in the user community who wanted to design their own circuits and who considered the job of the semiconductor industry to be to supply them with transistors and diodes so they could get on with their jobs. I was trying to emphasize the fact that integrated circuits really did have an important role to play.

ULSI has heavily depended on develops in optical lithography. However, optical lithography is now facing a major obstacle due to exposure wavelength limitations. To overcome this obstacle, not only the use of shorter wavelengths, but also such new technologies as super-resolution techniques, electron beams, and X-ray lithography are being intensively investigated. This paper reviews recent developments in these technologies and discusses the major issues. The difference in lithographic activities between Japan and the U.S. is also be discussed. Finally, recent developments in lithography for experimental 1 -Gb DRAMs are presented.

In 1994, The National Technology Roadmap for Semiconductors was revised, updated to reflect the latest information, and published in December 1994. Lithography continues to be a cornerstone for semiconductor process and productivity. The roadmap activity is sponsored by the SIA. The Technical Working Group (TWG), which does the actual work of updating the roadmap, is comprised of members from SEMATECH, SEMATECH FTABs, SRC, industry, universities and government. One of SEMATECH's key goals is to insure that the activities we pursue are well-aligned to the roadmap. The roadmap is focused on requirements and needs of the industry. SEMATECH is focused on solutions. The Lithography roadmap makes clear the desire of semiconductor manufacturers to stay with optics as long as possible since shifts to non-optical technology are viewed as much higher investment and risk. It is perceived that optics can be a viable manufacturing solution through at least the 0. 1 8uni generation. The TWG and SEMATECH FTAB meetings highlighted the need for greatly increased emphasis and resources on 193nm technology to bring it to pilot production in 1998 for the 0.1 8um generation. SEMATECH has since worked with the industry to define more detailed needs and requirements for a successful 193 program and is working with industry and the government to insure that the technology is available when needed in manufacturing. The needs are clear...cost-effective lithography for each generation. The challenges are...how to get there from here.