Electron-beam tools have been an important part of the semiconductor industry since its beginnings especially in the area of metrology and are a major contributor towards the goals of process control and yield enhancement. However, in order to stay on the Moore's Law curve instrument performance must be improved by a factor of two every two or three years and with the proximity of fundamental physical limits such major improvements in performance become increasingly difficult to obtain. This paper examines the opportunities that exist for continued improvement, and considers what other options are available for tools to replace the CD-SEM and its associates.

The production of 70nm devices is projected for the year 2008. With this projection, optical lithography will become more challenging since as the device size goes down, the potential for introducing killer defects also increases dramatically. Wafer inspection will play a key role in controlling the defect mechanisms and keeping an acceptable yield for next generation VLSI manufacturing. Metrology tools for the new generation lithography will have the following features. First, the projection and collection lenses will have higher numerical apertures (NA) to obtain high-resolution images. Typically, the NA will be as high as 0.9. Second, the wavelength used for wafer inspection will be much smaller so that common wafer materials will become highly absorptive. Last, with the increased number of process steps, wafer inspection will need to provide information for more critical processes. Based on these features, accurate modeling of the next generation wafer inspection schemes is needed to aid in the characterization and optimization of the inspection tools. The simulation tool must be able to simulate inspection systems with high NA lens, DUV wavelengths, and highly absorptive wafer materials accurately and quickly. At Carnegie Mellon University, a simulator called METRO-3D was developed into a defect inspection simulator for DUV lithography processes. This simulator is able to successfully model various types of defect mechanisms. However, it has experienced occasional numerical instabilities, with discontinuous dielectric structures composed of highly absorptive materials. In order to provide a tool to simulate the wafer inspection scheme of SUB-70NM NODE LITHOGRAPHY, we have incorporated a new algorithm to model the wafer inspection system more accurately and robustly. Numerical experiments shows that the algorithm is capable of simulating topographies with discontinuous dielectric functions, yielding stable results even when the material is highly absorptive. To verify the accuracy of the simulator, several simulations were compared with both analytical models and results form other existing simulators. The results show good matches between METRO-3D and these well-established results. Finally, we performed simulations on industrial data and the results exemplified the ability of METRO-3D to model complex 3-D structures. In this paper we will present is efficient and stable EM solver and the results of the simulator applied to various sub-70nm node wafer inspection schemes.

In this paper the simulation of wafer images for Attenuated Phase Shift Masks (ATTPSM) and repaired binary masks are performed by Virtual Stepper System in a real production environment. In addition, the Automatic Defect Severity Scoring module in Virtual Stepper is also used to calculate the defect severity score for each defect. ADSS provides an overall score that quantifies the impact of a given defect on the surrounding features. For the binary masks, the quality of reported defects is studied. For the ATTPSM three types of programmed defects on both line/space and contact hole patterns are assessed. Wafer exposures are performed using 248 nm imaging technology and inspection images generated on a KLA-Tencor's SLF27 system. These images are used by the Virtual Stepper System to simulate wafer images under the specific stepper parameters. The result are compared to SEM images of resist patterns and Aerial Image Measurement System simulated results.

Many techniques are used to reduce k1 for the 90nm node, including phase shift masks (PSM), assist features and optical proximity correction (OPC) features. Today, in addition to CD line widths, critical measurements of assist features, contact areas, and corner rounding, are now required to verify reticle integrity. New algorithms have been developed and implemented on the KMS100 DUV optical metrology tool to correct for iso/dense bias (optical proximity correction), assess corner rounding effects, and verify contact fidelity and printability. This paper presents new CD metrology studies for Chrome-on-Glass (COG) performed on a KMS100 DUV optical tool using these new metrology algorithms.

Driven by Moore's Law, the quest to double the number of transistors on a given chip every 18-24 months, the density and complexity of patterns on photomasks has increased steadily and significantly. To maintain the fidelity of shrinking features, reticle enhancement techniques such as OPC (Optical Proximity Correction) and Phase Shift Masks are now widely used in optical lithography to extend the lifetime of the existing technology. These techniques (or advanced reticles) provide the desired improvements in spatial resolution, but also complicate the task of reticle defect inspection. In this paper, we present results from ongoing contamination inspections of Embedded (Attenuated) Phase Shift Masks (EPSMs) for 248nm and 193nm lithography. A variety of 248nm masks have been successfully inspected on the KLA-Tencor STARlightTM SL3UV and SL3 tools at Intel Mask Operations (IMO) in Santa Clara, CA. Lessons learned from inspection of 248nm masks are being applied towards inspection of 193nm masks. A representative sample of inspection parameters such as algorithm options, and corresponding inspection results (defect types, capture rates etc.) are presented and discussed in the paper.

Carbon nanotubes (CNT) are among the candidates for atomic force microscopy probes for use in high aspect ratio critical dimension metrology (CDM). Their mechanical strength at small diameters makes them ideal probes for narrow and deep features. The synthesis of CNT has been making great progress in recent years. The use of CNT in scanning probe microscopy, however, has been limited due to a number of problems. While the CNT probes generally appear to be long lasting, the manufacture of precisely aligned CNT of defined length, diameter and number of walls poses a number of challenges. Yet, such precisely defined CNT probes seem to be required if the cantilevers are to be used for CDM. Our result demonstrate, for example, that the attachment angle of CNT with respect to the cantilever beam is crucial for their application in CDM. We report about our efforts to overcome these problems by growing well-defined CNT on standard Si cantilevers using chemical vapor deposition in combination with focused ion-beam machining techniques.

The continental scaling down of CMOS feature size to 100 nm and below necessitates a characterization technique to resolve high aspect ratio features in the nanoscale regime. This paper reports the use of atomic force microscopy coupled with high aspect ratio multi-walled carbon nanotube scanning probe tip for the purpose of imaging surface profile of photoresists. Multi-walled carbon-nanotube scanning probe tip for the purpose of imaging surface profile of photoresists. Multi-walled carbon-nanotube tips used in this work are 5-10 nm in diameter and about a micron long. Their exceptional mechanical strength and ability to reversibly buckle enable resolution of steep, deep nanometer-scale features. Images of photoresist patterns generated by 257 nm interference lithography as well as 193 nm lithography are presented to demonstrate multi-walled carbon nanotube scanning probe tip for applications in metrology.

We have constructed a prototype calorimeter that will serve as a primary standard for measurements of 157 nm excimer laser power and energy. The construction and performance of the prototype will be discussed. In addition, we have performed a series of thermal characterization measurements on the prototype. From these measurements, we deduce that the uncertainty in the prototype's electrical calibration factor is less than 0.2 percent. This number is less than or comparable to the uncertainty of the NIST primary standards for use with 193 and 248 nm excimer lasers. The 157 nm standards are part of a beamsplitter-based measurement system for laser power and energy calibrations. To control and determine the ambient measurement conditions, we have constructed a nitrogen-purged enclosure for this system. We are able to achieve O2 concentrations of less than 3 parts per million (ppm) inside the enclosure.

Acintic measurements of optical transmission are critical for determining the quality of repairs on advanced ArF- generation photomasks. We describe a new 193 nm tool designed to provide mask makers and uses the capability to resolve and measure photomask features with dimensions less than 0.20 micron. Measurements of transmission with sub-1 percent error are achieved in 30 seconds by using a laser probe beam imaged onto the top of the mask surface. We present high-resolution actinic images of embedded attenuator phase shift masks and of binary masks and discuss various methods and results of measuring mask transmission. A scheme to add phase retardation measurement capability to the tool is described, and preliminary results are discussed.

A Caliper CD-AFM under development at Xidex uses dual atomic force microscope (AFM) probes that operate together as a caliper for accessing vertical and highly reentrant sidewalls of high aspect-ratio features. This virtually eliminates the ubiquitous effect of probe width, which has been a large component of uncertainty in optical, scanning electron microscope (SEM), and scanning probe microscopy (SPM) linewidth measurements for decades. The resulting tool architecture eliminates the main model-dependent uncertainties associated with nanometer-scale length measurements. By drastically reducing the magnitude, complexity, and variability of model dependence, the tool is made more robust and easier to calibrate. We have already demonstrated the functionality of a single tilted probe. Future developments include dual tilted probes that scan in coordinated fashion.

As photolithography platforms move from 248nm to 193nm resist systems, the industry's established dimension measurement technique (CD-SEM) causes significant shrinkage of the resist structures during measurement. Many studies have been done to characterize this effect and look for the factors that influence / reduce this shrinkage. While numerous anecdotal mechanisms have been proposed to explain the shrinkage, few theoretical / empirical equations have been proposed to connect the observed effects to fundamental mechanisms. Models are proposed relating physical properties (accelerating voltage, photoresist density, resist e-beam film shrinkage) to the commonly observed CD 'hammer test' shrinkage profiles. The validity of the model assumptions is tested via Monte Carlo simulations, FTIR, e-beam curing, SPM and ellipsometry. These models explain the shape of the CD response to repeated measurements (exponential decay curve) and the magnitude of the shrinkage. These models also offer insight into why lower accelerating voltages cause reduced CD shrinkage, although the models predict that accelerating voltage should be a much more dominant parameter for CD shrinkage than literature has shown to date. Mass loss and density changes were also characterized during e-beam cure to check the validity of the model assumptions.

The Advanced Metrology Advisory Group (AMAG) is a council composed of the chief CD-metrologists from the International SEMATECH consortium's Member Companies and from the National Institute of Standards (NIST). The AMAG wrote and, in 2000, updated the 'Unified Advanced CD-SEM Specification for Sub- 0.18 micrometers Technology to be a living document which outlines the required performance of advanced CD-SEMs for vendor compliance to the International Technology Roadmap for Semiconductors, and also conveys other Member companies' collective needs to vendors. Following this specification, a benchmarking effort of the four currently available advanced CD-SEMs has been performed. This paper presents the result of this effort. Previous studies under this same project have ben published.

Prototype linewidth reference materials with Critical Dimensions (CDs) as narrow as 70 nm have been patterned in silicon-on-insulator films. The sidewalls of the reference features are parallel, normal to the substrate surface, and have almost atomically smooth surfaces. Linewidth calibration begins with the measurement of the electrical CDs of multiple reference features located at a selection of die sites on a wafer. The absolute widths of the cross sections of a sub-set of reference features on several chips that are diced from the wafer are then subjected to high resolution transmission electron microscopy (HRTEM) imaging to determine their physical CDs by lattice-plane counting. Sample preparation for lattice-plane counting by HRTEM Is destructive, and other reference features on the same chip become unusable for reference-material purposes. However, a calibration curve for converting the measured electrical CDs of reference features on other chips on the wafer, known as 'product reference features', to their physical values is obtained. The uncertainty attributed to the physical CD values of the product reference features generally varies inversely with the linear correlation between the cross- section lattice-plane counts and the corresponding electrical CD measurements of the sub-set of reference features that were selected for HRTEM imaging. A linear correlation value of approximately 0.97 has been obtained from a sub-set of 12 HRTEM measurements. In this case, the uncertainty attributed to the physical CD values of the product reference features is believed to be responsible for most of the product reference feature uncertainty. However, it has now been found that a forming-gas annealing treatment appears to prevent the referenced time dependence and thus has the potential for reducing the uncertainty level.

As critical-dimension shrink below 0.18 micrometers , the SPC (Statistical Process Control) based CD (Critical Dimension) control in lithography process becomes more difficult. Increasing requirements of a shrinking process window have called on the need for more accurate decision of process window center. However in practical fabrication, we found that top-down CD-SEM showed its limitations in process window center determination, especially for the best focus. For instance, in some extreme focus situation, resist pattern will show a severe undercutting profile which will affect the DICD reading by top-down CD-SEM with fixed measurement algorithm. This kind of DICD measurement error will finally affect the process window center determination (especially best focus) and in-line DICD monitoring, which will lead to the cost of scrap and loss of time for trouble-shooting. In this paper, we will present a detailed study of DICD best focus determination in case of top-down DICD by experiment and simulation. Further a possible solution to this problem will be described in the latter part of this paper.

One goal of CD metrology is to monitor lithographic process control and how it relates to post-etch results. At present, in-fab process control for this purpose is achieved through top-done CD measurements. To acquire profile information requires destructive cross-section SEM measurements or time- consuming atomic force microscope (AFM) measurements. To find height and profile information about a resist or etched structure directly on a CD-SEM, new techniques using the combination of in-column beam tilt and stereo graphic imaging have been developed and implemented on the Applied Materials VeraSEM-3D.

Optical scatterometry has been proven to be a useful tool for the inspection and assessment of lithographic processes. The characteristic signature of the scattergram provides information about the surface relief profile and can be rapidly and non-invasively acquired. Currently, attention is being focused on the inversion of scatterometric data to determine the surface profile of the relief structure. To overcome the highly non-linear relationship between the properties of the diffracting structure and the diffraction measurements, we present a linearized solution based on a least-squared refinement method. This approach relies on the reliability of lithographic processes, which allows us to determine the departures of structural key parameters from a set of expected design values. This linearized inversion, as shown by the mathematical formalism, is independent of the scatterometric measurement configuration. Hence it can be applied whether, for example, the incident angle or wavelength of the light is varied to acquire either irradiance of phase information. This is validated in this paper by various examples handling the retrieval of three geometrical parameters (groove depth, line width, and sidewall angle) from reflectance data simulated using the rigorous coupled-wave theory (RCWT) for both angular and wavelength scans. An additional benefit of the proposed linearized solution is the ability to examine mathematically the significance of measurement errors. An analytical propagation of error is presented, connecting measurement noise to the soughed parameter precisions and uncertainties. We applied this formalism to the cases mentioned above, where we simulated the retrieval of three parameters from measurements containing various levels of noise. This studies allows us to draw preliminary conclusions on the sensitivity of scatterometry with respect to the number and types of structural parameters to be retrieved, but also to the technique employed to gather the measurements.

We have studied the ability of ellipsometric spectroscopy to determine slight feature shape variations. In this paper, we show that an optical technique can be a sensitive technique to detect slight profile variations, especially at the top of the features. Simulated results as well as experimental results are presented. A sensitivity metrics, that is better suited to ellipsometric measurements than the conventional mean square error, is proposed. It is used to quantify the sensitivity to measure the feature shape of interest and to examine preferable experimental set-up. Computer simulations and ellipsometry signal libraries have been generated using a common rigorous electromagnetic modal approach (MMFE, Modal Method by Fourier Expansion). Experimental verification has been performed on resist features fabricated using DUV lithography. The ability and sensitivity to measure overhanging profiles such as notched polysilicon gates that are impossible to measure using top-down SEM is also mentioned.

Scatterometry is anew optical technique for profile measurement, capable of providing 2 or 3 dimensional profile information. The scatterometry optical system can be made small and economic enough to enable integration into each production tool and is therefore ideally suited for integrated process control. In addition, the range of applications for which this method applicable is wide: lithography photoresist profiles, etch structures, and even measurements of transparent layer thickness above arrays after CVD or CMP steps. Scatterometry is base don the measurement of the optical diffraction characteristics of aperiodic structure. Interpretation is done by evaluating the expected spectrum for a given profile, comparing the result to the measurement, and continuously fitting the profile parameters until a best fit is achieved. The leading physical model used in scatterometry is the Rigorous Coupled-Wave Analysis (RCWA). It is based on dissecting the profile into several layers and treating each as a waveguide with a fixed cross-section. The accuracy of the model is determined by the resolution of the dissection, and the number of Fourier modes retained in the calculation. RCWA gives very accurate results provided sufficient number of modes and layers are used. As calculation time scales linearly with number of layers and roughly cubic with number of modes, this usually leads to long calculation time, thus inhibiting real time interpretation which is especially required for integrated metrology. One possible solution is to consider alternative physical models to replace RCWA, such as Green's Function Integral Formulation. The selection between calculation methods should be determined by the calculation time per given spectrum accuracy. The latter should be determined for each application, considering the required parameters to be measured. In this paper we examine the use of two different models for several applications, which will be briefly described. It will be shown that is some important applications use of GFI may be preferable for real time calculations relative to RCWA.

A systematic study has been conducted to evaluate accuracy and precision of spectral scatterometry used for two-dimensional (2D) characterization of trenches formed in fluorinated silicon glass (FSG). Experiments were done on short-flow dual-damascene Cu interconnect material. Trench critical dimensions (CD) obtained using KLA-Tencor's spectral scatterometer were correlated with data collected using CD atomic force microscope (AFM), CD scanning electron microscope (SEM) and transmission electron microscope (TEM). 3 major trench characteristics were analyzed: trench width, trench depth and sidewall angle. Spectral scatterometry demonstrated an excellent correlation (above 0.96) with CD AFM and SEM in tested trench width range of (80-240) nm and trench depth range of (410-450) nm. Spectral scatterometry showed acceptable correlation of 0.55 and minimal offset of 0.05 degrees with AFM in tested sidewall angle range of (87.5-89) degrees. Spectral scatterometry has demonstrated better than 1.0 nm and 0.2 degrees dynamic precision (3s) for both width and height and sidewall angle, respectively. We conclude that KLA-Tencor's SpectraCD system is capable of accurate and precise 2D characterization of FSG trenches. We recommend scatterometry as a high throughput and non-destructive metrology for trench linewidth and depth monitoring in low-K dielectric interconnect manufacturing.

Scatterometry is gaining acceptance as a technique for critical dimension (CD) metrology that complements the more established scanning electron microscopy (SEM) techniques. Scatterometry determines the dimensions of the submicron structures by inverse diffraction calculations. SEM and scatterometry are complementary in many respects. Therefore, they are likely to coexist in the foreseeable future. Scatterometry and CD-SEM instruments can be integrated to take advantage of the complementary nature of the two techniques. To explore the joint use of scatterometry and SEM measurements, we measured a set of photoresist grating samples with CDs ranging from 240 nm through 40 nm by scatterometer (Sensys CD-i) and cross-section SEM (Hitachi S-4700). Although a cross-section SEM was used as an absolute standard for comparison of profiles and CDs, our conclusions range to include CD-SEM techniques. It was found that for measurements of profiles that were patterned with high uniformity within the measurement area, scatterometry was very effective, and correlated best with SEM measurements. However, in cases of substantial line-to-line profile variations, or for isolated or non-periodic lines, SEM is the more appropriate measurement method.

Pattern distortion caused by resist heating can be a significant contributor to errors in feature size and pattern placement. The Green's function method is suitable for determining the temperature evolution during resist exposure, but the previous published solutions are based on different simplifications and lead to different results. In this paper we present a solution of the multi-layer Green's function with clearer physical meaning. This model takes into account the thermal diffusion in all layers of resist chromium and substrate and can be used to calculate the four-dimensional temperature profiles that exist during and after electron beam exposure. Different writing strategies can cause differing degrees of edge displacement and this can lead to errors in feature size and feature position and to edge roughness.

Conventional optical development rate measurement techniques are generally unsuitable for monitoring the dissolution very tin resist films. Monochromatic systems have inadequate thickness resolution to capture the details of surface and standing wave effects, while traditional polychromatic techniques are generally unable to measure thicknesses below 250 nm. The failure of polychromatic analysis methods occurs when there is an absence of turning points int eh relative reflection spectrum. The exact thickness at which this happens is a function of the wavelength range utilized and the resist material's optical characteristics. A novel measurement method is introduced which allows a polychromatic DRM system to measure any resist thickness. Rather than placing the film under analysis directly on a reflecting substrate, it is spun on a wafer that has a relatively thick transparent film on its surface. The transparent film induces turning points in the relative reflection spectrum. The position of these turning points is modified by the presence of thin resist films in a predictable way, allowing accurate measurement of the resist film, providing the optical and thickness details of the intermediate film are known. Experimental results are presented demonstrating the capability of the technique to measure the dissolution rates of films with initial thickness ranging from 56 nm to 4400 nm. The ability of the method to resolve fine dissolution detail, such as standing waves and surface effects is also presented.

As geometrical dimensions of semiconductor devices decrease, the need to introduce Cu processes into the fabrication cycle becomes increasingly important as a means of maintaining line resistances and circuit time constants. However, the success of implementing such as fabrication process is dependent on the ability to characterize it through quantitative means, such as Overlay metrology. In this paper we examine the overlay measurement results which have been obtained on a Cu based CMOS process at the 0.12 (Mu) m technology node. Overlay measurements were taken over a wide range of process conditions, and included wafers exhibiting extreme image contrast reversal, grainy conditions and low contrast. These factors have traditionally led to a decreased ability to make repeatable measurements, if the measurements could be made at all. Our results cover the important metrics of overlay metrology, and include precision, recipe portability, and measurement success rates. The results suggest that the overlay metrology issues encountered with such leading edge processes need not pose intractable barriers to obtaining reliable overlay metrology data.

The measurement of optical constants for materials intended for use at wavelengths less than 200 nm is critical for production monitoring and process control when manufacturing devices with features of 0.15 micrometers or smaller. The challenges of organic ARC film measurement at 193 nm are due to the more complicated optical spectral signatures of the organic films and the necessary signal to noise improvements in the ultraviolet. Sharp features in the spectra of the organic ARC films are used to control the extinction value of the films, and they complicate the analysis of the optical constants. Research grade instruments available have shown the ability to accurately measure these films at 193 nm. We report here the results of using next generation optical metrology equipment to measure the optical constants on these difficult organic ARC films on a production level. Specific examples of production level monitoring of the optical constants of these films will be presented.

Conventional overlay control minimizes the registration error between aligned level pairs. Post alignment corrections are applied to the adjustable parameters on the exposure tool. These can be grouped into grid terms relating to stepper stage control and field terms related to mask, stage and lens control. To minimize alignment time and alignment mark real estate on the wafer, common practice is to align at a fixed field location over a sufficient number of fields to represent the grid. A problem with conventional overlay control is that it places no constraint on the magnitude of field terms at first-level or any subsequent level; it requires only that they be matched between the alignment pairs. Incorporation of field-stitching overlay control within each level establishes the constraint that field terms remain consistent with the wafer stage stepping at each process level. A novel target layout and analysis method enables the simultaneous determination of both conventional overlay and neighboring-field error. Application of our method at first level and subsequent alignment levels precludes the random walk of field terms, thereby improving the consistency of field matching. The net result is an improvement in overlay control, particularly in cases where 'single-point' alignment or mixed tool set matching is required.

The Overlay Metrology Advisory Group (OMAG), which includes representatives from International SEMATECH Member Companies and the National Institute of Standards and Technology, has collaborated to create a unified specification for overlay measurement tools [1]. The methodology and results of an overlay benchmarking comparison of several tools are discussed in this paper. As device technologies shrink below the sub-130nm range, a critical need arises to develop more precise tools to measure overlay. Overlay metrology capability needs to be available for detecting and controlling total device overlay regardless of the source of error. The misregistration measurement uncertainty introduced by the overlay tool can be compared for several systems. A benchmarking study is currently underway and focuses on the existing technique of optical measurement of centerline offsets in different target designs. The critical parameters that the study analyzes include precision, accuracy, throughput, through focus measurements, and recipe portability. Imaging issues such as low contrast targets, across wafer thickness variation, CMP effects, and grainy metal targets can contribute greatly to overlay errors. Several process stacks were designed to incorporate some of these imaging issues and test the limitations of the overlay tools. The same set of wafers and test locations were measured at each supplier site and the results were analyzed. This paper focuses on the methodology used for overlay benchmarking and examples of the results generated with respect to the parameters tested.

While overlay precision has received much focus in the past, overlay accuracy has become more significant with shrinking process budgets. One component of accuracy is the difference between pre-etch (DI) and post-etch (FI) overlay, which is a function of wafer processing parameters. We investigated a specific case of overlay between metal and contact layers of a 0.16 mm SRAM process. This layer was chosen because a significant amount of wafer contraction was observed between DI and FI, resulting in as much as 30nm of DI-FI overlay difference. The purpose of the study was to characterize the systematic DI-FI differences and gain understanding of the wafer processing parameters that affect the DI-FI differences. A designed experiment showed how certain overlay mark widths were less sensitive to processing parameters. AFM profiles of the prior-level overlay marks identified issues with mark widths 1.0um or smaller. By performing localized etches on the inner vs. outer marks of the overlay targets, it was noted that the majority of the wafer contraction was induced by etching the outer (prior level) mark. Production measurements at photo and etch showed the wafer contraction to be fairly stable over a month timeframe and independent of device and exposure tool, though large fluctuation shifts in wafer contraction were noted over a nine-month period. The methods used in this study can be helpful in understanding other DI-FI processing issues.

As the increasing demand of thinner CD for higher IC density and faster signal response, the techniques, such as OPC, PSM, and OAI was introduced to extend the optical resolution-limit below sub-100-nano. The lens quality is more important as critical dimension (CD) becomes smaller even using scanner instead of stepper. Pattern displacement error (PDE) resulting from lens aberration will be a phenomenon worthy to emphasize when technology design rules reaches sub-quarter micron generation and beyond1, 2. Because of inevitable coma aberration of lens, conventional 2úgm bar-in-bar overlay pattern cannot accurately represent real displacement of fine patterns in device. In this paper, evaluation of overlay pattern displacement with respect to lens coma aberration and its dependence on different feature size and structure are carried out. Different generation technology should combine with overlay pattern of proper feature size. Several illumination methodologies and NA/Sigma setting are also discussed in this paper.

An evaluation of the ASML ATHENA alignment system was performed using marks placed in the scribeline of an Intel test chip used for 130 nm node processing. Exposures were performed using two ASML scanners at IMEC at the isolation, gate, contact and first metal layers, with extreme process splits performed at Intel at critical steps in between. The splits were arranged as factorials in order to evaluate the process sensitivity of the alignment system. The modeled overlay terms with the largest sensitivities are discussed. In addition, data taken by the scanners on various mark/recipe combinations is analyzed to provide insight into overlay optimization and potential alignment system limitations.

In order to control and minimize overlay metrology errors, we have to deal with a number of design parameters both in the metrology tool domain and in the overlay target domain. For enhancing the rate of performance improvement vs. technology investment, simulation can be used for modeling both the effects of overlay metrology tool behavior and the impact of target designs on the ultimate metrology performance.

Optical scatterometry has attracted a great deal of interest for linewidth measurement due to its high repeatability and capability of measuring sidewall shape. We have developed an analogous and complementary technique for the scanning electron microscope. The new method, like scatterometry, measures shape parameters (e.g., wall angles) as well as feature widths. Also like scatterometry, it operates by finding a match between the measured signal from an unknown sample and a library of signals calculated for known samples. A physics-based model of the measurement is employed for the calculation of libraries. The method differs from scatterometry in that the signal is an image rather than a scattering pattern, and the probe particles are electrons rather than photons. Because the electron-sample interaction is more highly localized, isolated structures or individual structures within an array can be measured. Results of this technique were compared to an SEM cross section for an isolated polycrystalline silicon line. The agreement was better than 2 nm for the width and 0.2{degrees} for wall angles, differences that can be accounted for by measurement errors arising from line edge roughness.

The in-line and at-line measurement tools for critical dimension (CD) metrology in semiconductor manufacturing are technologically advanced instruments that exhibit excellent measurement repeatability - below one nanometer in some cases. Accuracy, however, is largely dependent upon the availability of traceable standards. Because the standards requirements of this fast-paced industry are particularly demanding and application-specific, metrology traceability is sometimes lacking. International SEMATECH (ISMT) and the National Institute of Standards and Technology (NIST) are working together to improve this situation. We are developing a reference measurement system (RMS) at ISMT using a critical-dimension atomic force microscope (CD-AFM). We are performing measurements needed to establish the traceability chain and develop uncertainty budgets for this tool. Monitoring of tool performance has been improved and we have performed preliminary checks of lateral and vertical scale calibration. Preliminary uncertainty budgets for pitch and height measurements have been developed. At present, the standard uncertainty due to scale calibration and non-linearity is estimated to be approximately 0.2 percent for pitch measurements and 0.5 percent for step height measurements. Our initial checks of scale calibration were performed using samples for which a full traceability chain is not available. We expect to reduce these uncertainties once we are able to use samples with a complete traceability chain. Ultimately, our major objective in developing an RMS is to provide a traceable metrology reference for other major projects at ISMT - including CD-SEM benchmarking, AMAG wafer development, and overlay tool benchmarking.

Traditionally, CD-SEM resolution has been measured using edge and spacing measurements, where images are subjectively examined and evaluated as to edge transition width between arbitrary amplitudes or minimum measurable spacings between particles. Occasionally, methods form traditional optics, i.e., application of the Rayleigh criterion, have been quoted, but the more conservative results have caused users either to relax these traditional definitions or to abandon them altogether. More recently, as in light optics, Fourier methods, where frequency space results are used to define the resolution of the system, have been applied. In this paper, these methods are surveyed and applied to experimental data. In particular, Fourier transform approaches are examined, and the difficulties in their application vis-a-vis separation signal from noise, maintaining repeatability, and eliminating system and processing artifacts addressed. A method based on the Rayleigh criterion for diffraction limited systems is proposed that minimizes these difficulties by processing out system effects and distancing the evaluation region form both the noise and system artifacts. This method has the particular advantage of not requiring operator intervention, ensuring consistent results form user to user.

In this work we describe a simple and easy way to implement line edge roughness measurement using the Applied Materials VeraSEM 3D CD SEM. The outcomes of the measurement are the roughness amplitude and its main spatial wavelength components. In the first step we test this method on an e-beam resist wafer with known roughness and correlate the results with AFM measurements. In the second step, we measured the line roughness of a 193 nm resist FEM wafer. It was found that negative defocus is characterized by long wavelength components while positive defocus shows line roughness at all wavelengths.

The shrink of semiconductor fabrication ground rule continues to follow Moore's law over the past years. However, at the 100 nm node, the fabrication cost starts to rise rapidly. This is mainly due tot he increase of complexity in the fabrication process, including the use of hard masks, planarization, resolution enhancement techniques, etc. Smaller device sizes require higher alignment tolerances. Also, higher degree of complexity makes alignment detection more difficult. For example, planarization techniques may destroy mark topography; hard masks may optically bury alignment marks, and more film layers makes the alignment signal more susceptible to process variations. Therefore in order to achieve reliable alignment, it is absolutely critical to develop an accurate and fast simulation software that can characterize alignment performance based on the film stack structure. In this paper, we will demonstrate that we have built an extremely fast alignment performance based on the film stack structure. In this paper, we will demonstrate that we have built an extremely fast alignment signal simulator for both direct imaging and diffractive detection system based on simple optical theory. We will demonstrate through examples using our advanced DRAM products that it is capable of accurately mapping the multi-dimensional parameter space spanned by various film thickness parameters within a short period of time, which allows both on-the-fly feedback in alignment performance and alignment optimization.

As overlay budgets quickly approach a total allowed variation of around 30 nanometers serious consideration must be given to the technique used to measure overlay between successive layers. This problem becomes particularly acute if different machines are used to expose alternating layers, or if different illumination settings are used between successive layers. As a result of overlay degradation observed between two production layers, a comprehensive study was initiated to evaluate the accuracy of conventional 'frame in frame' and 'box in box' alignment schemes. The insufficient accuracy of these structures is demonstrated using experimental techniques and through the use of photolithographic simulations. Using these simulations, new overlay markers were developed to minimize the sensitivity to Zernike fringe polynomial lens aberrations in both steppers and scanners. The improved accuracy of these overlay markers is finally demonstrated by experimental techniques. Despite significant improvements in the measurement accuracy of these new overlay markers, the accuracy of present techniques may not be sufficiently accurate to measure overlay as budgets approach 30 nanometers, and new techniques may have to be developed.

As overlay budgets shrink with design rules, the importance of overlay metrology accuracy increases. We have investigated the overlay accuracy of a 0.18mm design rule Copper-Dual-Damascene process by comparing the overlay metrology results at the After Develop (DI) and After Etch (FI) stages. The comparisons were done on five process layers on production wafers, while ensuring that the DI and FI measurements were always done on the same wafer. In addition, we measured the in-die overlay on one of the process layers (Poly Gate) using a CD-SEM, and compared the results to the optical overlay metrology in the scribe-line. We found that a serious limitation to in-die overlay calibration was the lack of suitable structures measurable by CD-SEM. We will present quantitative results from our comparisons, as well as a recommendation for incorporating CD-SEM-measurable structures in the chip area in future reticle designs.

The continuing reduction of IC device dimensions puts stringent demands on the corresponding overlay performance. As part of the total overlay budget, the effects of the different process parameters need to be characterized and well understood. In a joint development program between IMEC and ASML, the robustness of different alignment strategies to process parameters has been evaluated using the ATHENA alignment system. This paper looks at both Front-end (Shallow Trench Isolation) and Back-end (W-CMP and copper dual damascene) processing. To investigate the effect of STI processing on alignment marks in Front-end processing an extensive evaluation has been performed in which both mark design and process parameters have been varied. The robustness to typical long term process variation at the STI CMP step in a production environment has also been evaluated. To improve the robustness of alignment marks in Back-end processing, new mark designs have been evaluated. These designs have been evaluated for two different processes. The first uses traditional W-CMP and sputtered aluminum. The second uses copper dual damascene, with layer stacks consisting of both conventional and low-k dielectric materials. This knowledge will be used to generate alignment strategies for future technology nodes.

In this paper, we describe a new method for the separation of tool-induced measurement errors and sample-induced measurement errors. We apply the method to standard overlay target configurations. This method is used to separate the effects of the tool and sample errors in the measured optical intensity profiles and to obtain the best estimate of the correct intensity profile for a given sample geometry. This most accurate profile is then compared to calculated profiles from two different theoretical models. We explain the modeling in some detail when it has not been previously published.

In this paper, variability in the photolithography pattern transfer process is analyzed by means of a large number of CD measurements spread across the stepper field and across the wafer. The variability is found to be highly systematic in nature, and methods are developed to extract the parameters of this systematic variation. Knowledge of the structure of the systematic variance allows for the selection of an optimum sampling plan to best capture the variance in future measurements of wafers from the same process.

This paper studies the resolution limits of optical lithography using phase-shift mask technology. Alternating phase-shift masks are studied in the context of the 70nm lithographic node. The performance of alternating phase-shift masks is measured at the limits of its resolution. Printing issues that result from phase errors on the reticle are observed. These phase errors result in a loss of depth of focus and ultimate resolution. A lithography-system-based solution to the mask phase errors is proposed. A small obscuration aperture that matches the partial coherence of the illumination is placed in the pupil plane of optics to block the zero-order light leakage that results from the phase error. This paper reports the performance and effects of using this obscuration aperture.

Electronic structure of chromium aluminum oxynitride is calculated with the embedded cluster method within the framework of discrete variational (DV)-Xa method. Then calculation is verified by comparing with the experiments of photoelectron spectroscopy such as X-ray photoelectron spectroscopy (XPS) and ultraviolet photoelectron spectroscopy (UPS). Electronic structure calculated theoretically is well consistent with photoelectron spectra under valence band. When nitrogen is substituted for oxygen in Cr-Al-O system, N2p level appears in the energy range between O2p and Cr3d level. So the valence band of chromium aluminum oxynitride becomes broader and the band gap becomes smaller than that of chromium aluminum oxide. We expect that this valence band structure of chromium aluminum oxynitride change the transmittance slope in mask application. We can properly elucidate the electronic structure of chromium aluminum oxynitride system through the theoretical calculation and experimental verification.

This paper introduces a compact but rigorous theoretical formulation of the mask error factor. The results compare favorably to experimental data as well as to first-principle simulations. This work provides an insight into the MEF causes, and it explores its dependency on process settings in order to control critical dimension (CD) variation.

With the advent of alternating aperture phase shift masks it has become necessary to Monitor quartz etch depth to ensure quality mask fabrication. To date, optical tools have been primarily used for the phase metrology of these masks. Optical tools have the advantage of using actinic light, but they also have several limitations, most notably: restrictions on the minimum aperture size to be measured, prior knowledge of which aperture is etched and alack of automation. Atomic force microscopes have a long history of depth metrology and the current tip technology is small enough to extend for the next four reticle generations while new tips are being developed to meet semiconductor device roadmaps. We examine several practical issues of using an AFM based system for the phase metrology of APSM masks; matching of AFM to optical tools automation of phase measurement and an examination of the theoretical difference between AFM and optical measurements.

Automated top down critical dimension scanning electron microscopy (CD SEM) remains the tool of choice for critical dimension process control and dispositioning. Although CD SEMs have limitations, their combination of throughput, resolution, precision, ability to measure any feature of interest and automation has been unmatched in recent years. As geometries shrink in non-uniform ways and manufacturing processes become more complicated, questions continue to recur concerning the degree of correlation between post- develop and post-etch CD measurements. For the particularly critical processing steps of forming the transistor gate, the correlation of these measurements to subsequent electrical measurements is also of major concern. Can CD SEM measurement parameters be better optimized to improve these correlations. This study attempts to answer this question by gathering CD SEM raw waveforms and AFM linescans at gate develop and etch and comparing to electrical test measurements for an advanced device fabrication process. Optimal settings of the CD algorithms working on the waveforms are then sought to improve these correlations. The tradeoff with measurement precision will also be discussed.

In this paper, we discuss a methodology for characterizing and comparing the lithographic process from the CD control point of view. In contrast to the traditional method of exposure latitude vs. depth of focus widely used in lithography, we use the statistical line width variation Cpk, and Cp as the measures of the lithographic process performance and capability. We have developed a software package called Megastat, which estimates CD variation based on experimental focus and exposure matrix (Bossung Curves) and assumed or independently measured focus and dose stochastic behavior. A similar approach was considered in. The impact of focus and dose control on across chip line width variation (ACLV) and across wafer line width variation (AWLV) was quantified with Megastat and compared to the experimentally obtained CD control for a particular 248nm lithographic process for the 130nm technology node. The statistical approach utilized in the Megastat software produced a very good match with the experimental results, which confirmed the usefulness of the statistical approach to characterizing the lithographic processes. We have additionally devised methods for decoupling and determining the actual experimental statistical variations inherent in lithographic processes. These methods allow the input 1 sigma variations for the statistical window analysis to be accurately determined. The usefulness of these methods will also be shown.

New imaging technique in AFM has been developed to suppress bending of the sharpened probe during scanning. After analyzing the bending of probe, it is clear that the bending is caused by response of servo control and slipping of the probe on the slope. It is needed that we do not scan the probe under contact of it on a sample surface for friction- free and we approach it to the surface at a contact force of < 10nN for slip-free. The technique controls the probe such that approaching and gap-controlling are done without scanning after only one step xy-scanning is competed, and the xy-scanning is done without servo controlling after lifting the probe up from the sample surface. This technique permits to use very sharped and slim probe and to observe a steep structure such as a dry-etched groove and hole, and a photoresist pattern with a high aspect ratio faithfully. We can clarify that it is possible to apply this technique to monitoring the steep structures as an in-line process monitor in the LSI process without cracking the wafer.

Optical proximity correction using both model-based and selective size adjust techniques was used as means to correct the distortion of patterns in silicon from drawn data. The performance of these corrections on silicon was evaluated using top-down SEM for both active area and gate wafers post-pattern and post-etch steps. Figure of merit such as corner rounding radius and line-edge roughness were measured. By overlapping the drawn data to the processed image, metric such as overlapping area and critical shape difference were extracted. This resulted in a range of metrics that quantified the performance of the proximity correction that was applied on the mask with reference to drawn data.

Lithography process control is one of the greatest challenges of modern semiconductor manufacturing. Due to the complexity of the overall lithography process, the number of process elements that can contribute to the process yield is large. One area of chief concern ins the control of focus on the lithography tool. Determination of the center of focus for a fixed dose during the photoresist development step in wafer processing is critical. Furthermore, dose variations can compound the difficulty in determining this center of focus. The lenses that are used in steppers have a very limited depth of focus, so utmost precision is necessary. Lense that are in focus will yield sharper printed photoresist images, and lack of focus will result in improperly developed photoresist features. Being at the center of focus also improves process repeatability. Scatterometry is a technology well-suited for lithography inspection and process control because it provides rapid measurement data and can be used for the measurement of resist line profiles. In this research we will report on several different methods for lithography focus control, including approaches that use conventional scatterometry models as well as a model-less algorithm. Comparisons will be drawn between commanded focus offsets, stage tilts and the focus response from the scatterometer. Use of the focus control techniques to map illumination modes and field non- linearity will also be shown.

The Post-Exposure Bake (PEB) is one of the critical pattern transfer steps in deep ultra-violet 248nm, 193nm (DUV) photolithography. The objective of the PEB is to activate the photo-acid produced during the exposure, which, in a self-catalyzing sequence, attacks the bonds of the organic compounds within the resist, making them soluble to the developer solution. This sequence generates more photo-acid, and the cycle continues until the process self-quenches. The acid is also rapidly diffusing along its concentration gradients, further impacting the final dimensions of the resulting pattern. All these phenomena are very sensitive to the PEB temperature trajectory, especially during the early stages of the process. Using existing state of the art equipment calibrated specifically to measure PEB effects, it has been reported that as much as 30 percent of the critical dimension (CD) error budget is due to the PEB variations from plate to plate or across one plate. Virtually all practical PEB procedures call for the rapid heating of the wafer to a recommended target temperature and time, followed by the rapid cooling down to near room temperature. The temperature and timing of the PEB and subsequent cooling are optimized for the particular photoresist chosen for processing. Due to the transient nature of the post exposure bake, aspects such as time-to-temperature, temperature overshoot, as well as steady-sate behavior must be measured, understood, and optimized in order to improve CD control, while minimizing, and if necessary, accommodating any variation across the PEB plate. Until recently, reliable measurements of the actual PEB temperature trajectories have been difficult to obtain. The PEB temperature mapping system provided by OnWafer Technologies allows for these conditions, since it is capable of processing inside the lithography track system as if it were a product wafer due to its design and its wireless nature. Furthermore, it has more than 40 rapid response sensors spread across the wafer surface, and thermal mass close to that of a product wafer. In this way, it can collect the temperature as a function of time during loading, transporting, heating and cooling, and it can provide an accurate view of the thermal history across the surface of a wafer. This information, coupled with physical models of the equipment and of the process can be used to more accurately characterize PEB behavior and its impact in CD control. Experiments done to date using the OnWafer system confirm that typical PEB steps can be too short for reaching steady-state conditions. More specifically, state of the art bake plates were found to require several seconds to heat a typical wafer within a degree of the target temperature, and they can overshoot the target temperature by a degree or more. Moreover, these plates may need a few minutes to stabilize within 0.1 degrees C from that target. Similar time constants apply during the cooling sequence. Experiments involving detailed gate layer CD measurements further validate the hypothesis that transient effects are responsible for much of the CD errors due to PEB, especially since these transient effects have a traceable non-uniformity. This paper will highlight the metrology involved in detecting PEB non-uniformities, demonstrate how the results of real-time post exposure bake data collection differ from 'static bake data test' collection and provide in-line bake data alongside collected CD data.

As the number of varied devices produced by a fab increase, coupled with an increased complexity in those devices which call for an ever increasing number of process layers, in- line process control via metrology can become an impossible task, unless metrology recipe management schemes are implemented. Logic fabs are now introducing more than 1 new device per day, which can result in the writing and management of thousands of recipes, which in turn can lead to the costly consumption of tool and personnel resources sand a general loss in productivity. In this paper we present the productivity gains to be made in the recipe creation process through off-line recipe generation, as well as a method of decreasing the recipe optimization time. We will also outline the concept of Just In Time recipe creation, its contribution to productivity gains, and its generalized implementation with respect to Overlay Metrology recipes.

As critical dimensions continue to decrease with each technology node, the precise characterization of line width and profile becomes an increasingly challenging task. Small angle neutron scattering (SANS) offers several advantages for the characterization of sub-100 nm structures, particularly as a calibrating measurement method. In this work, SANS is used to characterize three samples prepared with the same mask and focus conditions, but different photoresist formulations. The mask pattern consists of parallel lines with a nominal line width of 180 nm and a 1:2 line to space ratio. Scattering data are taken using both a focused neutron beam instrument (two-dimensional data) and a perfect crystal diffraction ultra-high resolution small angle neutron scattering (USANS) instrument. From the location and intensity of observed diffraction peaks, both the periodicity of each grating pattern and the average line widths are determined from simple analytic expressions with nanometer resolution.