The growing complexity of reticles and continual tightening of defect specifications causes the reticle defect disposition
function to become increasingly difficult. No longer can defect specifications be distilled to a single number, nor can
past simple classification rules be employed due to the effects of MEEF on actual printing behavior. The mask maker
now requires lithography-based rules and capabilities for making these go/no-go decisions at the reticle inspection step.
We have evaluated an automated system that predicts the lithographic significance of reticle defects using PROLITHTM
technology. This printability prediction tool was evaluated and tested in a production environment using both standard
test reticles and production samples in an advanced reticle manufacturing environment. Reference measurements on
Zeiss AIMSTM systems were used to assess the accuracy of predicted results.
The application, called the Automated Mask Defect Disposition System, or AMDD, models defective and non-defective
test and reference images generated by a high-resolution inspection system. The results were calculated according to the
wafer exposure conditions given at setup such that the reticle could be judged for its 'fitness-for-use' from a lithographic
standpoint rather than from a simple physical measurement of the film materials.
We present the methods and empirical results comparing 1D and 2D Intensity Difference Metrics (IDMs) with respect to
AIMS and discuss the results of usability and productivity studies as they apply to manufacturing environments.
The combined impact of longitudinal chromatic aberrations, focus-drilling, and Z-noise on several lithographic
performance metrics is described. After review, we investigate an improved method for simulating the lithographic
behavior of longitudinal chromatic aberrations stemming from the finite bandwidth of excimer laser pulse-spectra
using PROLITHTM v. 9.3.3. Additionally, we explore two methods for modeling the lithographic improvements related
to focus-drilling and new PROLITH functionality for modeling the effects of Z-noise. Our case studies involve reinvestigating
the RELAX process and providing a framework for accurate lithographic simulation using machine
specific pulse-spectral data, modified Lorentzian, and Gaussian models. After presentation and analysis, we discuss
potential applications including methods for improved focus budgets and improved mask design.
The use of hardware-based and software-based reticle defect printability simulation systems is expanding as the cost and complexity of reticles increases. Without such systems it has become increasingly difficult to predict the lithographic significance of a defect found on a reticle. The viability of such systems can be judged using several criteria including accuracy, ease of use, level of automation, and the degree to which they can be applied to a wide range of reticle types. Simulation systems have improved in each of these areas. Automated and semi-automated systems have now been developed and integrated into reticle manufacturing. We report on advances made in a software-based simulation system which uses high-resolution reticle inspection images as the basis for the description of the reticle. We show that the simulated aerial images can be compared quantitatively to results from a hardware-based simulation system (the Zeiss AIMSTM tool) for both 193 and 248 nm EPSM reticles. The development of a new set of metrics to judge lithographic significance will be explained. Common procedural mistakes in evaluating the impact of a defect will be discussed.
The present approach to Optical Proximity Correction (OPC) verification has evolved from a number of separate inspection strategies. OPC decoration is verified by a design rule or optical rule checker, the reticle is verified by a reticle inspection system, and the final wafers are verified by wafer inspection and metrology tools. Each verification step looks at a different representation of the desired device pattern with little or no data flowing between them.
In this paper, we will report on a new inspection system called DesignScan that connects the data between the various abstraction layers. DesignScan inspects the OPC decorated design by simulating how the design will be transferred to the reticle layer and how that reticle will be imaged into resist across the full focus-exposure process window. The simulated images are compared to the desired pattern and defect detection algorithms are applied to determine if any unacceptable variations in the pattern occurs within the nominal process window. The end result is a new paradigm in design verification, moving beyond OPC verification at the design plane to process window verification at the wafer plane where it really matters.
We will demonstrate the application of DesignScan to inspect full chip designs that utilized different Resolution Enhancement Technique (RET) and OPC methods. In doing so, we’ll demonstrate that DesignScan can identify the relative strengths and weaknesses of each methodology by highlighting areas of weak process window for each approach. We will present experimental wafer level results to verify the accuracy of the defect predictions.
Photomasks with small dense features and high mask error enhancement factor (MEEF) lithography processes require stringent reticle quality control. The ability to quickly and accurately measure reticle defects on a high-resolution inspection system and to simulate their impact on wafer printing are key components in ensuring photomask quality. This paper discusses the correlation of measurements made with UV and DUV-based inspection systems; simulation performed with a 193nm aerial image review tool and aerial image simulation software. Ease-of-use is discussed for each technique. Data accuracy is compared to measurements performed by a Scanning Electron Microscope (SEM) on mask and wafer. Tests show that the inspection system can quickly and accurately determine sizes of most defects. The study also indicates that the simulation techniques can accurately tract the lithographic results, and can be used to reduce or eliminate the use of test wafers and expensive lithography and wafer metrology time. The outcome of this study leads to better defect dispositioning by providing techniques to determine the size and printability of reticle defects.
Lithography simulation is being used in a wide range of applications to help lithographers solve an equally wide range of problems. A necessary input to optical lithography simulation is the specification of the mask transmittance function, m(x,y), which forms the basis for the aerial image calculation. Various methods are used to specify m(x,y). The choice of method depends, in part, on the capabilities of the simulation software package and the available information. To maximize effectiveness, efficiency and accuracy, users should choose a method of specifying m(x,y) which considers the requirements of their application. In many cases, a simple expression for m(x,y) is all that is needed. In other cases, finer detail is desirable or even necessary. This paper reviews many techniques to generate m(x,y) for the PROLITH family of lithography simulators and presents current research for the defect printability application.
Today's reticle inspection tools can provide a wealth of information about defects. We introduce here a system called DIVAS: Defect Inspection Viewing, Archiving, and Simulation that fully uses and efficiently manages this wealth of defect information. In this paper, we summarize the features of DIVAS and describe in more detail PRIMADONNA, one of its components. Current reticle defect specifications are based, primarily, on defect size. Shrinking design rules, increasing MEEF and use of Optical Enhancement Techniques cause size to be an inadequate criterion for disposition. Furthermore, visual disposition of defects is not automated, strictly reproducible, or directly tied to wafer lithography. To compensate for these inadequacies, reticle specifications are set conservatively adding direct and hidden costs to the manufacturing process. PRIMADONNA, utilizing Prolith as the simulation engine, retrieves all defect and reference images saved from a KLA SLF77 inspection tool and processes them through a series of increasingly rigorous simulation stages. These include pre-filtering, aerial image formation, and post filtration. Difference metrics are used to quantify a defect's wafer impact. We will report results comparing PRIMADONNA decisions to manual classifications for a significant volume of inspections. Correlation between PRIMADONNA results and AIMS metrology will be presented.
Lithography at its limit of resolution is a highly non- linear pattern transfer process. Typically the shapes of printed features deviate considerably from their corresponding features in the layout. This deviation is known as Optical Proximity Effect, and its correction Optical Proximity effect Correction or OPC. Although many other so-called optical enhancement technologies are applied to cope with the issues of lithography at its limit of resolution, almost none of these can re-store the linearity of the pattern transfer. Hence fully functional OPC has become a very basic requirement for current and future lithography processes. In general, proximity effects are two-dimensional (2d) effects. Thus any measurement of proximity effects or any characterization of the effectiveness of OPC has to be two- dimensional. As OPC modifies shapes in the data for mask writing in a way to compensate for the expected proximity effects of the following processing steps, parameters describing the particular OPC-mask quality is a major concern. One-dimensional mask specifications, such as linewidth mean-to-target and uniformity, pattern placement, and maximum size of a tolerable defect, are not sufficient anymore to completely describe the functionality of a given mask for OPC. Two-dimensional mask specifications need to be evaluated. We present in this paper a basic concept for 2d metrology. Examples for 2d measurements to assess the effectiveness of OPC are given by the application of an SEM Image Analysis tool to an advanced 130nm process.
For some applications, the usefulness of lithography simulation results depends strongly on the matching between experimental conditions and the simulation input parameters. If this matching is optimized and other sources of error are minimized, then the lithography model can be used to explain printed wafer experimental results. Further, simulation can be useful in predicting the results or in choosing the correct set of experiments. In this paper, PROLITH and ProDATA AutoTune were used to systematically vary simulation input parameters to match measured results on printed wafers used in a 193 nm process. The validity of the simulation parameters was then checked using 3D simulation compared to 2D top-down SEM images. The quality of matching was evaluated using the 1D metrics of average gate CD and Line End Shortening (LES). To ensure the most accurate simulation, a new approach was taken to create a compound mask from GDSII contextual information surrounding an accurate SEM image of the reticle region of interest. Corrections were made to account for all metrology offsets.
A complete evaluation of the optical proximity effects (OPE) and of their corrections (OPC) requires a quantitative description of two-dimensional (2D) parameters, both at resist- and at reticle-level. Because the 2D behaviour at line-ends and at line-corners can become a limiting factor for the yield, it should be taken into account when characterising a process, just as the CD- and pitch-linearity are already kept under control. This implies the measurement of 2D-metrics in a precise way. We used an SEM Image Analysis tool (ProDATA SIAM) to define and measure various OPC-relevant metrics for a C013 process.
For the METAL (M1) process, we show that the overlap between line-ends of M1-trenches and underlying nominal contacts is a relevant metric to describe the effectiveness of hammerheads. Moreover, it is an interesting metric to combine with the CD process window. For the GATE process, we demonstrate that for a given set of metrics there is a degree of OPC aggressiveness beyond which it is not worth to go. We considered both line-end shortening (LES) and corner rounding affecting the poly linewidth close to a contact pad, and this on various logic circuits having received different degrees of fragmentation. Finally the knowledge of the actual line-end contour on the reticle allows one to simulate separately the printing effect of that area loss at reticle line-ends. The area loss measured by comparing the extracted contour to the target one is regarded as a combination of pull-back and area loss at corners. For our C013 gate process, and for the 130nm lines at a 1:1.25 duty cycle, those two parameters contribute together to approximetely 40% of the measured LES in the resist. This fact raises the question of specifications on 2D reticle parameters. We also find a linear correlation between the area loss at reticle line-end corners and the corresponding increase of LES on the wafer, which suggests a way towards putting specifications on the reticle line-ends.
With the increased resolution of today's lithography processes, reticle pinhole defects are much more printable. Measuring the size of small pinholes using the current SEM method often produces erroneous results when compared to pinhole energy transmission. This is mainly due to the fact that SEMs do not accurately account for edge wall angle and partial filling which can dramatically reduce the pinhole transmission and subsequent printability. Since reticle inspection tools, like wafer steppers and scanners, use transmitted illumination, pinhole detection performance based upon top surface SEM defect sizing is often erroneous for small pinhole diameters. This study first uses simulation to predict printability. Then, a pinhole test reticle is developed with a variety of sub-200nm pinholes. The reticle pinholes are measured with an improved method incorporating transmission and imaged to wafer in order to assess printability.
We are jointly examining several aspects of reticle blank inspection. Previously, we have reported on our progress inspecting PBS blanks with the KLA-Tencor SL300 STARlight system. Using a simple test reticle we have measured, in a reproducible and qualitative way, the probability that a reticle blank defect will transfer to the finished reticle. Our study evaluated samples containing 452 blank defects. In some cases, the transfer probability is higher than 80 percent. We have now expanded our research to include reticles with optical resist such as 895I. Using eight optical blanks and a CD uniformity test pattern, we have found that the scanning process of the inspection does not measurably change the optical blank characteristics. However, there is clear evidence that the calibration process does impose some limitations on the use of non- destructive optical blank inspection. Our proposals to overcome those limitations are discussed.
Photronics and KLA-Tencor are jointly examining several aspects of reticle blank inspection. PBS blank quality has been examined using the KLA-Tensor STARlight reticle inspection system. PBS blanks were inspected using a 500-nm pixel with the highest sensitivity settings. Data from the initial phases of this study show conclusively that blank defects with certain characteristics can, with high probability, `transfer' to the finished reticle. These conclusions were drawn from a systematic study of several test samples containing 452 blank defects. These defects were classified using three different characteristics and correlation studies were completed to determine which factors most significantly influence transfer rates. This study has now been expanded to include production reticles. The results are being used to develop a comprehensive blank inspection protocol in a commercial mask production facility. We report on all phases of the project including the trial program.
In this study, we used production reticles and a new programmed defect test mask to characterize the sensitivity and false defect performance of several algorithms (ALM100, ALM200 and ALM300). ALM300's sensitivity spec is 60 nm. The inspection results were transferred to a CD SEM for analysis. SEM measurements were taken to validate the sensitivity of the algorithm and to quantify the calibration accuracy of the review tools of the inspection system.
As semiconductor lithography nodes become increasingly difficult to achieve with traditional optical lithography, several new technologies have emerged. SCALPEL (SCattering with Angular Limitation Electron beam Lithography) is at the forefront of the NGL technologies. SCALPEL technology uses an electron beam rather than laser light to produce images on the wafer. The SCALPEL mask is non-traditional in the sense that it is silicon-based instead of glass-based and the patterns are written on a membrane. SCALPEL provides unique challenges for the mask maker as well as the semiconductor manufacturer. In this study, we have demonstrated that the KLA-Tencor 3XX platform is capable of inspecting prototype SCALPEL reticles for pattern defects. The inspections were performed with two light wavelengths: 488 nm and 365 nm. Included are the difficulties faced and a projected roadmap for the inspection tool when SCALPEL enters at the 100 nm technology node.
Stringent specifications require that reticle makers carefully examine the role blanks play in reticle quality. Photronics and KLA-Tencor are jointly examining several aspects of this issue. As part of this investigation, PBS blank quality was tested in a production environment using the KLA-Tencor STARlight inspection system. PBS blanks were inspected using a 500-nm pixel with the highest sensitivity settings. We completed a comprehensive study using an effective blank defect test pattern. The test pattern was chosen to maximize the probability that a blank defect will fall on a chrome-to- quartz transition. Several test reticles were inspected and reviewed before writing, and reviewed a second time after processing. 452 defects were classified using three variables: blank defect size, blank defect type and reticle defect type. Some blank defect sizes and types transferred to the test reticles with probabilities exceeding 80%. False defect rates were less than 0.5%. Defect statistics for two blank suppliers are presented. We outline the phases of the research, present the results and discuss the implications for production reticles. We demonstrate techniques that can be used before writing and processing to assess the probability of defect transfer. Plans for a trial protocol for blank inspection are presented.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.