The CNTech Advanced Lithography Toolset uses a beam propagation method to calculate the intensity profile as it propagates through the mask and into the photoresist. One can construct the membrane, absorber, gap, and resist, each as a series of n-slices to achieve unusually precise calculations. Here a clear X-ray phase mask is modeled with silicon nitride in a configuration called a Bright Peak Enhanced X-ray Phase Mask (BPEXPM).
For the optimized structure of this mask, which relies on both diffraction and phase shifting to produce the reduced wafer image, four factors must be controlled; these are: absorber thickness - material and wavelength dependent, absorber wall slope, gap, and resist threshold. A central composite experimental design showed that a 100 nm mask would print a wafer at 35 nm CD using the 70% maximum intensity threshold when the wall slope was 0.5° from the vertical. Additionally: 1) a 100 nm increase in absorber thickness decreased the CD by 1.0 nm; 2) every 1.0 um increase in gap decreased the CD 0.8 nm; and 3) every 1.0 nm increase in mask CD increased the linewidth only 0.1 nm. Other mask processing materials were examined in addition to the 180° (π) phase-shift absorber thickness. Experimental verifications of the modeling results are in progress to demonstrate device construction for devices with lower wafer coverage than would be required for memory devices.
We report some new results in the use of high energy radiation in proximity x-ray lithography for the 50 nm node. The higher energy of the incoming radiation, 2.6-2.7 KeV, has two primary benefits: (1) it reduces the diffraction and allows printing of higher resolution features, and (2) it increases the effective depth of exposure allowing larger gap setting; however, the absorption of these photons creates hot electrons, which redistributes the energy in the resist, thus creating a uniform blur that limits the resolution by reducing contrast. The impact of the energy redistribution needs to be investigated when increasing the energy of the radiation, and in considering the materials used in both the optics and the mask and resist combination.
Phase-shifting masks have been applied in optical lithography and various phase-shifting techniques in X-ray Lithography (XRL) have been demonstrated. In this study, we compare different phase-shifting technologies for XRL, such as clear phase mask, attenuated phase-shifting mask and alternating aperture phase-shifting mask through computer simulation. The control of critical dimension is of primary importance as the CD shrinks to the sub-100 nm region. We have reported our design and fabrication of a more robust X-ray Phase Mask, which is capable of sub-70 nm imaging. The processing latitude of this design is investigated in terms of the X-ray source broadening, phase-shifter uniformity, mask-to-wafer gap and sidewall slope of the phase-shifter. The latitude is compared with those results from an attenuated phase shifting mask and an alternating aperture phase shifting mask.
Availability of production-worthy x-ray masks is of great concern to the lithographic community in anticipation of insertion of x-ray lithography as the leading contender among the next generation lithographies.
In semiconductor processing, the modeling of the process is becoming more and more important. While the ultimate goal is that of developing a set of tools for designing a complete process (Technology CAD), it is also necessary to have modules to simulate the various technologies and, in particular, to optimize specific steps. This need is particularly acute in lithography, where the continuous decrease in CD forces the technologies to operate near their limits. In the development of a 'model' for a physical process, we face several levels of challenges. First, it is necessary to develop a 'physical model,' i.e. a rational description of the process itself on the basis of know physical laws. Second, we need an 'algorithmic model' to represent in a virtual environment the behavior of the 'physical model.' After a 'complete' model has been developed and verified, it becomes possible to do performance analysis. In many cases the input parameters are poorly known or not accessible directly to experiment. It would be extremely useful to obtain the values of these 'hidden' parameters from experimental results by comparing model to data. This is particularly severe, because the complexity and costs associated with semiconductor processing make a simple 'trial-and-error' approach infeasible and cost- inefficient. Even when computer models of the process already exists, obtaining data through simulations may be time consuming. Neural networks (NN) are powerful computational tools to predict the behavior of a system from an existing data set. They are able to adaptively 'learn' input/output mappings and to act as universal function approximators. In this paper we use artificial neural networks to build a mapping from the input parameters of the process to output parameters which are indicative of the performance of the process. Once the NN has been 'trained,' it is also possible to observe the process 'in reverse,' and to extract the values of the inputs which yield outputs with desired characteristics. Using this method, we can extract optimum values for the parameters and determine the process latitude very quickly.
It is well known that a point source produces a small magnification of the mask pattern on the wafer due to divergence; the effect becomes especially noticeable at the corners and edges of the field. In this article we model the image formation due to a point source and compare the deviations of the aerial images and developed patterns, from those at the center of the field. We illustrate the behavior with several computational results.
In this paper we report the results of simulations and experiments on application of phase-shifting mask to x-ray lithography (XPM). We have built an XPM with PMMA as a phase shifter; we printed patterns using the XPM and characterized the linewidth dependence on gap and dose. Small lines around 70 nm were printed at 25 micrometer gap. The resist lines have good uniformity, and aspect ratio as high as 4. The characterization experiment was performed on the Karl Suss X- ray Stepper installed at the Center for X-ray Lithography. The results show that the exposure-gap latitude window is obviously increased compared with traditional x-ray mask. XLITH has been used to analyze the aerial image under the XPM and predict linewidth. Different phase shift materials, such as PMMA, Si, SiC and Al have been simulated and characterized with respect to blur, dose and gap. The model predicts a mask contrast 6 - 7, and resolution of 50 nm using gaps around 5 micrometer. XPM is a path to nanostructures with more reasonable process latitudes. We printed sub-0.1 micrometer dots using XPM. The XPM can also be used for gate-level lithography of ultrafast MOS devices to share its high resolution and large process latitude. It is possible to form both of a fine gate line and a contact pad at the same time with the combination of phase and transmission in a single mask. An experimental demonstration of printing unclosed lines is given in this paper. The further combination of phase-edge effects and phase-shifter size can also make it possible to form a wide range of patterns, ranging from line-space to isolated bright or dark areas with varied sizes, in a single exposure.
The printability of defects in x-ray masks was simulated in three dimensions using the CXrL toolset software developed at the University of Wisconsin and resist dissolution software developed in a collaboration between University of California at Berkeley and Motorola. Isolated defects on mask membranes and isolated defects on pellicle membranes mounted behind the mask membrane were modeled. Defects close to x-ray absorber features and absorber fabrication defects were also considered. Spheres and parallelepiped defect shapes composed of PMMA, ammonium sulfate, and stainless steel were modeled at exposure gaps in the range 10 - 50 micrometers. Attenuation of a variety of potential defect materials was calculated for the IBM Advanced Lithography Facility Helios synchrotron source and beam-line x-ray spectrum. The dose-to-clear for 400 and 500 nm thickness APEX-E films was then used to predict what thickness of defect material would result in a printed defect. Image formation model predictions of defect printability in APEX-E resist were compared to attenuation calculations, indicating that defect shape and x-ray phase shift in the defect material has a profound impact on defect printability for materials that are not highly attenuating. Spheres printed more readily than parallelepipeds. Increasing the exposure gap reduced printability slightly. Experiments to determine the printability of organic spheres added to x-ray masks were compared to simulation to verify its accuracy. Based on modeling results, the minimum size of isolated defects on x- ray masks that printed are presented. The minimum size of defects that changed printed line-width were also discussed. Based on these results, defect inspection sensitivity, cleaning capability, and repair resolution for less than or equal to 175 nm line-width x-ray masks can be established.
The problem of accurately describing the topography of a mask arose in x-ray lithography, when the absorber structures have dimensions of several hundred wavelengths and may have topography. However, the application of phase shifting structures and a refinement of the image formation calculation even in optical masks has recently raised the same issue for the case of optical lithography. This paper reports the design and implementation of Topography Description Model (TDM), a rule-based 3D topography generator that allows the user to describe the exposure system, such as a system comprising of mask, gap, resist and wafer. The paper illustrates some applications of TDM to real-world studies, such as studying the effect of absorber side-wall slope on the final image and optimizing XRL masks.
The optimization of the lithographic process requires the definition of a figure of merit to gauge the quality of the image. Several ad-hoc approaches have been used in the past, but there is not yet a method which is clearly the most effective. Here we introduce the concept of energy error map, obtained by comparing the aerial image with an ideal target pattern. One is target aerial image which we want to define on the wafer, the other is aerial image which is calculated at each process conditions. Given a target pattern B, in the form of a matrix, and an aerial intensity map A, we define the energy error map as (A - A (DOT) B) plus (A - A (DOT) B). We also used the total sum of error map between target aerial image and calculated aerial image as figure of merit in quantifying the quality of images. With the calculated energy error map, we also proposed first-order proximity-corrected mask pattern which is based on the distribution of the energy error in error map.
The actinic spectra of two beamlines of the University of Wisconsin's Center for X-ray Lithography (CXrL) at the Aladdin storage ring were studied in three configurations. Some beamlines optimized for particular bandwidths are presented and their impact on mask making, aerial image quality, printed image quality, and device damage discussed. Resist performance dependence on the actinic spectrum is investigated. The exposure-gap tree response of 0.25 micron features is presented for different spectra. Resist characteristic curve data were collected for these conditions and are compared.
This paper presents the results of a simple orthogonal matrix experiment testing photoresist performance as a function of post exposure bake temperature and time. The dose latitude of quarter micron line/space pairs is found under these conditions. These empirical results are compared against those produced under identical process conditions but utilizing simulated images based on resist dissolution rate data. The matrix responses of the empirical and simulated data sets are compared. Also, these linewidth results are compared against resist characteristic data produced under identical process conditions. The matrix responses of the three data sets are compared.
We report on a new modeling tool for the prediction of the cumulative dose delivered to any given layer of a semiconductor process involving x-ray lithography (XRL) steps. In such a process, the layers receive various doses at different steps in the manufacturing. In order to determine the total dose delivered at any given level it is necessary not only to compute the dose for the lithographic step of that particular level, but also to keep track of the past history of the process in order to accumulate the total radiation. For this, one must know the full process, including sacrificial layers, resist layers and layout geometries, since different areas of the surface being patterned will receive different doses because of shielding by other layers. We apply the model to the case of a standard 0.5 micrometers NMOS device process and to an advanced 0.1 micrometers NMOS process. Results will be presented and discussed in detail, for the various assumptions of the lithographic process. The program, written in C under UNIX, is compatible with standard process modeling tools and device layouts.
KEYWORDS: Systems modeling, Data modeling, Data conversion, X-ray lithography, Human-machine interfaces, X-rays, Manufacturing, Visualization, Data processing, Process modeling
EXCON is a software package developed to control and define the modeling of experiments. In this case, EXCON is used to integrate various individual modeling systems, in a user- definable manner, to simulate the semiconductor manufacturing process. These numerous systems model specific aspects of the integrated circuit fabrication process. Each can be a large complex software program requiring many system resources to reliably emulate the physical processes, in many cases at the atomic level, in an analytical manner. There are many different program data formats and user interfaces within the modeling systems used. EXCON addresses the automatic insertion of configuration and process data, the conversion of data formats between modeling systems, and the sequence of model execution. EXCON also has a mechanism to re-run the sequence of models with variations in one or several configuration or data parameters, thereby creating an environment to do controlled experiments. EXCON assists in the visualization of the data in the experimental data sequence. EXCON allows for coarse-grained parallelism by connecting processes with an interprocess and inter-machine communication mechanism, thereby allowing for concurrent execution of processes on multiple machines. System performance enhancement is done with a incremental directed graph analysis technique. EXCON will, when appropriate, transparently convert file formats between modeling systems.
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.