Developing a new lithographic process routinely involves usage of lithographic toolsets and much engineering time to perform data analysis. Process transfers between fabs occur quite often. One of the key assumptions made is that lithographic settings are equivalent from one fab to another and that the transfer is fluid. In some cases, that is far from the truth. Differences in tools can change the proximity effect seen in low k1 imaging processes. If you use model based optical proximity correction (MBOPC), then a model built in one fab will not work under the same conditions at another fab. This results in many wafers being patterned to try and match a baseline response. Even if matching is achieved, there is no guarantee that optimal lithographic responses are met. In this paper, we discuss the approach used to transfer and develop new lithographic processes and define MBOPC builds for the new lithographic process in Fab B which was transferred from a similar lithographic process in Fab A. By using PROLITHTM simulations to match OPC models for each level, minimal downtime in wafer processing was observed. Source Mask Optimization (SMO) was also used to optimize lithographic processes using novel inverse lithography techniques (ILT) to simultaneously optimize mask bias, depth of focus (DOF), exposure latitude (EL) and mask error enhancement factor (MEEF) for critical designs for each level.
A new method to calculate Mask Error Enhancement Function, or MEEF, from the intensity slope of the unperturbed
geometry and intensity offset of the perturbed mask is derived. In the limit of small perturbations, the intensity slope
technique is predicted to be the same as MEEF values calculated from the ratio of wafer to mask CD differences scaled
by the magnification. Full chip process window simulations were done to compare the accuracy of this new approach
for 45 to 90nm mask designs for line, space and contact features. The standard deviation was less than 0.11 and the
largest deviation was only 12% for over 5200 MEEF calculations. Below MEEF values of 20, the standard deviation
was less 0.065 and all simulations were within ±0.5.
A significant discovery in this work is the inverse relationship between image intensity slope rather than NILS or ILS at
the location of the printed feature edge and MEEF. Since the image slope decreases closer to the intensity extrema, high
MEEF regions are predicted to be those that print closest to the minimum and maximum intensities.
New methods for visualizing process window effects on simulated lithography violations are shown. Three types of
analysis of simulation errors are discussed. Worst site violations are those geometries in which at least one process
condition shows largest deviations from target. For these errors, variations of Cleveland dot charts are useful for
showing key attributes such pinpointing which process condition(s) cause the largest violations and the distribution of
violations among focus and exposure conditions. Modifications of dot charts are also useful to visualize violations
across the process window for the entire chip as opposed to selected sites. Lastly, linearity charts combined with
box/whisker objects can be used to show deviations from target over a range of drawn dimensions.
An integrated methodology for developing recipes for optical proximity correction (OPC) is demonstrated. A complete implementation of software programs for generating the OPC corrections, determining mask and layout errors and automatically displaying contours of the worst violations has been accomplished. Integration of these elements facilitates recipe development by quantifying the effect of recipe changes on the overall critical dimension (CD) control. In this paper, a 65nm alternating aperture phase shift test mask is used for illustration of the method. The concept of a recipe comparison matrix is introduced to quantify the effect of recipe changes on across-chip metrics.
For leading edge technologies, mask critical dimension (CD) errors consume a substantial part of the total wafer CD budget. Moreover, the strong optical proximity effects (OPE) can make the impact of a CD error on the mask significantly worse on wafer. At the same time, the mask making capabilities as far as CD control can barely keep up with the wafer fab requirements. To assess the overall mask quality ever more mask CD measurements are taken in the mask qualification process. These measurement points are increasingly placed in the main die area and are often selected in a more or less random fashion.
An improved assessment of the mask CD quality can be achieved by taking advantage of the lithography verification step. The wafer simulation capability in the Silicon versus Layout (SiVL) tool is used to identify the high mask error enhancement factor (MEEF), error prone locations on a critical layer. The mask CD qualification process can be improved by including these poor MEEF and error prone sites. In this work, an automated flow is presented in which mask qualification sites are selected based on simulated wafer image contrast.
A new general method to predict resist maxima and minima of linewidth swing curves using UV reflectance spectra near the actinic wavelength is proposed. In this technique, the reflectance spectra in the vicinity of the exposure wavelength are used to determine the effective phase shift due to the underlying substrate. With this
parameter, resist thicknesses yielding minima and maxima linewidths can be predicted. In principal, resist minima and maxima can be determined from the UV spectra of a single wafer. This technique was used to predict minima and maxima on both simple (bare silicon) and multifilm (polybuffer LOCOS and gate) stacks for 2 different resists, with and without topography. Three of the film stacks included a top antireflecting coating. Results were compared with swing curves determined from measured linewidths versus resist thicknesses. In general, predicted resist maxima and minima were within 50 Angstroms of the corresponding values measured from linewidths which was well within the experimental error. Additionally, the peak height of the reflectance spectra in the vicinity of the actinic wavelength was generally correlated to linewidth swing determined from multiple wafers. This implies that process stability with respect to cd swing can be compared directly from reflectance spectra. Theoretically, results determined from a nearly normal incident UV reflectometer require a correction to account for the finite NA of the lens. Simulation was used to determine this correction. However, two
different simulation programs (Prolith 6.0 and Solid C 6.3) overcorrected the case of an i-line stepper with 0.63NA and 0.65 partial coherence. In this case, agreement with experiment was better with a "half" correction determined from simulation.
A new algorithm for correcting misalignment between layers is introduced which is capable of compensating for interdependencies and arbitrary conventions of correctable factors. In this approach, optimal corrections are determined from solving a set of linear equations that exactly negate the effect of the observed misalignment. A series of calibration runs were performed by measuring the effect of a balanced set of forced input corrections on the resulting alignment in order to determine elements of a transformation matrix. This technique was able to calculate the average corrections required to reverse the input offsets within an average of 10nm for translation and 0.1ppm for magnification and rotation offsets. Estimated standard deviations between calculated and input offsets were smaller for y parameters than x, presumably because of better stage reproducibility in the vertical direction. The transformation matrix for Canon EX4 steppers highlighted that chip rotation is affected by inputting corrections to x axis wafer rotation. Calibration wafers with alignment sites that simulated wafer rotations of +/- 0.8585ppm were fabricated. These wafers verified that chip rotation occurs for EX4 steppers as a consequence of automated adjustment for x wafer rotation. The observed responses from these calibration wafers agreed within nanometers of the relevant element of the transformation matrix for that stepper family.
Optical end of line metrology, OELM, is a new method to measure relative line shortening effects using conventional optical overlay instruments. In this technique, a frame which has two adjacent sides that are constructed of lines and spaces is imaged onto a wafer. Since sub 0.5 micrometers gratings cannot be resolved using conventional optics,the alignment tool sees the sides compared of lines and spaces as solid edges. The purpose of this paper is to characterize errors implicit with this approach. First we show a general error analysis for determining best focus using OELM measurements. From this, we introduce the concept of local image quality as the inverse of the minimum lien shortening, and curvature of line shortening with focus.
Complementary alignment metrology (CALM) is a new metrology technique to visually measure stepper alignment correctable factors such as horizontal, vertical and rotation offsets as well as magnification errors. CALM is based on the concept that a line and space pattern exposed into resist will e completely cleared if, prior to development, it is exposed a second time by shifting the grating by exactly its half- pitch. We have used this principle to fabricate test wafers that visually indicate correctable factors. The estimated 3(sigma) accuracy of CALM readings compared to box-in-box measurements is 0.03 micrometers . Linearity between CALM readings and box-in-box measurements is maintained for misalignments of +/- 0.13 micrometers . Using such a technique allows baseline corrections to be performed on a more frequent basis.
We have investigated the effect of resist thickness, linewidth and pitch on UV reflectance spectra. This technique exploits the property that conventional novolak resists are very absorptive from 200 - 300 nm while substrates are significantly more reflective. For line/space developed resist features of constant pitch, we observe that the reflectance in this wavelength range varies periodically with (lambda) and increases linearly with decreasing linesize. The dominant factor in wavelength dependence is the constructive/destructive interference of the measurement light from the air/resist and air/substrate interfaces. Linesize dependence at constant pitch and resist thickness is predominantly controlled (within proper boundary condition regimes) by the percentage of the substrate exposed. The gross periodicity of the DUV reflection spectra for patterned films is correlated with resist thickness in a manner similar to the resist thickness dependence of UV reflection spectra for unpatterned films. Simulation of DUV reflectance from patterned films showed semiquantitative agreement with experimental results.
A generalized approach to modeling resist performance is introduced and applied toward characterizing a negative chemically amplified resist system. The Generalized Characteristic Model for lithography is used to extract parameters to evaluate easily development rates from characteristic curves. The model suggests that two lumped parameters, αn and E0, dominate lithographic response for negative chemically amplified resists. Both αn and E0 were regressed from characteristic curves over a postexposure bake temperature and time range from 110 to 150°C and 30 to 90 s and develop times from 30 to 150 s. The parameter E0 showed the predicted postexposure bake temperature and time and develop time dependencies over the processing window, while αn did not. Possible explanations for this discrepancy are discussed. These parameters were used to simulate linewidths that were compared with experimental results. Linewidth predictions using parameters extracted with the generalized characteristic model agreed to within 15% of experimental results over the entire processing window.
This paper discusses issues concerning the use of a thin absorbing organic anti-reflective (AR) coating for 0.5 micrometers excimer laser lithography. Linewidth profiles, CD control, substrate dependence and etching are all affected by the presence of the AR coating. The advantages of implementing highly absorbing organic AR coatings are that CD control with resist thickness (swing curve) is improved, substrate reflectivity effects becomes negligible, adhesion is improved and reflective notching is decreased, leading to better CD control over topography. For example, the swing curve was reduced from 0.18 micrometers to 0.04 micrometers by using an absorbing AR film. Exposure/focus latitudes were modeled using Obelisk software. This gave an exposure dose of 26 mJ/cm2 for printing 0.5 micrometers dense lines on aluminum, poly- silicon, oxide, nitride and tungsten. In addition, the exposure latitudes for these substrates was found to be +/- 10% when the total depth of focus was 1.2 micrometers . Statistically designed experiments were used to optimize resist profiles of the SNR248 resist images on AR coating. Issues relating to implementing organic AR coatings under a 1300 angstrom absorbing AR coating reduced reflectivity variation on poly-silicon from 38% to 6% over a resist thickness range of 350 angstrom. By using the AR coating the swing curve variation was reduced from 0.18 micrometers to 0.04 micrometers .
A novel statistically designed tracking technique is described which utilizes multiple exposure systems and photoresists to decompose tool, material, and process contributions to variation in lithography. The application of this technique is described as a means of characterizing an overall lithography process. In one application, the primary source of variation in the photo process was attributed to resists and resist processes rather than related to exposure tool. Process variation in one resist system was tracked unambiguously to changes in developer normality. Finally, results from this work extended previous studies showing that open frame exposure measurements are sensitive measures of photoresist process variation for both conventional and chemically amplified resists.
A generalized approach towards modeling resist performance is introduced and applied towards characterizing a negative chemically amplified resist system. The Generalized Characteristic Model for Lithography is used to extract parameters to easily evaluate development rates from characteristic curves. The model suggests that two lumped parameters, (alpha) nn and E0, dominate lithographic response for negative chemically amplified resists. Both (alpha) nn and E0 were regressed from characteristic curves over a postexposure bake temperature and time range from 110-150 C and 30-90 s and develop times from 30-150 s. E0 showed the predicted postexposure bake temperature and time and develop time dependencies over the processing window while (alpha) nn did not. Possible explanations for this discrepancy are discussed. These parameters were used to simulate linewidths that were compared with experimental results. Linewidth predictions using the Generalized Characteristic Model agreed to within 15% of experimental results over the entire processing window.
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.