Performing Optical Proximity Correction (OPC) in ATHENA

Introduction

One of the main goals in lithography processes is to transfer the pattern on the photomask onto the wafer surface as accurately as possible. However, due to the diffraction from the edges and the interference between neighboring mask features, the mask pattern and its printed image are bound to be different. As shown in Figure 1, corner rounding, line shortening and non-uniform linewidth are few of the proximity effects that have been observed. For a binary mask, making small modifications on the mask pattern by adding serifs to the corners and by moving the pattern boundaries are the basic vehicles in counteracting these proximity effects. In conjunction with MaskViews and TonyPlot, Optolith has been extended to perform optical proximity correction (OPC). This new OPC capability will enable the process engineer to examine various aspects of the proximity effect on the lithography process window. Details of these enhanced tools are discussed in the following sections.

The OPC Tools

The procedure to performing OPC is similar to simulating the aerial image of a given mask pattern. Use MaskViews to draw or to import the layout file, and then use Optolith to calculate the aerial image. Visualization of the image profile is done in TonyPlot. If the deviation between the image and the pattern is not acceptable, the layout pattern can be modified and these steps repeated. To achieve the final corrections on the mask pattern, several iterations may be required. In order to expedite the process, special features have been incorporated into the above mentioned interactive tools.

MaskViews Enhancements

As a visual guide for the user in making correction to the mask layout, MaskViews can now import the aerial image output file from Optolith and then overlay it onto the mask pattern [Files-Import-Optolith format], see Figure 1. Furthermore, a new serif object has been created [Define-Objects-Object] to let users to adjust the size of the serif. Adding user-defined serifs to the corners of a mask feature is done just at the push of a button [Edit-Add serifs].

Figure 1. The image from a mask does not match the
intended layout. Importing the aerial image into
MaskViews acts as a guide for OPC.

 

Optolith Enhancements

In designing this OPC generator, we have assumed a threshold model in the response of the photoresist to the energy dosage, which means the chemical characteristics of the positive photoresist will change above a user-defined threshold level. The normalized intensity threshold level is specified by the opc-option as in the following statement,

structure outfile=anopex25_2.str / infile=anopex25.sec opc=0.5

This syntax also signifies ATHENA to search for an layout section file anopex25.sec, generated by MaskViews for the unbiased original mask layout, and to include the information of the unbiased mask into the output structure file for later use in TonyPlot. This unbiased mask pattern is required for calculating the percentage of the area of the mask pattern that is below the threshold level (AB), and the percentage of the image area that is lying outside of the mask pattern (AO). These are defined as,

 

and

 

It is our belief that these two figures of merit are much easier to calibrate than the distribution of distance deviation. For the unbiased mask shown in Figure 1, AB and AO are equal to 32.4% and 0.7%, respectively. For the corrected mask displayed in Figure 2, AB and AO become 12.0% and 3.7%, respectively. There is approximately 20% improvement in filling up the mask pattern. A cross-section of the intensity profile, Figure 3, reveals that the image of the corrected mask not only conforms better to the ideal case, but also has a better contrast and steeper slope. These are very good indications that OPC will be able to enlarge the existing process window without using the phase-shifting mask technology or demanding for more exotic optical system and light source. The potential advantages for the semiconductor industry are tremendous extending the lifetime of the existing process equipment and knowledge base, and reducing the pressure on new process technology development, to name a few.

Figure 2. Outline of the unbiased mask pattern overlayed
with the layout and image of the biased mask.


Figure 3. Image intensity profiles for
the biased and unbiased mask.

 

Besides calculating the two figures-of-merit, an image output file is automatically generated for MaskViews to overlay with the mask layout as mentioned above. The filename of the image file has the following syntax: x_img.sec, where x is the filename specified in the outfile-option, for example, anopex25_2_img.sec.

TonyPlot Enhancements

To give a visual inspection on the integrity of the image of the biased mask pattern, the outline of the unbiased mask layout can be drawn on top of the usual contour plot of the aerial image intensity profile.

 

Additional Optolith Improvements

In addition to the OPC capability, Optolith has been upgraded in two other aspects. Firstly, the speed in calculating the aerial image has been greatly increased without sacrificing the accuracy. This improved algorithm is slightly better than the n-log-n type technique. Secondly, the accuracy of the ray tracing technique employed in the photoresist exposure modeling has been enhanced. This technique is especially applicable to evaluate the latent image formation in the photoresist over a realistic nonplanar device geometry without rendering into other time consuming algorithms, for example, the time domain finite difference method [1]. The reflective notching effect from the sidewalls of the trench is clearly shown in the developed resist profile depicted in Figure 4.

 

Figure 4. Note the reflective notching effect at the lower
half of the photoresist profile inside the trench.

 

Reference

  1. R. Guerrieri, K. H. Tadros, J. Gamelin, and A. R. Neureuther, IEEE Trans. Computer-Aided Design, 10, 1091-1100 (1991).