Sub-resolution Assist Features in Photolithography Process Simulation

Summary form only given. As modern photolithography feature sizes reduce, the use of sub-resolution assist features (SRAFs) to improve the manufacturing process window has become more prevalent. To support this trend in conjunction with Optical Proximity Correction (OPC), it is vital to precisely simulate the manufacturing process with SRAF placement. However, for computational reasons, it is necessary to model features with SRAFs placed using specially developed approximation algorithms. This need arises for three reasons. First, SRAFs are normally difficult to verify physically because of their small sizes which leads to a relatively large uncertainty on SRAF layout compared to the printed features. Second, the effective transmission of SRAFs may not be the same as the transmission of main features. Third, the thickness of the mask is comparable to the feature sizes of SRAFs, so three-dimensional (3D) mask effects may play an important role in the process simulation. To compensate for these effects, three algorithms for SRAFs are proposed. The experimental data for this study were acquired from an optical system with hyper NA of 1.2. The cQuad source has a wavelength of 193 nm and sigma_in of 0.80 and Sub-resolution assist features in photolithography process simulation Sub-resolution assist features in photolithography process simulation sigma_out of 0.96. The source polarization is chosen to be x-y sector polarization. The smallest line width of the test pattern is 44 nm with a pitch of 100 nm. The empirical process data were fit into a process model by minimizing the selected cost function derived from the root mean square (RMS) of the prediction errors. Figure 1 shows the results obtained by treating the features with SRAFs the same way as those without SRAFs. It is clearly seen in the plot that the residuals (CD errors) can be separated into two groups. For all the features without SRAFs (red), the best fitting residuals have a RMS error o- f 0.76 nm and range of 3.9 nm. However, for the features with SRAFs placed (black), there is a significant offset from the 0 target and the RMS of the residuals is 4.99 nm with a range 6.6 nm. To account for the effects mentioned above, three functions are added to SRAF models so the main feature fits remain unchanged, but SRAF feature fits improve. The three treatments are: a) edge bias is applied to all SRAFs edges; b) the attenuation of SRAFs is determined independent of the main features; c) a loading effect is added to SRAFs layer. Figure 2 shows the results after these functions for SRAFs features. Significant improvement on the fitting residuals with SRAFs is obtained, resulting in an RMS of 0.79 nm and range of 3.55 nm. The details on how the special treatments are added and the theoretical explanation on how they compensate those effects being missed in the main features will be discussed in the full paper.