Computer-simulated resist profiles

Figure 3.17. Computer-simulated resist profiles (SAMPLE). Operating input parameters include matched substrate, AZ1350J resist, 4358 A, 90 mjlcm, NA = 0.35, a = 9.99, defocus 0.0, development 80 s. The open image (B = 0.058) simulates AZ1350J performance. The shallow profile (B = 1.96) was generated from identical input parameters with the exception that the un-bleachahle absorbance (B) was adjusted to the value corresponding to the absorbance of 1 xm of novolac at 254 nm. (Reproduced with permission from reference 37. Copyright 1981 Institute of Electrical and Electronics Engineers.]...

The authors gratefully acknowledge helpful discussions with D. Hofer particularly regarding the computer simulation of resist profiles. J. Michl and J. Downing also acknowledge partial financial support from the NSF (Grant No. CHE 81-21122). 

The most important application of SCM is two-dimensional dopant concentration profiling in semiconductors. However, the raw capacitance-voltage data obtained from SCM must be converted by a mathematical model into a dopant concentration. Therefore, development and validation of appropriate models represents a large part of SCM methodology. To validate the various models, other experimental techniques must be employed to measure and verify the dopant profiles independently. SCM data are usually compared to secondary ion mass spectrometry (SIMS) measurements of dopant concentrations. Spreading resistance profiling (SRP) and computer simulations are also employed to check model validity. 

There is a striking similarity between the experimentally observed and the theoretically calculated profiles, and all four characteristic features occur in both. The calculated location of the minimum, which mainly depends on the vacancy mobility, is close to the location observed in the experiment. The computed temperature dependence of the depth of the minimum corresponds with the results of the measurement. Obviously, the stoichiometry polarization model of resistance degradation correctly predicts the conductivity variations. In particular the almost quantitative agreement of the very characteristic shape of the conductivity distribution proves the validity of the existing model described above. It should be noted that in the calculations only the hole mobility is chosen such that the theoretically and the experimentally observed depth of the minimum is similar, but all other parameters used in the simulation are taken from literature [77, 336, 338],... 

---