The models and algorithms used for simulation of chemical vapor deposition (CVD) profiles in Stanford profile emulator for etching and deposition in integrated circuit engineering are described. In the CVD simulation direct deposition, re-emission, and surface diffusion are considered as the mechanisms for near surface mass transport. The re-emission process is characterized by a single surface reaction coefficient (or sticking coefficient) which condenses the complex physico-chemical mechanisms (physisorption, chemisorption, desorption) in a single probability that defines the final attachment of a reactive to a surface. For desorption process, different desorption models can be selected (cosine re-emission, specular reflection). Surface diffusion is modeled with a gaussian distribution function characterized by the diffusion length. A Monte Carlo approach is used to determine the deposition rates in arbitrary topologies and a string algorithm is used for the evolution of the surface. Iteration between them allows taking into account self-shadowing effects. Comparisons of experimental versus simulation results for SiO2 trench filling, that can be characterized by direct deposition and re-emission with a pure cosine desorption law and no surface diffusion are included.

A new, physical based 3-D profile simulator has been developed that includes the dominant effect of re-emission. This simulator is part of the Stanford Profile Emulator for Etching and Deposition in ! C -Engineering (SPEEDIE). Unlike previous simulators which consider only the arrival of deposition precursors by unshadowed direct transport and by surface diffusion, SPEEDIE also considers transport into shadowed areas by adsorption and re-emission. The importance of re-emission was established by using overhang test structures to separate the roles of surface diffusion and re-emission For the depositions investigated (SiO,, poly-Si and W) it was found that re-emission dominates over surface diffusion in controlling surface contours. Using the simulator to fit experimental LPCVD SiO, profiles, it was found that a single constant sticking coefficient (SJ model with a cosine re-emission distribution gave excellent fits independent of geometry for a given deposition condition. Both Monte Carlo and analytic methods are used to calculate the precursor flux along the growing surface.

No abstract

A great deal of photolithographic activity in recent years has been centered on thick photoresist films. Thin film heads (TFH), micromachining and sensor fabrication are examples of applications requiring this type of processing. The needs of the TFH industry are currently the technology driver for thick photoresist processing. Modem TFH manufacturing processes require 1 im resolution in layers ranging in thickness from 5 to as much as 25 tim. These large aspect ratios not only make the lithographic process difficult, but add complexity to the evaluation and measurement of experimental wafers. This is particularly true for the large number of measurements needed for process optimization and control. Well-calibrated and easy to use modeling techniques for analysis of the impact of optical system design and photoresist process changes would be extremely valuable for process lithography engineers.The photoresist development process involves complex dissolution and polymer chemistry. It forces simulator developers to implement empirical models with definitions and assumptions that only indirectly reflect the underlying physical and chemical processes. However, with appropriate calibration such an approach provides results with accuracy better than 90% at reasonable computational time for any given combination of a particular photoresist base material, photoactive component, development and bake conditions. A method has been developed that allows accurate simulation of pattern profiles in photoresist in excess of 10 xm thick. The method uses the DEPICT® photolithography simulator to model i-line exposure, bake and development of Shipley SJR®5740 thick film photoresists with an Ultratech 2244i Wafer Stepper®.Kim model inputs were estimated from a family of development rate curves obtained by processing wafers with a range of expose energies for logarithmically increasing develop times and measuring thickness change as the develop process occurred. These results were compared with dissolution results obtained using a laser-based dissolution rate monitor. Uncertainties in the measured photoresist absorbence, photosensitivity and refractive index coefficients were estimated and their influence on the simulated results were considered. An optimization 0227-786X/97/$ 10.00 SPIE Vol. 3049 / 789 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 06/26/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

We discuss computational techniques for calculating aerial image intensity distributions from large GDS II files recently implemented in Depict, a photolithography simulator for projection imaging, resist exposure, post-exposure bake and development. In particular, an algorithm for rapid and accurate evaluation of the mask Fourier transform over large domains containing non-uniformly positioned mask elements is implemented. By controlling aliasing errors within the context of a multiple level scheme, this algorithm renders feasible the simulation of aerial images across large portions of integrated circuits. The algorithm also allows overlapping phase mask elements obeying multiplicative transmission rules, and mask element merging. Accuracy for integration of the extended light source is also reported.