Historically modeling and simulation has started from stand-alone programs for unit process steps, namely, oxidation, lithography, ion implantation, diffusion, etching, deposition, and metalization. The purpose of these simulators is to identify the decisive effects of the single process and to quantify the impact and the cross-dependences of the process parameters. Beyond this, simulations are useful for providing insight into technology directions at early stages of the development, when the technological requirements for a new node are not yet met.
In a further step these point solution simulators were combined to an integrated process simulation flow which is aimed to follow as closely as possible the process recipes applied in manufacturing. The final result of such a simulation flow is a device representation which in the ideal case exactly mirrors the fabricated chip. Not only serves this device representation as input for device simulators as well as for resistance, capacitance and thermal extractions for the interconnects, the final simulation result also gives insight in device properties which sometimes are not or only very difficultly accessible for measurements as in the case of doping distributions and interconnect capacitances.
The latest developments combine a great variety of tools to an integrated simulation framework, summarized under the term Technology Computer-Aided Design (TCAD). TCAD provides the complete path from the mental image of the reality, formalized in mathematical models and implemented with simulators, to calibration and validation by comparison with relevant experimental data, carried out in order to determine numerical values for parameters and to demonstrate the suitability of the approach. It comprises equipment modeling including the physical environment, conditions and processes affecting the wafer, feature scale modeling with front end processes, lithography, and topography modeling including structural, mechanical, and thermal aspects, physical device modeling for physically based models for active devices and interconnects, circuit element modeling including compact models for active, passive, and parasitic circuit components as well as parameter extraction, package modeling for electrical, mechanical, and thermal characterization, simulation environments providing graphical user interfaces, tool integration, statistical modeling, as well as coupling to experimental and fab data by means of calibration, optimization, design of experiments (DOE) and response surface models (RSM), and numerical methods like grid generators, matrix solvers, parallel algorithms, and surface-advancement techniques, to note only the most important sections without claiming to be complete.
Topography simulation, preferably coupled with lithography tools and directly connected with the layout, plays the most important role in linking TCAD to the design and the layout of ICs, realized with Electronic Computer-Aided Design (ECAD) [41]. ECAD provides the tools necessary for the generation of a physical representation of the circuit diagram which is a symbolic description which cells are used and how they are connected. It guarantees the complex functionality of the circuit, required by the operational specifications of the customer.
In order to reduce the complexity of this transformation procedure, ECAD makes
use of hardware description languages (HDL) which combine the design
rules of the single components provided by the selected technology to
operational blocks and design libraries. Up to now, these design rules on
their part have relied on circuit models describing the behavior of single
devices, extracted from measured 
-
curves of single transistors and delay
times of ring-oscillators. Most recently the development in this area is
directed towards the virtual wafer fab. The vision is to directly
include the layout into a complete process simulation followed by device and
interconnect simulation for extracting the compact circuit models for devices
and interconnects. These compact models are in turn fed back into the chip
design rules leading to a high level of optimization long before the IC is
casted into the first silicon wafer.
Curios enough and worth mentioning within this context is, that improved IC fabrication technology directly influences the hardware available for the simulation environments, allowing more detailed analysis of all affiliated processes, leading to a better understanding of device physics and better IC technology, on their part supplying faster processors for even more detailed investigations...
To get more serious again, the justification for the investments in such sophisticated, highly cross-connected, and complex simulation environments has to be queried. Not unexpectedly the answer is costs. Computational power has increased drastically going along with decreasing costs for desktop personal computers (PCs), workstations, super-computers, as well as networking hardware and is heavily contrary to the exploding costs for semiconductor manufacturing equipment, especially in the experimental, pre-fabrication stage. Moreover experiments are very time-consuming, and as we all know, time is money. By its mathematical description, simulation provides a very fast way for testing new approaches and for playing with concepts and ideas long before wasting expensive silicon. And simulation is non-polluting.
On the other hand it would be wrong to exclusively see costs as the reason for the emerging of TCAD. It is clear that there is also a massive scientific interest in both process and device physics as well as in concepts for modeling, simulation, and information technology (IT) itself. Investigation and simulation of single electron devices, Monte Carlo simulations of particle-particle interactions which describe transport phenomena of sputtered particles, implanted doping atoms and carriers, as well as quantum chemical investigations of diffusion, molecule-surface interactions, and gas phase chemistries are only a few of the many fields of scientific research and are broadened on the IT side by model description languages, prototyping systems, and parallelization schemes. Knowledge from previous work on a more phenomenological level or from work carried out under different modeling assumptions at a different physical level directly flows into new theories on the one hand and application software for day-to-day use on the other hand.
As an illustrative example of the effects which have gained significance for modeling and simulation, investigations on device physics have reached atomic level, provoked to some part by the shrinking devices. Undoubtedly continuum models are still valid for state-of-the-art devices with conventional gate lengths. But when the channel becomes that short and the doping concentration so low, that only about 10 dopant atoms are located in the current path under the gate, continuum models become at least questionable.
Prev: 1.1 Process Technology
Up: 1. Introduction
Next: 1.3 Topography Simulation