TCAD stands for Technology Computer-Aided Design, and it refers to software that simulates how semiconductor devices like transistors are built and how they behave electrically. Instead of physically manufacturing a chip to test a new design idea, engineers use TCAD to model the entire process on a computer, from the chemical steps that create microscopic structures on silicon to the electrical currents that flow through the finished device. It saves enormous amounts of time and money in an industry where a single fabrication run can cost millions of dollars.
What TCAD Actually Does
TCAD covers two broad types of simulation that work together. The first is process simulation, which models how a chip is physically manufactured. Every step of chipmaking, from depositing thin films to etching patterns to implanting atoms that change silicon’s electrical properties, gets recreated mathematically. The output is a virtual 3D structure showing exactly where different materials and impurity concentrations end up inside the device.
The second type is device simulation, which takes that virtual structure and predicts how it will perform electrically. The software calculates how current flows, how the device switches on and off, how much it leaks when it’s supposed to be off, and how it responds to changes in voltage or temperature. Together, these two simulation layers let engineers go from “here’s how we’d build it” to “here’s how it would work” without ever entering a fabrication facility.
Beyond those core functions, TCAD is also used for device modeling (creating compact mathematical representations of a device for use in circuit simulators), manufacturing yield optimization, and process centering, which means fine-tuning a fabrication recipe so that small, unavoidable variations in manufacturing don’t cause chips to fail.
The Physics Under the Hood
Device simulation relies on solving fundamental physics equations across a mesh of thousands or millions of tiny points inside the virtual device. The most common approach is called the drift-diffusion model, which tracks how electrons and holes (the two types of charge carriers in a semiconductor) move under the influence of electric fields and concentration gradients. It combines a charge conservation law for each carrier type with an equation that relates the electric potential to the distribution of charges throughout the device.
For most conventional transistors, drift-diffusion is accurate enough. But as devices shrink to just a few nanometers, the physics gets more complex. Carriers can start behaving more like waves than particles, and energy distributions stop following simple assumptions. TCAD tools handle this by offering more sophisticated models: hydrodynamic models that account for carrier energy, and even full statistical approaches that track individual particle interactions. Engineers pick the level of physics complexity they need based on the device they’re simulating.
Why It Matters for Modern Chips
The semiconductor industry has spent decades shrinking transistors to fit more of them on a chip. That scaling has pushed designs far beyond simple flat structures. Today’s leading chips use FinFET transistors, where the channel stands up vertically like a thin fin, and the newest designs use gate-all-around nanowire or nanosheet transistors, where the gate electrode wraps completely around the channel for maximum control over current flow.
TCAD is essential for developing these architectures. Researchers use 3D simulations to compare how different structures perform before committing to a design. For example, simulations comparing FinFETs and nanowire transistors have shown that current density in a FinFET concentrates heavily at the outer walls of the fin, with the center of the fin carrying two to three orders of magnitude less current. In a nanowire transistor, more than half the cross-sectional area carries significant current, which is one reason the gate-all-around approach offers better control over leakage. At the same time, those simulations revealed a tradeoff: the tightly wrapped gate in a 14-nanometer nanowire structure produced peak temperatures about 35 degrees higher than an equivalent FinFET, information that directly shapes thermal management decisions.
These are exactly the kinds of insights TCAD provides. Engineers can explore a new device concept, identify its advantages and drawbacks, and extrapolate performance to the next technology generation, all computationally.
The Software Landscape
Commercial TCAD began in 1979 with the formation of Technology Modeling Associates (TMA), which commercialized simulation tools originally developed at Stanford University. Silvaco later licensed those same Stanford programs and built competing products. A third vendor, Integrated Systems Engineering (ISE), developed its own suite independently.
Through a series of acquisitions, TMA became part of Avant!, which was then acquired by Synopsys. Synopsys later also acquired ISE, consolidating most of the market. Today, Synopsys and Silvaco are the two remaining major commercial TCAD vendors, with Synopsys holding more than 80% of the market. Synopsys Sentaurus is the dominant tool suite used across the industry and in academic research, offering the full range of process, device, and interconnect simulation capabilities.
How Engineers Use TCAD in Practice
A typical TCAD workflow starts with defining a process flow: the sequence of manufacturing steps that will create the device. The process simulator generates a structure file representing the device geometry and material composition. That structure then gets fed into the device simulator, where the engineer defines operating conditions (applied voltages, temperature) and the software solves the physics equations to produce electrical characteristics like current-voltage curves.
Engineers iterate on this loop constantly. They might adjust a doping concentration, change a material thickness, or modify the shape of a gate electrode, then re-run the simulation to see how performance changes. This is far faster than building real devices, which can take weeks or months per iteration in a fabrication plant. TCAD doesn’t replace physical experiments entirely, but it dramatically reduces the number of expensive test runs needed. The combination of simulation and selective experimental validation is the standard approach at every major semiconductor company.
TCAD vs. Other Types of CAD
People familiar with general CAD software (used for mechanical parts, buildings, or PCB layouts) sometimes wonder how TCAD differs. The key distinction is that TCAD operates at the physics level of individual semiconductor devices. It models atomic-scale processes like how implanted atoms diffuse through a crystal lattice and how quantum effects influence current flow in a nanometer-scale channel. Traditional electronic design automation (EDA) tools, by contrast, work at a higher level of abstraction: they simulate circuits made of thousands or millions of transistors, using simplified mathematical models for each one. TCAD is what generates and validates those simplified models in the first place.
A Quick Note on the Acronym
If you searched “TCAD” expecting information about tricyclic antidepressants, the standard medical abbreviation for that drug class is TCA (not TCAD). TCAs are a class of medications originally developed in the late 1950s for major depressive disorder and now considered a second-line treatment option. The two acronyms are unrelated.