Compact Modeling of Oxide-Semiconductor, 2D Material, Carbon Nanotube, and Cryogenic Transistors with Experiment Verification

The relentless drive to make computers faster and smaller has long relied on shrinking the tiny switches inside them, known as transistors. For decades, these switches have been built from silicon, a material that has served the industry well. However, as engineers push these components to their physical limits, they are looking for new materials that can be stacked on top of existing chips or operate in extreme cold without failing. These emerging materials include amorphous oxide semiconductors used in screens, atomically thin two-dimensional sheets, and tiny carbon tubes. The challenge is not just building these devices, but understanding exactly how they behave so engineers can design circuits around them. To do this, scientists use compact models, which are simplified mathematical descriptions that act like a blueprint for how a transistor will respond to electricity. Without a reliable blueprint, it is impossible to predict how these new technologies will perform in a real computer.

In this work, a researcher has created a single, unified blueprint that can describe four very different types of these emerging transistors: oxide-semiconductor switches, two-dimensional material switches, carbon nanotube switches, and standard silicon switches operating at extremely low temperatures. Instead of building a separate set of rules for each material, the author developed a flexible framework that accounts for the unique quirks of each. The model considers how electrons are trapped in imperfections within the material, how they behave when squeezed into incredibly thin spaces, and how they move when the device is cooled down to near absolute zero. By testing this framework against real-world measurements from fabricated devices and published data, the researcher demonstrated that a single set of principles could accurately predict the behavior of all these diverse technologies.

The core of this achievement lies in how the model handles the invisible forces inside the transistor. In these tiny devices, electrons do not just flow like water in a pipe; they interact with the material in complex ways. The model includes a way to calculate the charge of electrons that get stuck in defects, which is crucial for oxide semiconductors that tend to have more of these imperfections than silicon. It also accounts for the fact that in very thin materials, the electrons behave differently than they do in bulk silicon, a phenomenon known as quantum confinement. Furthermore, the model bridges the gap between two ways electrons move: the slow, bumping-around motion seen in long channels, and the fast, unimpeded flight seen in very short channels. This allows the same equation to describe a transistor that is nearly a micrometer long and one that is only a few nanometers wide.

To prove the model works, the researcher first tested it on a newly fabricated transistor made from indium gallium oxide, a type of oxide semiconductor. The device was built on a silicon chip with a gate length ranging from 40 nanometers to nearly a micrometer. The model successfully matched the measured electrical data, but only when the researcher included the effect of trapped charges. Without accounting for these trapped electrons, the model failed to predict how the device turned on. The fit was so precise that it revealed the energy level of these traps, equivalent to a temperature of 700 Kelvin, a detail that would have been missed with a simpler model. This validation confirmed that the model could handle the messy reality of real-world materials, not just idealized theory.

The researcher then applied the same model to two other cutting-edge materials: molybdenum disulfide, a two-dimensional material, and carbon nanotubes. For the molybdenum disulfide transistor, which had a channel length of 10 nanometers, the model accurately predicted its behavior, revealing that the contact resistance between the metal and the material was quite low. For the carbon nanotube transistors, the model captured a unique electrical property where the capacitance, or the ability to store charge, changes in a non-linear way due to the one-dimensional nature of the tubes. The model successfully reproduced the measured data for devices with channel lengths of 90 and 10 nanometers, showing that the same underlying physics could explain both materials despite their different structures.

Finally, the model was tested under extreme conditions by simulating a silicon transistor cooled from room temperature down to 8 Kelvin. In this cold environment, the movement of electrons is heavily influenced by "band-tail states," which are energy levels that appear just below the main energy band where electrons usually sit. The model incorporated these states and successfully predicted that the transistor's ability to switch on and off would stabilize at very low temperatures, matching experimental observations. This confirmed that the model could handle the complex physics of cryogenic operation, which is essential for future quantum computing applications.

With the model validated across these different materials and conditions, the researcher used it to simulate how these transistors would perform in a simple digital circuit, specifically a ring oscillator made of 17 stages. The simulation assumed ideal conditions without external wiring resistance to find the best possible performance. The results showed that the carbon nanotube transistor could switch incredibly fast, with a delay of just 0.15 picoseconds, while the molybdenum disulfide transistor was slower, with a delay of 1.7 picoseconds. However, the simulation also highlighted a weakness in the carbon nanotube design: it suffered from a large leakage of current when the voltage was high, suggesting that while it is fast, it may need better engineering to control that leakage. The oxide semiconductor transistor, while not the fastest, showed excellent control over the current flow, making it a strong candidate for specific applications.

This work provides a powerful tool for the semiconductor industry, offering a single, consistent way to design and evaluate the next generation of transistors. By proving that a unified approach can accurately describe everything from oxide semiconductors to carbon nanotubes and cryogenic silicon, the research removes a significant barrier to developing new technologies. It allows engineers to predict how these devices will behave before they are even built, accelerating the path toward more advanced and efficient electronic systems. The findings suggest that while each material has its own strengths and weaknesses, they can all be understood through a common physical lens, paving the way for a future where diverse materials work together in the same chip.