Device Modeling

[Mirtha Shikles]

Semiconductor device modeling creates models for the behavior of the electrical devices based on fundamental physics, such as the doping profiles of the devices. It may also include the creation of compact models (such as the well known SPICE transistor models), which try to capture the electrical behavior of such devices but do not generally derive them from the underlying physics. Normally it starts from the output of a semiconductor process simulation.

Physics driven device modeling is intended to be accurate, but it is not fast enough for higher level tools, including circuit simulators such as SPICE. Therefore, circuit simulators normally use more empirical models (often called compact models) that do not directly model the underlying physics. For example, inversion-layer mobility modeling, or the modeling of mobility and its dependence on physical parameters, ambient and operating conditions is an important topic both for TCAD (technology computer aided design) physical models and for circuit-level compact models. However, it is not accurately modeled from first principles, and so resort is taken to fitting experimental data. For mobility modeling at the physical level the electrical variables are the various scattering mechanisms, carrier densities, and local potentials and fields, including their technology and ambient dependencies.

By contrast, at the circuit-level, models parameterize the effects in terms of terminal voltages and empirical scattering parameters. The two representations can be compared, but it is unclear in many cases how the experimental data is to be interpreted in terms of more microscopic behavior.

IC development for more than a quarter-century has been dominated by the MOS technology. In the 1970s and 1980s NMOS was favored owing to speed and area advantages, coupled with technology limitations and concerns related to isolation, parasitic effects and process complexity. During that era of NMOS-dominated LSI and the emergence of VLSI, the fundamental scaling laws of MOS technology were codified and broadly applied.[6] It was also during this period that TCAD reached maturity in terms of realizing robust process modeling (primarily one-dimensional) which then became an integral technology design tool, used universally across the industry.[7] At the same time device simulation, dominantly two-dimensional owing to the nature of MOS devices, became the work-horse of technologists in the design and scaling of devices.[8][9] The transition from NMOS to CMOS technology resulted in the necessity of tightly coupled and fully 2D simulators for process and device simulations. This third generation of TCAD tools became critical to address the full complexity of twin-well CMOS technology (see Figure 3a), including issues of design rules and parasitic effects such as latchup.[10][11] An abbreviated perspective of this period, through the mid-1980s, is given in;[12] and from the point of view of how TCAD tools were used in the design process, see.[13]

The semiconductor integrated circuit (IC) design industry is witnessing a monumental shift with the emergence of chips harboring over 1 trillion transistors. This advancement is largely due to the development of advanced production nodes, which also introduces a new realm of complexities in semiconductor device modeling.

Device modeling engineers encounter novel effects and massive datasets that challenge traditional, manual modeling approaches. The accuracy required in semiconductor device modeling has also reached unprecedented levels. Device modeling engineers are responsible for creating accurate SPICE models and process design kits (PDK) for IC designs based on both silicon and III-V semiconductors under various operating conditions such as radio frequencies (RF), voltages, and temperatures. To address these challenges, Keysight has empowered semiconductor foundries and design companies with a highly flexible, automated device modeling solution from measurement to verification for over two decades.

Device modeling is the creation of mathematical models and formulas that describe the characteristics of semiconductor devices. It is a critical aspect of the semiconductor integrated circuit (IC) design process, enabling design engineers to predict the behavior of semiconductor devices like transistors, diodes, and capacitors before physical prototyping.

At its core, a semiconductor device model consists of a set of rigorously defined equations that describe how characteristics of a transistor or diode change under various conditions. Users can externally access and set the values of parameters within these equations to reflect different operating environments.

The determination of parameter values is a meticulous process that involves either direct extraction from empirical data or optimization techniques to ensure the model equations closely fit the observed data from experimental results, including varied sizes, temperatures, and operating voltages.

Accurate device modeling serves as the foundation for developing accurate simulation models and process design kits (PDKs). Design teams can significantly reduce the time and cost associated with the design and testing of ICs.

Compact models find wider application in high-level electrical simulations of integrated circuits (ICs) and circuit boards. Some are derived from or simplified versions of physics-based models. Others are empirically determined from experimental data. Compact models offer faster simulations due to their simpler nature.

Next, various real device effects are added to the core model, including velocity saturation effect, mobility field dependence, sub-threshold slope degradation, non-linear series resistances, channel-length modulation, drain-induced barrier lowering, and self-heating effect temperature dependence.

Device characterization involves measuring the physical and electrical properties of the devices. This step involves collecting a massive amount of measured data from different wafers over several temperatures. Therefore, it is important for device modeling teams to automate wafer-level measurements for efficient device characterization.

One of the major challenges is automated measurements over temperature. Due to wafer and hardware expansion (or contraction), the wafer mapping software may lose control of the alignment in the X and Y directions. Pattern recognition technology is critical to overcome these problems.

In this step, specific parameters such as DC, CV, S-parameters, and noise that define the behavior of the device are extracted from the characterization data. The accuracy of the model largely depends on how precisely these parameters are extracted, which is impacted by the type and accuracy of the available device characterization data.

This step involves creating the mathematical model that represents the device's behavior based on the extracted parameters. The merits of the model are partially determined by the quantity and the type of characterization required.

To cope with increasingly complex design requirements, design teams rely on device modeling software to thoroughly understand, predict, and optimize the behaviors of devices. The ability to model and simulate device behavior under various conditions allows design engineers to reduce the dependency on costly physical prototypes, saving both time and resources.

To understand the essential role of device modeling in IC design, consider a simplified resistor model. In this scenario, you have a set of measured data points for current (I) and voltage (V). The relationship between current and voltage in a resistor is linear, described by the equation I = V/R, where R is the resistance. By adjusting the parameter R, the best-fit results with the data points can be obtained when R=R0.

Although foundry models have achieved impressive quality, they may not be sufficient for certain specific applications not characterized during extraction. Therefore, fine-tuning foundry libraries in-house becomes critical for design houses to speed up the model/design iterations, saving IC design costs.

With 5G applications, typical circuit operating frequencies continue to advance well into the RF and microwave frequency range. Design teams need models that can accurately predict device behaviors at DC, RF, and millimeter-wave regions.

Moreover, the amount of data measured for device modeling purposes has been increasing exponentially. With measurements taking several hours or even days, it is essential to be as efficient as possible without compromising measurement accuracy. Measurement control software must work in conjunction with the prober's native control software and each instrument to allow automated measurements across temperature.

Today's most advanced semiconductor foundries and Integrated Device Manufacturers (IDMs) rely on Keysight for modeling silicon CMOS, Bipolar, compound gallium arsenide (GaAs), gallium nitride (GaN), and many other device technologies.

Within a single environment, modeling teams can use IC-CAP to automate measurements, simulate device performance, extract data, optimize model parameters, perform statistical analysis, and generate best-case and worst-case models. IC-CAP provides extraction routines for industry-standard and Keysight proprietary models such as diodes, BJT, MOSFET, MESFET, HEMT, noise, thermal, and more.

IC-CAP offers complete coverage for GaN devices, from the traditional, empirical-based, Angelov-GaN model to the most recent physics-based models recently promoted to industry standards by the Compact Modeling Council, the ASM-HEMT, and MVSG models.

The Device Modeling Language (DML) is a domain-specific language for creating fast functional transaction-level virtual platform models. The first version of DML was launched in 2005, and it has been the standard way to build device models for the Simics simulator ever since.

The Simics use of DML as the primary modeling tool is an interesting example for the virtual platform community. The most common approach for virtual platform modeling is to use a general-purpose language (like C++ or C#) along with a modeling library and simulator API. Designing a domain-specific language is philosophically different, and in our experience DML has provided benefits for programmer productivity and model quality that provide a clear return-on-investment.

c80f0f1006