Global TCAD Solutions

Predictions based on Physics

Device and Circuit Variability

Due to continuous down-scaling in CMOS-technology, device feature size is approaching atomistic scales. Hence, self-averaging of device properties for individual devices becomes less effective and variability has become a major concern for CMOS scaling and integration. For current and future device generations, careful analysis of intrinsic fluctuation on circuit and system level is vital to reduce variability and increase yield.

Impact on Device and Circuit Level

3D image of atomistic FinFET model, showing discrete random traps and dopants
Discrete atomistic random traps and dopants
FinFET with applied metal grain roughness model
Metal grain roughness (MGR)
Line edge roughness shown at example of FinFET gate
Line edge roughness (LER)
Transfer characteristics (IDlin, IDsat)
Transfer characteristics (ID,lin, ID,sat)
Ion/Ioff diagram, DIBL (drain induced barrier lowering) histogram
Post-processing: ION/IOFF ratio and DIBL histogram

GTS Framework offers an accurate yet efficient model to predict and also "design" fluctuations in the electrical device characteristics. Mixed-mode analysis allows for investigation of circuits, such as 6T or 8T SRAM cells. The GTS variability module automatically introduces major/dominant variability sources to the 3D device- and circuit simulation model

Atomistic Traps and Dopants (or Random Dopant Fluctuations)

For simulation of random discrete dopants (RDD) and discrete oxide and interface traps, Minimos-NT is capable of generating random configurations, in order to capture the atomistic nature of traps and dopants [Nobuyuki01]. Random dopant fluctuations are dominant in recent planar CMOS technologies.
See Example: SOI-FinFET-Discr.Dop.

Metal Grain Roughness

The grain roughness model allows to describe poly or metal gate granularity (MGG) and the induced work function fluctuation (WKF). In order to achieve patterns with the desired statistical properties, random surface grains are grown from random seeds to the average size. The WKF due to grain orientations of gate materials such as titanium nitride (TiN), molybdenum nitride (MoN), and tungsten nitride (WN) becomes dominant for undoped channel devices [Cheng10].

Line Edge Roughness

As a further component, the line edge roughness (LER) model allows to introduce geometrical variation to a semiconductor surface or interface [Asenov03]. With its flexible generic implementation, LER can describe variations due to the lithographic process like gate edge roughness (GER) or trench edge roughness, as well as other process-induced fluctuations like oxide thickness variation (OTV). In FinFET structures, fin edge roughness (FER) is another major variability source that can be captured using the LER model.

Distributed Computing

In order to obtain meaningful statistical values, variability analysis typically requires 100 to 1000 cases (=device implementations). Here, our in-house cluster solution GTS JobServer automates setup and execution of such simulation jobs, including task scheduling, distribution and collection of results. Running simulation tasks can be easily monitored, and intermediate results are instantly accessible at runtime as well as after a job has finished.

Post-Processing

GTS Framework offers a range of integrated post-processing functions for calculating characteristic values like threshold voltage, DIBL, ION/IOFF-ratio, sub-threshold slope and more. It allows to extract meaningful statistical results in a homogeneous work-flow within seconds.

Conclusion

Considering the impact of variability sources on device properties is vital to control variability and yield for current and future device generations.

The GTS variability module allows simulation and thorough analysis of various variability sources such as discrete random dopants (RDD), metal grain roughness (MGR), line edge roughness (LER), etc., on device and circuit level.

Example in Detail

For a step-by-step description of an application example, see Variability Simulation.

Key Features

  • Random discrete dopants (RDD)
  • Discrete oxide and interface traps
  • Metal grain roughness (MGR)
  • Line edge roughness (LER)
  • Integrated post-processing
  • Automatic job distribution on cluster

Applications and Benefits

  • Planar and FinFET transistors
  • Variability on device level and circuit level
  • Automatic simulation work flow
  • Fully integrated in GTS Framework R.2013
  • Efficient yet very affordable solution

Download Application Flyer: Variability & Reliability

File link icon for GTS-App-Reliability-Web_03.pdf

GTS-App-Reliability-Web.pdf

GTS Framework Application Flyer: Reliability (web version)

578 kB

References

[Nobuyuki01]: Sano, Nobuyuki; Tomizawa, M., "Random dopant model for three-dimensional drift-diffusion simulations in metal–oxide–semiconductor field-effect-transistors," Applied Physics Letters , vol.79, no.14, pp.2267,2269, Oct 2001

[Cheng10]: Hui-Wen Cheng, Fu-Hai Li, Ming-Hung Han, Chun-Yen Yiu, Chia-Hui Yu, Kuo-Fu Lee, Yiming Li, "3D device simulation of work function and interface trap fluctuations on high- k / metal gate devices." in Int. Electron Device Meeting Tech. Dig., Dec. 2010, pp. 15.6.1-15.6.4.

[Asenov03]: A. Asenov, S. Kaya, A.R. Brown, "Intrinsic parameter fluctuations in decananometer MOSFETs introduced by gate line edge roughness," Electron Devices, IEEE Transactions on , vol.50, no.5, pp.1254,1260, May 2003.