Python speeds up GaN models

March 14, 2013 // By Janine Love
Anyone following developments in highpower RF knows that GaN is hot (pun intended). According to research firm Yole Développement, there is a great deal of R&D activity in this market. In March 2012, the company projected the power GaN market would grow to nearly $10 million in 2012 and $500 million in 2016—up from less than $2.5 million in 2010.

When new semiconductor processes are introduced, they are usually slowly integrated into device modeling and characterisation tools. If a process takes off, however, the capabilities of the device modeling and characterisation tools must keep pace, or even anticipate the needs of the new market. Agilent Technologies, for example, recently announced a new version of its Integrated Circuit Characterisation and Analysis Program (IC-CAP) for high-frequency device characterisation and modeling, offering parameter extraction, data analysis, instrument control, and interface responsiveness. It also now includes Angelov-GaN modeling and Python scripting. This announcement actually includes two noteworthy topics: GaN and Python.

Angelov-GaN is an industry-standard compact device model for GaN semiconductor devices. Since GaN devices typically operate at high power, it is important to be able to model thermal issues and their impacts on device characteristics. Designers working with GaN quickly realised that GaAs models were not good enough. Fortunately, Professor Iltcho Angelov at Chalmers University of Technology (Gothenburg, Sweden) developed his Angelov-GaN model as an alternative.

In IC-CAP 2013.01, Agilent has embraced the Angelov model with the W8533 Angelov-GaN extraction package. An interface lets users execute a step-by-step extraction flow to obtain model parameters. A turnkey flow aims to provide quick-start modeling of GaN devices. The Agilent product uses Python scripting. Why Python? The Python program may soon lay claim to the title of most used interpreter language. Roberto Tinti, device modeling product manager with Agilent EEsof EDA, thinks an advantage of Python is the support of a large, open-source community. “Some of our major customers have started to use Python for their projects,” says Tinti, “and we decided to support it natively in our platform in addition to our own native language, called PEL.” Python is 100× faster than Agilent’s Programming Extraction Language, the language that IC-CAP users have been using to customise their applications.

The new IC-CAP release includes support for Smartspice simulations, as well as support

for gain compression and