SI - Modeling of High-Frequency Silicon Transistors
Polynomial noise modeling of silicon-based GaN HEMTs
Article first published online: 24 MAY 2013
Copyright © 2013 John Wiley & Sons, Ltd.
International Journal of Numerical Modelling: Electronic Networks, Devices and Fields
Special Issue: Modeling of High-Frequency Silicon Transistors
Volume 27, Issue 5-6, pages 812–821, September-December 2014
How to Cite
Colangeli, S., Bentini, A., Ciccognani, W. and Limiti, E. (2014), Polynomial noise modeling of silicon-based GaN HEMTs. Int. J. Numer. Model., 27: 812–821. doi: 10.1002/jnm.1907
- Issue published online: 20 AUG 2014
- Article first published online: 24 MAY 2013
- Manuscript Accepted: 17 APR 2013
- Manuscript Received: 6 APR 2013
- automated test bench;
- GaN-on-silicon HEMT;
- polynomial noise model;
- source pull
In the framework of silicon (Si) technology, evolution towards high-frequency analog applications – which involves innovative solutions such as SiGe BiCMOS and FinFET devices – wide bandgap semiconductors grown on Si substrates are likely to represent a valid option in those cases wherever high-power handling and low noise figures are required. Although such active devices have been extensively investigated in the last years, much of interest has been devoted in developing nonlinear models for high-power applications, whereas reliable noise models still lack, in particular, the validity of traditional (i.e. equivalent temperature-based) approaches for noise modeling of wide bandgap devices has not been sufficiently probed yet.
In this contribution, a quite general, black box noise model of active devices is proposed and applied to a family of gallium nitride-on-Si high-electron-mobility transistors fabricated by Selex ES. The model is based on a polynomial approximation of the device correlation matrix and does not require that an accurate small-signal equivalent circuit is available; instead, it can be extracted from multifrequency source pull data. Experimental results demonstrate that a typical behavior of the noise parameters is obtained, both versus frequency and gate periphery. Copyright © 2013 John Wiley & Sons, Ltd.